University of Huddersfield Repository

University of Huddersfield Repository
University of Huddersfield Repository
Dean, Lionel Theodore
Futurefactories: the application of random mutation to three-dimensional design
Original Citation
Dean, Lionel Theodore (2009) Futurefactories: the application of random mutation to threedimensional design. Doctoral thesis, University of Huddersfield.
This version is available at http://eprints.hud.ac.uk/8799/
The University Repository is a digital collection of the research output of the
University, available on Open Access. Copyright and Moral Rights for the items
on this site are retained by the individual author and/or other copyright owners.
Users may access full items free of charge; copies of full text items generally
can be reproduced, displayed or performed and given to third parties in any
format or medium for personal research or study, educational or not-for-profit
purposes without prior permission or charge, provided:
•
•
•
The authors, title and full bibliographic details is credited in any copy;
A hyperlink and/or URL is included for the original metadata page; and
The content is not changed in any way.
For more information, including our policy and submission procedure, please
contact the Repository Team at: [email protected]
http://eprints.hud.ac.uk/
FUTUREFACTORIES: THE APPLICATION OF RANDOM MUTATION TO
THREE-DIMENSIONAL DESIGN
LIONEL THEODORE DEAN
A thesis submitted to the University of Huddersfield
in partial fulfillment of the requirements for
the Degree of Doctor of Philosophy
The University of Huddersfield
September 2009
2
Abstract
The title of the project, ‘FutureFactories’, describes an exploration of direct digital manufacturing
and the use of this technology in creating new models for consumer product design practice.
In additive fabrication itself, there is no economic advantage in producing identical artefacts: given
this, and the free-form potential of a technology that can deliver almost any form imaginable, the
project examines the possibility of modifying the design with every artefact produced. The aim is to
create automated systems capable of volume production, establishing mass individualisation: the
industrial scale production of one-off artefacts.
This work explores the potential to combine parametric CAD modelling with computer programming
to create animated meta-designs that change in real time. These scripts introduce a random
computer generated element into each physical product ‘printed out’ using direct digital
manufacturing. The intention is to combine qualities normally associated with the vagaries of the
hand-made with the technical resolution of industrial mass-manufacture; whilst at the same time
maintaining a coherent design and identity.
The outputs from this practice-based research project consist of inspirational products ranging from
gallery pieces to commercial retail products and, alongside the real-world artefacts, the scripted
meta-designs from which they are created.
The use of such software processes and real-time networks as generative tools, questions existing
transient boundaries of practice, and exposes the irrelevance of conventional definitions of role. It
is clear that the outcomes of such a new model of creative production cannot be thought of as
traditionally conceived pieces. The outcomes of the research suggest that the resulting artefacts
can be considered both functionally useful and as art. Outside of that, existing definitions convey
little of the reality of their production, as they lie in some new, as yet unspecified, arena of
production.
3
4
Contents
1.0 Aims and Objectives of the Thesis
Page
21
1.1 Key Terminology
Page 21
1.2 Chapter Summary
Page 22
1.3 Introduction
Page 23
1.4 The Technological Landscape
Page 25
1.4.1 Computer Aided Design, CAD
Page 26
1.4.2 Rapid Prototyping, RP
Page 26
1.4.3 Direct or Rapid Manufacture
Page 29
1.5 Computer Generated Art
Page 31
1.5.1 William Latham
Page 31
1.5.2 Richard Dawkins’ Biomorph
Page 32
1.5.3 Latham and Todd
Page 33
1.5.4 Karl Sims
Page 34
1.5.5 Digital Sculpture, Digital Art Practice and Design Art
Page 35
1.6 Commercial Examples of Digital/Direct/Rapid Manufacturing
Page 38
1.4.1 Materialise
Page 38
1.2.2 Freedom of Creation, FOC
Page 40
1.2.3 Patrick Jouin
Page 40
1.2.4 Bathsheba Grossman
Page 41
1.7 Mass Customisation and Individualisation
Page 43
1.5.1 Ron Arad
Page 43
1.5.2 Celestino Soddu
Page 44
1.5.3 Fluidforms
Page 47
1.5.4 Front
Page 48
5
1.6 Software Developed for Generative Design
1.6.1 Genometri – Generative Design Software
Page 52
Page 52
1.6.2 Bentley Systems, Generative Components, and the Smart
1.9
Geometry Group
Page 53
Literature Review Summary
Page 55
2.0 Project Overview
Page 57
2.1 Design Residency Program
Page 57
2.2 Expansion of the Project into a PhD Study
Page 59
2.3 Justification of the Practice Based Elements of the Project
Page 59
2.4 Researcher’s Established Practice
Page 60
3.0 Mass Individualisation: Industrial Scale Production of One-off Artefacts
Page 61
3.1 Why Individualise?
Page 61
3.2 Industrial Production Versus Craft Making – The Need for Automation
Page 62
3.3 Mass Individualisation Distinct From Mass-Customisation
Page 63
3.4 Consumer Input to the Individualisation
Page 63
3.5 Random Morphing, Control or Happenstance
Page 65
3.6 Meta-Designs
Page 66
4.0 Computer Generation of Variants
Page 67
4.1 Key Frame Animation
Page 68
6
4.2 Procedural Animation: Rules, Ranges and Relationships
4.2.1 Procedural Animation Principles as Applied in FutureFactories
Page 69
Page 70
4.2.2 The Nature of Morphing: Micro Changes, Macro Changes and
Alterations of the Geometrical Structure
4.2.3 Structural Changes to Geometry
Page 74
Page 75
4.3 The Use of Evolutionary and Genetic Algorithms: The Introduction of
Page 77
Selection
4.3.1 A Model for Mutation and Selection
Page 77
4.3.2 The Introduction of an Evolutionary Pressure: Aesthetic
Evolutionary Design
Page 79
4.3.4 Ranking
Page 80
4.3.5 Step Size – Micro-Mutation, Macro-Mutation and the Balance Between
Different Transformations
Page 81
4.3.6 Assessing Functionality and Manufacturability
Page 81
4.3.7 Scoring the Aesthetic
Page 82
4.4 The Constructive Solid Geometry, CSG ‘Building Block’ Approach
Page 83
4.4.1 DNA: Design Case Study of a Constructive Solid Geometry
‘Building-Block’ Approach
Page 85
4.4.2 DNA II
Page 91
4.4.3 Holy Ghost: Case Study - Combining the Building Block Approach With
Morphing
Page 97
4.4.4 SuperKitch Bangle – Case Study
Page 105
4.5 Section Summary
Page 111
5.0 Summary of Design Work and Exhibitions 2003 – 2006
Page 113
5.0.1 Lighting Objects
Page 114
5.0.2 Materials and Processes
Page 114
5.0.3 The Driving Computation
Page 115
5.1 The ‘First Collection’
Page 117
5.1.1 Lampadina Mutanta
Page 117
5.1.2 Nautilus
Page 133
5.1.3 Tuber
Page 139
5.1.4 Let’s Twist Again Candle Holder
Page 145
7
5.1.5 Twist
Page 149
5.2 The ‘First Collection’ Exhibitions
Page 155
5.2.1 Barnsley Design Centre 27/10/03 – 21/11/03
Page 156
5.2.2 Dean Clough 01/12/03 – 16/01/04
Page 157
5.2.3 The Media Centre, Huddersfield 23/01/04 – 13/02/04
Page 157
5.3 Designersblock Milan 14/04/04 – 19/04/04
Page 161
5.4 Evolution of the ‘First Collection’
Page 167
5.4.1 Tuber9
Page 169
5.4.2 Lightbikes
Page 173
6.0 Commercial Retail Products
Page 177
6.1 RGB.mgx
Page 177
6.2 Creepers.mgx
Page 183
7.2.1 Tangle
Page 189
6.3 Entropia
Page 191
6.4 Section Summary
Page 199
7.0 Later works
Page 201
7.1 Artenoma
Page 201
7.2 Cornuta
Page 207
7.3 Holy Ghost
Page 209
7.3.1 Metal Plated Holy Ghost
Page 219
7.4 Pallavi
Page 221
7.5 Jewellery
Page 225
7.4.1 Aorta
Page 225
7.4.2 Puja
Page 226
8
7.4.3 Icon
Page 227
7.6 Puja Table Lamp
Page 241
7.7 Section 7 Summary
Page 243
8.0 Summary and Conclusions
Page 245
8.1 Technology
Page 246
8.1.1 Additive Fabrication
Page 246
8.1.2 Bureau Culture
Page 248
8.1.3 Materials
Page 248
8.1.4 Sustainability
Page 250
8.2 The Consumer Interface
Page 253
8.2.1 Virtual Reality, VR
Page 253
8.2.2 The User-Script Interface
Page 254
8.3 The Role of the Designer
9.0
Page 255
8.3.1 Design for Manufacture
Page 257
8.3.2 Communication
Page 258
8.3.3 Authorship
Page 258
8.3.4 The Geometry Comes Free ‘Myth’
Page 260
8.3.5 Art, Craft and Design
Page 261
8.3.6 Design Skills and Education
Page 262
8.4 Achieving a Balance Between Order and Chaos
Page 263
8.5 Recommendations for Further Research
Page 267
8.5.1
CAD and Programming
Page 267
8.5.2
The Designs
Page 267
8.5.3
The Artefacts
Page 268
Technical Glossary
Page 269
Total 51,247 words.
10.0
Bibliography
Page 275
9
11.0 Appendices
Appendix 1
International Conference on Advanced Engineering Design
Paper, Prague, 2003
Appendix 2
Page 287
Evolving Individualised Consumer Products
6th International Conference of the European Academy
of Design Paper, Bremen, 2005
Appendix 3
Page 297
FutureFactories: Teaching Techné
5th International Conference of the European Academy
of Design Paper, Barcelona, 2003
Appendix 4
Newdesign Issue 19: FutureFactories is adding a
new dimension to the design process, 2004
Appendix 5
Page 315
Page 325
Supportive technologies as creative processes
9th International Design Conference paper, Croatia 2004
Page 331
Appendix 6
Time Compression Technologies Conference paper, 2006
Page 337
Appendix 7
Newdesign Issue 57: One and Only: Lionel T Dean
describes how FutureFactories is pushing the boundaries
of what the RP industry can produce
10
Page 345
List of Figures
Figure 1
Digital Practice Timeline
Gatefold Pages 19-20
Figure 2
A 3DSystems SLA machine
Page 27
Figure 3
The EOS P390 SLS machine
Page 28
Figure 4
Small Hand-Drawn FormSynth Tree,
Page 31
William Latham, 1992
Figure 5
Generations of the Dawkins’ Biomorph, 1986
Page 32
Figure 6
The Visitor, Desmond Morris, 1949
Page 33
Figure 7
A Set of “Children” from Mutator,
Page 34
Todd and Latham, 1992
Figure 8
A Still from Panspermia, Sims, 1990
Page 35
Figure 9
Attached to Light, Geoffrey Mann, 2005
Page 36
Figure 10
Blown, Geoffrey Mann, 2005
Page 36
Figure 11
Teacup, Robert Lazzerini, 2003
Page 37
Figure 12
Snot Vases, Marcel Wanders, 2001
Page 37
Figure 13
Lilly, Janne Kyttanen, 2003
Page 39
Figure 14
V-Bag, Freedom of Creation, 2004
Page 40
Figure 15
Chairs – Solid Collection, Patrick Jouin, 2004
Page 41
Figure 16
Quinse, Bathsheba Grossman, 2005
Page 42
Figure 17
Bouncing Vases, Ron Arad, 2000
Page 44
Figure 18
Computer Generated Chair Designs,
Page 45
Celestino Soddu, 2001
Figure 19
Digitally Manufactured Chair Models,
Page 46
Celestino Soddu, 2001
Figure 20
Serene Pepper Grinder, Fluidforms,
Page 47
Customisation Screen Shots, 2008
Figure 21
Rat Wallpaper, Front, 2006
Page 48
Figure 22
Sketch Furniture Creation Video, Front, 2005
Page 48
Figure 23
Chair - Sketch Furniture Collection, Front, 2005
Page 49
Figure 24
Tavs Jorgensen’s Data Glove
Page 50
Figure 25
One-Liner Glass Bowls, Tavs Jorgensen, 2008
Page 51
Figure 26
Scumak No 2, Roxy Paine, 1998
Page 51
Figure 27
Holy Ghost chairs, 2006
Page 65
Figure 28
Phenotype Form and Genotype Data List
Page 68
Figure 29
Tuber Variants
Page 69
Figure 30
A Solid Formed, Lofted Square
Page 70
11
Figure 31
The Effect of Scale Variance
Page 71
Figure 32
The Effect of Differing Scale Variance on all Three Lofted Profiles
Page 71
Figure 33
The Effect of Rotating One Lofted Profile
Page 72
Figure 34
The Effect of Altering the Height of the Central Profile
Page 72
Figure 35
The Effect of Combined Transformations
Page 73
Figure 36
Twist Candlestick
Page 73
Figure 37
Sphere of Influence
Page 74
Figure 38
Modes of Mutation
Page 75
Figure 39
Changes in Model Structure
Page 76
Figure 40
The Mutation, Selection and Animation Process
Page 78
Figure 41
Aspects of Evolutionary Design by Computers,
Page 79
Bentley, 1999
Figure 42
Initial State and the form after 200 generations
Page 82
Figure 44
Example of Virtools’ Script
Page 84
Figure 45
DNA Luminaire
Page 85
Figure 46
DNA Components
Page 86
Figure 47
Boolean Union
Page 86
Figure 48
DNA Schematic
Page 88
Figure 49
Lens Size Selection within the Virtools Script
Page 89
Figure 50
DNA II Iterations
Page 90
Figure 51
Side Emitter LED Fitted to 30mm Rim
Page 91
Figure 52
Rim Link Rotations
Page 92
Figure 53
DNA II Build Volume Restriction
Page 92
Figure 54
Component Make up of DNA II Iterations
Page 93
Figure 55
DNA II #1 ‘Growth’ Iterations 1 – 10
Page 94
Figure 56
DNA II #1 ‘Growth’ Iterations 20 – 100
Page 95
Figure 57
Holy Ghost
Page 96
Figure 58
Virtools Holy Ghost Script: Phase 1,
Page 98
Placing ‘Buttons’
Figure 59
Virtools Holy Ghost Script: Phase 2,
Page 98
Uniform Axial Expansion
Figure 60
Virtools Holy Ghost Script: Phases 3,
Page 98
Irregular Expansion
Figure 61
Placed Buttons
Page 99
Figure 62
Sequential Full Expansion
Page 100
Figure 63
Balanced Results from Step-by-Step Expansion
Page 100
Figure 64
Blending Out of Expansion Amplitude
Page 101
12
Figure 65
Uniform Compared with Non-Uniform Expansion
Page 101
Figure 66
The Prevention of ‘Spikes’
Page 102
Figure 67
Holy Ghost Polyamide Ink Springs
Page 103
Figure 68
‘Mapping’ in the Tuber Animation
Page 104
Figure 69
SuperKitsch Bangle Configuration
Page 105
Figure 70
Library of Charm Elements
Page 106
Figure 71
SuperKitsch Script Schematic
Page 108
Figure 72
SuperKitsch Iteration
Page 109
Figure 73
SuperKitsch Iteration
Page 109
Figure 74
SLS Sample Section
Page 110
Figure 75
Lampadina Mutanta
Page 118
Figure 76
Lampadina Mutanta Cross-Section
Page 119
Figure 77
Lampadina Mutanta ‘Drops’
Page 120
Figure 78
Lampadina Mutanta ‘Tentacles’
Page 120
Figure 79
Lampadina Mutanta ‘Risers’
Page 120
Figure 80
Lampadina Mutanta Morphing Features
Page 121
Figure 81
Casting Problems at the Tentacle Tips
Page 123
Figure 82
Wax Halves for Casting
Page 124
Figure 83
The First, Single Piece, Castings, March 2003
Page 124
Figure 84
Thermojet Wax Patterns
Page 125
Figure 85
Stainless Steel Cast Halves
Page 125
Figure 86
Halves Welded and the Weld Dressed
Page 126
Figure 87
Lampadina Mutanta #3, Barnsley Design Centre, October 2003
Page 126
Figure 88
SLS Lampadina Mutanta
Page 127
Figure 89
Lampadina Mutanta Variants #1 -#5
Page 128
Figure 90
Lampadina Mutanta Tentacle Integration
Page 130
Figure 91
Lamadina Mutanta #5 Pictured in Milan, 2004
Page 131
Figure 92
Lampadina Mutanta Variants, September 2003
Page 131
Figure 93
Constant Fillet Radius Problems
Page 132
Figure 94
Nautilus Cut-Away
Page 133
Figure 95
Nautilus Lighting Effect, Viewed From Above Looking Downwards
Page 134
Figure 96
Nautilus Morphing Features
Page 134
Figure 97
Nautilus #1, September 2003
Page 135
Figure 98
Nautilus #4 in Bronze
Page 135
Figure 99
Nautilus Iterations
Page 136
Figure 100
Nautilus Tentacle Development
Page 137
13
Figure 101
Nautilus Iterations, Barnsley Design Centre, October 2003
Page 138
Figure 102
Nautilus Body Geometry
Page 138
Figure 103
Tuber #5
Page 139
Figure 104
Tuber HP LED mounting
Page 140
Figure 105
Tuber Morphing Features
Page 140
Figure 106
ZCorp Process Tuber Iterations
Page 141
Figure 107
Tuber Iterations
Page 142
Figure 108
Let’s Twist Again Iterations
Page 144
Figure 109
Let’s Twist Again #1, Cast in Bronze
Page 145
Figure 110
Let’s Twist Again Morphing Features
Page 146
Figure 111
Twist #1 Candlestick in Cast Aluminium
Page 149
Figure 112
Twist Morphing Features
Page 150
Figure 113
Twist Cross-Section Morphing, Square to Circular
Page 150
Figure 114
Twist Iterations, Photographed in Milan, April 2004
Page 151
Figure 115
Twist Iterations
Page 152
Figure 116
Exhibition Layout, Barnsley Design Centre, 2003
Page 156
Figure 117
The Dean Clough Board Room, December 2003
Page 157
Figure 118
The Media Centre Exhibition Set-Up, Huddersfield, Spring 2004
Page 158
Figure 119
The Media Centre, Huddersfield on Opening Night (i)
Page 159
Figure 120
The Media Centre, Huddersfield on Opening Night (ii)
Page 159
Figure 121
The Media Centre, Huddersfield on Opening Night (iii)
Page 160
Figure 122
Studio Zeta, Milan, 2004
Page 161
Figure 123
DesignersBlock, Milan 2004, Ground Floor Plan
Page 162
Figure 124
Studio Zeta on the March Survey Visit, 2004
Page 163
Figure 125
Studio Zeta on the March Survey Visit, 2004
Page 163
Figure 126
Studio Zeta Installation, Milan, April 2004
Page 164
Figure 127
DesignersBlock, Milan 2004, Preview Night
Page 165
Figure 128
DesignersBlock, Milan 2004 (i), Photograph Core77
Page 166
Figure 129
DesignersBlock, Milan 2004 (ii), Photograph Core77
Page 166
Figure 130
DesignersBlock, Milan 2004 (iii), Photograph Core77
Page 166
Figure 131
Newdesign Issue Nineteen Front Cover, 2004
Page 167
Figure 132
Tuber9
Page 168
Figure 133
Tuber9 Configuration
Page 170
Figure 134
SLS Translucency in Tuber9
Page 171
Figure 135
Tuber9 LED Clip Fittings
Page 171
Figure 136
Tuber9 Morphing Features
Page 172
Figure 137
Tuber9 in Laser Sintered Nylon
Page 172
14
Figure 138
Lightbikes Bicycle Concept
Page 173
Figure 139
Lightbikes Tuber, Morphing Features
Page 173
Figure 140
Lightbikes Tuber Iterations
Page 174
Figure 141
Lightbikes, Milan, 2006
Page 175
Figure 142
Lightbikes, Rome (i), 2006
Page 175
Figure 143
Lightbikes, Rome (ii), 2006
Page 175
Figure 144
RGB Detail Showing the Colour Mix
Page 177
Figure 145
RGB Iterations
Page 178
Figure 146
RGB Morphing Features
Page 179
Figure 147
Stand Alone Mixed Colour RGBs
Page 180
Figure 148
RGB ‘Eyelash’ Feature
Page 180
Figure 149
RGB ‘Eyelash’ Feature Lighting Effect
Page 181
Figure 150
Colour Days Installation, Warsaw, 2007
Page 181
Figure 151
Creepers Iterations
Page 182
Figure 152
Early Creepers Concept
Page 183
Figure 153
Creepers Installation
Page 184
Figure 154
Creepers’ Reflector Group
Page 185
Figure 155
3D Envelope of Reflector Group
Page 185
Figure 156
Integral Spring Clips
Page 186
Figure 157
Revised Clip Detail
Page 186
Figure 158
New Conductor Button
Page 187
Figure 159
Creepers’ Installation Detail (i)
Page 187
Figure 160
Creepers’ Installation Detail (ii)
Page 188
Figure 161
Creepers’ Installation at the Milan Furniture Fair, 2005
Page 188
Figure 162
Tangle Photographed at ‘Funky’, Norsu Gallery, Finland, 2007
Page 189
Figure 163
‘Funky’ Event Invitation, Norsu Gallery, Finland, 2007
Page 190
Figure 164
Entropia
Page 192
Figure 165
Table, Wall and Pendant Entropia Variants
Page 193
Figure 166
Entropia Parts at the Break-Out Station, Protosystems, Palma
Page 194
Figure 167
Flower (top left) and Leaf (bottom right) Features
Page 195
Figure 168
Time Lapse Morphing of the Flower Feature
Page 196
Figure 169
Entropia Detail (i)
Page 196
Figure 170
Entropia Detail (ii)
Page 197
Figure 171
Process 2008
Page 197
Figure 172
Artenoma 2003 Collection
Page 201
Figure 173
Proposed Artenoma 2005 Collection
Page 202
Figure 174
Artenoma Iterations
Page 203
15
Figure 175
Artenoma SLS Prototype
Page 203
Figure 176
Artenoma Animation (i)
Page 205
Figure 177
Artenoma Animation (ii)
Page 205
Figure 178
Cornita Iterations
Page 206
Figure 179
Cornuta Cross-Section
Page 207
Figure 180
Cornuta Geometry
Page 208
Figure 181
Louis Ghost Chair, Philippe Starck, Kartell 2002
Page 209
Figure 182
Holy Ghost Concept Sketch
Page 210
Figure 183
Portions of Louis Ghost Chair Replaced
Page 211
Figure 184
Holy Ghost Arm Separation
Page 212
Figure 185
Perimeters, Boundaries and Borders, Lancaster, 2006
Page 213
Figure 186
Holy Ghost Iterations
Page 214
Figure 187
Digitability DesignMai, Berlin, 2007
Page 215
Figure 188
Trans-Form Exhibition, Paris, 2008
Page 216
Figure 189
Trans-Form Exhibition, Paris, 2008
Page 216
Figure 190
Trans-Form Exhibition, Paris, 2008
Page 217
Figure 191
Trans-Form Exhibition, Paris, 2008
Page 217
Figure 192
Nickel Plated Holy Ghost #2, Front
Page 219
Figure 193
Nickel plated Holy Ghost #2, Rear
Page 219
Figure 194
Pallavi
Page 220
Figure 195
Pallavi Geometry
Page 221
Figure 196
Pallavi LED Position Out of Direct View
Page 222
Figure 197
Pallavi Trumpet Cluster
Page 222
Figure 198
Pallavi Lighting Effects
Page 223
Figure 199
Pallavi Close-Up
Page 223
Figure 200
Digital Rendering of Aorta, 2007
Page 226
Figure 201
Digital Rendering of Puja, 2007
Page 226
Figure 202
Digital Rendering of Icon, 2007
Page 227
Figure 203
DMLS Support Requirement
Page 229
Figure 204
The Original Icon Design on the Left Alongside the DMLS Adaptation
Page 230
Figure 205
Icon DMLS Build Orientation
Page 231
Figure 206
18K DMLS Gold Chain, Towe Norlén, Lena Thorsson
Page 232
Figure 207
First Batch of Titanium Icons, January 2008
Page 233
Figure 208
Icon Iterations 1 – 12
Page 234
Figure 209
Icon Iterations 13-24
Page 235
Figure 210
Icon Surface Finish After Building and Shot Peening
Page 236
Figure 211
Conventional Polishing – Steel Shot
Page 237
16
Figure 212
Conventional Polishing – Wet Blast
Page 237
Figure 213
Conventional Polishing – Ceramic Cones
Page 238
Figure 214
A Full Z Height Icon in Cobalt Chrome
Page 238
Figure 215
MMP Polished Stainless Steel Icon
Page 239
Figure 216
DMLS Trophy, 2008
Page 240
Figure 217
Anodised Icons
Page 240
Figure 218
Puja Table Lamp Design
Page 241
Figure 219
Aluminium (right) and the First Stainless Steel Prototypes
Page 242
Figure 220
Puja Table Lamp, Stainless Steel, 2008
Page 242
Figure 221
Kundalini Brochure, 2006
Page 247
Figure 222
‘Redundant’ Creepers’ Reflectors
Page 250
Figure 223
Host Server and Client PC Approaches
Page 255
Figure 224
The Experimental User Interface
Page 260
Figure 225
Similar Pairs in Icon Production
Page 264
Figure 226
Tuber Animation Frames
Page 265
Figure 227
Adjustment vs Reconfiguration
Page 265
17
Acknowledgement
Throughout the period of this research I have had a huge amount of encouragement and
support……
I am grateful to the School of Art, Architecture and Design and The University of Huddersfield for
their ongoing support. In particular I would like to thank Ertu Unver for his patient supervision and
guidance over virtually seven years of study and Paul Atkinson for commissioning the initial
residency and supporting the project throughout its early development.
I would like to express my gratitude to Katie Bunnell, John Fieldhouse and David Swann for
examining this thesis and for their insightful comments. I would also like to thank Dave Tancock
and Ray Anable for their supervision and Ian Pitchford and his staff at the Research Office.
Finally I would like to express my gratitude to my wife and family for their love, support, patience
and understanding.
18
1.0
Aims and Objectives of the Thesis
The aim of this thesis is to demonstrate the potential for mass individualisation: the industrial scale
production of one-off artefacts. In order to achieve this, the research should:Combine CAD techniques with computer programming;
Create automated systems that need no further creative input to achieve unique outputs;
Create unique variants without repetition;
Achieve an obvious visible difference between iterations while maintaining a coherent
design identity;
Result in artefacts manufactured direct from computer data with the minimum of post
processing; and
Generate desirable and functional consumer products.
1.1
Key Terminology
The creative use of digital technologies is central to this thesis. These technologies are developed
by various rival commercial entities each trying to establish their own process and come with a
bewildering list of names and acronyms. These are detailed a Technical Glossary 9.0. However, in
the context of this project certain key terms, and their definition, require introduction from the onset.
The term Direct Digital Manufacturing (DDM) as used in this thesis refers to the automated
production of artefacts from CAD data. This may be shortened to Digital Manufacturing or Direct
Manufacturing depending on the context. Direct Digital Manufacturing most often refers to additive
manufacturing processes where material is added, where required, rather than cut away from an
oversize billet. Whilst the rise of these additive or so called Rapid Prototyping processes was the
trigger for this project, it is the principal of using computer data rather than fixed tooling (moulds,
dies, patterns and the like) that is important and this applies equally to computer numerically
controlled (CNC) subtractive machining.
The practice considered in the thesis is design for industrial manufacture which is referred to as
Product Design. The term Product Design is used rather than Industrial Design simply because the
desired outputs are consumer products, rather than any of other activities a practicing designer
might become engaged in.
Product Design in the context of this thesis is defined as the
specification of an artefact, which enables its subsequent commercial manufacture without further
creative input from the designer.
21
1.2
Chapter Summary
Section 1: Introduction. This section describes the project, the ideas behind it and the technology
that underpins it. Previous work across art design disciplines that has informed project thinking is
considered in a literature review.
Section 2: Project Overview. This section describes the projects’ origins as a Design Residency and
its contribution to the undergraduate teaching program
Section 3: Mass Individualisation. This section examines retail manufacturing and mass
consumption in the digital era considering where the impetus for such a mass individualisation
and/or customisation concept might originate.
Section 4: Computer Generation of Variants. This section examines, through case studies, the
creation of meta-designs which combine computer scripts with CAD models.
The design
implications of Differing methodologies are compared and evaluated.
Section 5: Design Work and Exhibitions 2003-2006. This section documents the designs created at
the end of the residency period, the dissemination of these outputs and public reaction to them. It
describes how the plausible retail finishes were obtained from budget visualisation processes and
how, towards the end of this period, the option of more exotic manufacturing processes allowed a
fuller realisation of the projects potential.
Section 6: Commercial Retail Products.
This section documents commercial retail design
opportunities that developed from the early research outputs demonstrating the commercial viability
of the project ideas.
Section 7: Later Works.
This section examines project outputs to-date, charting design
developments as the computer programming underpinning the work becomes more sophisticated
and, from a manufacturing perspective, an increasing palette of materials and finishes become
available. The section culminates with Icon, an edition of 100 jewellery pieces, 25 of which are
produced effectively proving the project ideas.
Section 8: Summary and Conclusions. The work is summarised and conclusions drawn in this
section.
22
1.3
Introduction
The title of the project, ‘FutureFactories’, describes a creative exploration of digitally technologydriven design and manufacture. Rapid Prototyping (RP) technologies were developed to compress
product development cycles and are now well established.
The new frontier is Direct Digital
Manufacturing or Rapid Manufacture (RM); Rapid Prototyping technologies applied to the
production of end-use artefacts. These technologies essentially allow functional real-world objects
to be ‘printed’ out in a series of fine layers direct from virtual Computer Aided Design (CAD) data.
The need for significant tooling investment, such as moulds and dies dedicated to a particular job,
is gone. This allows flexibility in production, unprecedented in the industrial era as the massproduction economics of standardization and uniformity cease to apply. Through Direct Digital
Manufacturing a revolution is underway in 3D Product Design that is likely to be as radical as that
brought about by digital technologies in 2D Graphic Design. Layer-build manufacturing costs are
principally governed by the build height i.e. the number of layers built.
They are largely
independent of the part’s geometry. There is no economic to producing repeats as there are no
physical tools or mould dedicated to producing a particular form that can be reused. The question
is therefore, why not produce something unique every time? There are clearly design implications
as a unique specification has to be generated somehow and these will be discussed. From a
purely manufacturing point of view however, the economies of scale rationale of mass-production
need not apply. There will be benefits to volume manufacture in the bulk ordering of material,
shared plant costs and the like, but these advantages would apply equally to volumes of differing
artefacts. Each time CAD information is entered into the build process there is a degree of data
processing which would be reduced or eliminated in repeat builds, this however is the work of
minutes as opposed to the builds themselves which take several hours.
In 2002, the researcher presented to Huddersfield University, an embryonic concept for the
industrial scale production of one-off artefacts. In essence the project proposed an inversion of the
mass production paradigm to one of individualised production; mass-individualisation. Rather than
defining a discrete form, the designer would create a meta-design capable of random, automated
change. This design template would yield a string of unique outcomes, each of which would be
constrained to remain true to the designers’ intent and the desired product identity. The concept
involves combining high-end parametric CAD (in which geometry is defined by relationships rather
than absolute numerical values) with computer scripting.
The variance in form may encompass a variety of design criteria including the relative positioning of
features, shape, scale, proportion, surface texture and pattern. These variations may be multiple
23
and interrelated. The intention is to achieve changes in the 3 dimensional form itself rather than
scaling the existing geometry or by the use of surface treatments such as texture or colour. In this
way, FutureFactories aims to overcome the split between the technological and the aesthetic,
between artistic creativity and mechanised production. While designs for mass production are
regularly highly creative and have strong aesthetics, their impact is diminished by familiarity. They
are by definition commonplace. In hand worked artefacts there will be differences at some level in
every piece. This allows for diversity within a design type and emotional responses to individual
outcomes.
A future is envisaged in which virtual reality (VR) boutiques are filled with ‘living’
products. At any given moment a product of choice is frozen to create a unique design that may be
ordered on-screen, digitally manufactured, and delivered to the door. An original… A one-off... A
work of art?
24
1.4
The Technological Landscape
The possibility of creating computer based virtual representations of three dimensional forms has
existed since the first Graphic User Interfaces appeared in the 1960’s. The use of comtemporary
computational power to animate such virtual objects can be seen as a logical extension of that
capability.
As the computer has increased in speed and power, the possibilities for manipulating
form have become ever more complex and visualisation of the virtual realm increasingly convincing.
However, until recently, computer generated forms existed purely in the virtual world as graphics or
computer art. It is only the advent of additive layer-build fabrication referred to as Rapid
Prototyping, or more importantly the wide spread availability of the technology, that has enabled
these forms to be realised as real-world products.
It is only the advent of additive layer-build fabrication referred to as Rapid Prototyping, or more
importantly the wide spread availability of the technology, that has enabled these forms to be
realised as artefacts.
Initially additive manufacture was limited to visualisation models whose
mechanical performance and longevity bore no relation to the functional parts they were simulating.
Increasingly, materials and processes have been developed that allow the assessment of functional
criteria to the point where certain high-end processes can yield artefacts suitable for retail sale.
Additive layer build Rapid Prototyping, RP is a relatively recent phenomenon. The first patents
where granted in the mid-1980’s and commercial service providers began appearing towards the
end of that decade. The term RP however, can also be applied to computer controlled subtractive
processes, where material is removed from a blank, such as milling. Computer Numerically
Controlled (CNC) milling has been available as an industrial process for decades. The principles
can be traced back as far as the Jacquard Loom and Babbage’s mechanical computers in the
1800’s. Generally however, Rapid Prototyping refers to additive fabrication in which forms are
created by the addition of material only where required with no waste. It is the rise of this
technology that has made Direct Digital Manufacture not only possible but commonplace. In
addition to developments in additive fabrication technology itself, parallel and more general digital
advances have facilitated uptake including:




powerful affordable computing;
developments in design software;
the ability to visualize though powerful graphics; and
the Internet and rapid digital data transfer.
It is now possible to manufacture almost any 3D form direct from an on-screen model allowing
virtual graphics to be ‘printed-out’ as functional artefacts with the properties expected of mass-
25
market consumer products. This can effectively remove manufacturing criteria from the creative
process and potentially allow freedoms normally open only to those manufacturing their own work.
A timeline of technological developments considered relevant to this project can be seen in Figure 1
on gatefold pages 9 and 10.
1.4.1
Computer Aided Design (CAD)
Even with the advent of Direct Digital Manufacturing the computer art of previous decades would
defy production. For successful manufacture three-dimensional models need levels of definition and
integrity not present in the visualisation models of computer art and games. It is common for
example for surfaces in the virtual world to be infinitely thin, for them to pass through one another,
or for a regular surface to give the appearance of having a complex texture via linked twodimensional reference data. Computer Aided Design data by contrast, in particular that of high end
packages with an engineering rather than a visualization bias, offer robust models with more
watertight data.
CAD packages, running on personal computers (PC) rather than industry
mainframe based systems, have evolved from the basic drafting packages introduced in the mid1980’s to sophisticated 3D design systems capable of producing complex irregular forms with the
necessary integrity for production.
These packages allow the parametric definition of models.
Parametric CAD models enable the designer to set-up relationships that define the character and
function of a design; rather than identifying a single, discrete, design solution. Parametric design
defines relationships between degrees of freedom rather than specifying absolute dimensions. This
allows the whole form to update in reaction to the modification of a single element. In the context of
this project parametric CAD potentially allows the morphing model to be run within the CAD
package driven by external script data.
1.4.2
Rapid Prototyping, RP
It is primarily the advent, and the availability, of Rapid Prototyping (RP) that makes the physical
production from on-screen models a realistic proposition. It was this technology that formed the
basis for the FutureFactories concept. Rapid Prototyping is a catch-all term that applies to the
digital manufacture of prototypes directly from CAD data. The focus for this project is additive
fabrication, as the limitations of manipulating a physical cutter means that subtractive manufacture
can never offer sufficient flexibility for the reproduction of ‘freeform’ models. It is the appearance of
additive RP that has allowed the unfettered production of virtual forms. In additive RP, software
‘slices’ the CAD model into thin layers (down to 0.05mm). The model then ‘grows’ one thin layer at
a time with the machine ‘forming’ the build material only where it is required. Each data ‘slice’ is
replicated in three-dimension (3D) from the bottom up. The layers are built on a moving platform,
26
each built on its predecessor as the platform steps down in layer thicknesses. There is no tooling or
cutting away of material. This allows unlimited geometry. Forms may be produced that would be
almost impossible to mould or machine.
a) ‘High-End’ RP Processes
The high-end processes require industrial scale capital equipment.
The equipment can cost
between £200K and £500K. They require skilled operators and ancillary processes/stations.

Stereo Lithography, SLA (Stereo Lithography Apparatus)
Stereo lithography was the first of the modern layer build systems. Developed by 3D
Systems in the Eighties, it accounts for 45% of new machine installations (Wohlers 2003).
In this process each layer is ‘drawn’ by laser in a photosensitive liquid polymer; the laser
energy solidifies the liquid.
The advantage of this process is the surface finish; the
disadvantage is that delicate features and overhangs require thin supports that must be
removed post-production. A 3DSystems SLA machine can be seen in Figure 2.
Figure 2
A 3DSystems SLA machine
27

Selective Laser Sintering, SLS
Selective laser sintering is powder based.
The production chamber is maintained just
below the melting point of the build material. The layers to be built are then ‘drawn’ on the
fusible powder which is sintered by the laser. The finished part is formed within a cake of
un-fused powder which supports the features. The final step is to remove the loose powder.
SLS components are much more than appearance models, they have sufficient durability to
be considered functional. A wide range of powder based materials are available to the
process including nylon, stainless steel, synthetic rubber and ceramics. An SLS machine
from EOS, Germany, can be seen in Figure 3.
Figure 3
The EOS P390 SLS machine

Direct Metal Laser Sintering (DMLS) and Selective Laser Melting (SLM)
Metal sintering has been available since the start of the project, but high cost, limited
availability and poor surface finish proved a barrier to its use. Improvements in the process
and increased availability have brought them within reach for the later stages of the project.
DMLS is marketed by EOS of Germany, SLM by MCP in the UK. Both systems offer
production in titanium, stainless steel and cobalt chrome. Unlike the sintering of plastics,
where geometry is supported by loose powder, metal sintering involves the forces of
contracting metal which requires firmer anchorage. Geometry must be attached to the build
platform by support structure at the base. If material is built with overhangs of anything
more that 45 degrees to the vertical then additional support is required.
Any support
structure requires manual post production removal and is not the light ‘snap off’ material of
the resin processes such as SLA and the like.
28
b) Mid Range RP Processes
Processes such as 3DSystems Envision offer similar quality photo curable resin parts to SLA only
with the Laser replaced by a UV light source. This greatly reduces the capital cost to around 25%
that of SLA. A Further advantage is that any support structure can be built in a second soluble
build material rather than the primary resin itself automating support removal.
c) Budget RP Processes
Three-Dimensional Printing (3DP) is at the budget end of the RP spectrum.
The processes use
inkjet printing technology and the equipment costs are around a tenth of those for STL/SLS. Each
layer is formed by a jet of material or binder from a multi-nozzle piezoelectric print head. 3DP tends
to be used for visualisation rather than functional models.
Fused Deposition Modelling (FDM) is another budget process that uses a two axis extrusion head
(X,Y), over a vertically moving platform (Z) to deposit material. This is a relatively inexpensive
process that can yield components with good mechanical performance. Experimental projects such
as [email protected] (Malone 2008) have produced entire robotic devices including batteries and
actuators using this technology. As a commercial process however it is slow and the material
expensive (including a separate soluble support structure material), these factors drive up build
costs in spite of a relatively low capital cost. The finishes are inferior to those of the high end
processes.
1.4.3
Direct or Rapid Manufacture
Rapid Prototyping (RP) as a bureau service exists to provide models and prototypes in advance of
high volume industrial production. It enables complex forms to be prototyped and trialled without the
delay or cost of machining production tools.
Disregarding the technologies’ current expense;
designers, artists and commentators have experimented with, and speculated on, the use of RP for
end-use manufacture, this project among them. In 1997 Professor Celestino Soddu wrote that,
“Digital manufacturing technology allows one to realize, at the same operational cost, unique
objects or repeated objects. We have examples of this before us every day: a printer costs the
same to run whether it prints ten identical pages or ten different pages” (Soddu 1997). In 2003,
29
less than one year after the start of this project, the first Direct Manufacture consumer products
went on sale. The RP industry, having reached a sales plateau as the prototyping market reaches
saturation, has been keen to foster embryonic Direct Digital Manufacture or Rapid Manufacture; the
use of RP technologies for end use manufacture. In 2006, Time Compression Technologies, an RP
industry journal, dedicated its annual conference to ‘Rapid Manufacture’. At the conference Rapid
Manufacture was defined as, “The use of a CAD based automated additive manufacturing process
to construct parts that are used directly as finished products or components” (Hopkinson et al
2006).
30
1.5 Computer Generated Art
Computer generated art is limited only by the ability to render images on screen rather than any
physical manufacturing restrictions. The technology to computer-render 3D graphic images has
been available around twenty years longer than rapid prototyping. Of particular relevance to this
project are works involving the creation of 3D forms, albeit virtually, and systems for the automated
or semi-automated generation of ‘art’.
1.5.2
William Latham
The Artist, William Latham, was inspired by natural systems and how they often relied on the
repetition of simple, small, steps (Todd and Latham 1992). He sought to create art systems going
beyond two dimensional drawing systems, such as Russian Constructivism, to create ‘synthetic
organic forms’. His first system, FormSynth, was hand drawn. It featured an evolutionary tree of
sketches, Figure 4. Starting with a parent generation of regular forms, a series of children would be
created from each parent sketch by combining the form in some manner with other parents of the
generation. Highlighted in Figure 4 are two children from the cone parent, one with the addition of
the cube, the other with it subtracted. The children of this first generation would in turn become the
parents of the next and so on, creating an exponential growth in ‘solution space’.
Figure 4
Small Hand-Drawn FormSynth Tree, William Latham 1992
These hand drawn FormSynth ‘trees’ could run to 10m in length. Interestingly the artist sculpted
some of the ‘more aesthetic’ results in wood and plastic, indicating that the sketches were seriously
31
considered as 3D objects. The artist refers to the sketches, which were the hand drawn equivalent
of virtual models, as ghosts of sculptures.
1.5.3
Richard Dawkins’ Biomorph
Dawkins is a scientist and author rather than an artist: however, the ‘Biomorph’ system he created
to illustrate the principals of evolution and natural selection, is significant both as a work in itself and
for its influence on Todd and Latham. The Biomorph program was developed to point out the power
of micro-mutations and cumulative selection. The Biomorphs are simple line drawings, 2D graphics
made up of a series of straight line vectors.
They are generated by a recursive subdivision
algorithm. The recursive function calls on itself during its execution allowing the repetition of
features. The program begins with a simple line or, in Dawkins’ natural world analogy, a ‘trunk’
(Dawkins 1986). Subdivision allows the initially simple graphics to grow ever more complex with
vectors being divided and divided again by intersecting ‘branches’. A generation of alternatives is
produced by adding these ‘branches’ according to differing rules. From the range of alternatives
generated, one ‘child’ is selected by the user to be the subsequent parent and the process
repeated, Figure 5.
Figure 5
Generations of the Dawkins’ Biomorph, 1986
32
The subjective aesthetic judgment of the user thus plays the role of natural selection. Dawkins
expected trees to develop in a form reminiscent of the various species seen in the natural world.
He was surprised to see insect like forms emerging after only a few generations and described
these ‘creatures’ as ‘Biomorphs’: the name coined by Desmond Morris for the vaguely animal-like
shapes in his surrealist paintings, Figure 6. Starting from the ‘trunk’, there is a sub-branch
corresponding to each iteration. Each sub-branch is defined by a vector. Dawkins likened these
vectors, cumulatively, to a genetic code.
Figure 6
The Visitor, Desmond Morris, 1949
Featuring the “Creatures” He Termed “Biomophs”
1.5.4
Latham and Todd
For Artist William Latham, it was a logical step to employ computational power to drive art systems
such as his FormSynth (1.5.2). In the late 1980’s he began collaboration with mathematician and
computer graphics expert Stephen Todd. Using what was considered then to be the extensive
computational resources of IBM’s UK Scientific Centre in Winchester USA, Todd developed
FormGrow, a system that arranged geometric primitives, such as spheres and torii, to create
computer based virtual forms (Todd and Latham 1992). A system called Mutator was then
developed to explore the ‘solution space’ created using FormGrow. Mutator in concept was a
combination of Latham’s FormSynth and the two dimensional ‘Biomorph’ system of Richard
Dawkins (Dawkins 1986, as detailed in 1.5.3). “Biomorph demonstrates, in zoological terms, the
power of natural selection. Mutator harnesses this power, and extends it to make a fast and
effective exploration tool” (Todd and Latham 1992). The result is a ‘survival of the most aesthetic’
evolutionary development. From each new generation of mutations, Figure 7, one ‘child’ could be
33
selected by the artist. This subjective selection would be used to ‘steer’ the next generation of
evolutionary mutation.
Figure 7
A Set of “Children” from Mutator
Todd and Latham 1992
The similarities of this process to natural evolution have led to Latham being referred to as a ‘Digital
Darwin’ (Cook 1996). The driving force behind ‘Mutator’ was the creation of art. As Todd and
Latham stated, “Some artists feel that it provides a genuinely new way of working, and it has
certainly led to the creation of forms that would not have been created by other methods” (Todd and
Latham 1992).
Although the resulting ‘sculptures’ were only ever intended to be seen as 2D
representations of complex 3D models, presented as art in a gallery context, the principle behind it
can just as easily be used to create variations on ‘usable’ forms to produce designs for “anything
from buildings to shampoo bottles” (Computer Artworks 2003).
1.5.5
Karl Sims
Karl Sims studied computer graphics at the MIT Media Lab, and Life Sciences as an undergraduate
at MIT. For several years, in the early nineties, Sims was artist-in-residence at ‘Supercomputer’
Manufacturer Thinking Machines Corporation, Cambridge, Massachusetts.
In his 'Panspermia' series (Sims 1990), Sims created scenes with 3D ‘vegetation’ controlled by
genetic parameters. ‘Plant’ forms were 'bred' together with 'survival of the prettiest' determining the
evolutionary direction.
Sims uses a similar approach to Todd and Latham, with comparable
34
selection criteria. In this work however, there is an attempt to create plausible structure in terms of
human experience resulting in realistic yet alien landscapes. The results are quite different to
Latham's work with plausible organic vegetation and an air of realism, Figure 8.
Figure 8
A Still from Panspermia, Sims 1990
In his research ‘Evolved Virtual Creatures’ (Sims 1994) he created simulated Darwinian evolutions
of virtual block creatures. The form of these ‘creatures’ was represented by a hierarchy of rigid
primitives.
Elements of the creatures’ genotype code determines firstly the particular primitive
employed and secondly the constraints placed on the relative motion between this primitive and its
neighbour. In this work, evolutionary optimisation methods where used to develop pseudo-physical
skills, walking, swimming, jumping and the like.
1.5.6
Digital Sculpture, Digital Art Practice and Design Art
The work of Todd and Latham, Dawkins and Sims in which computation is employed to create 3D
forms capable of change over time, is of obvious direct relevance. There are many other artists
experimenting with digital techniques aspects of whose work are of interest and relevance.
35
Geoffery Mann graduated from the Royal College of Art (RCA), London, in 2005. He describes
himself as a ‘Product Artist’. Mann uses motion capture to record movement and reproduces it
digitally in a 3D artefact. Examples of his work include ‘Attracted to Light’, in which the path
described by a moth circling a light bulb is captured and produced as a physical form, as detailed in
Figure 9.
Figure 9
Attracted to Light, Geoffrey Mann 2005
Another piece by Mann is ‘Blown’ from the ‘Natural Occurence Series’. This is a cup and saucer
digitally distorted by a cooling breath on a liquid surface, Figure 10.
Figure 10
Blown, Geoffrey Mann 2005
36
Mann’s Blown is reminiscent of New York Artist Robert Lazzarini, who works with commonplace
objects such as a violin, a telephone and a skull. These objects are scanned, deliberately distorted,
and then reproduced digitally, Figure 11.
Figure 11
Teacup, Robert Lazzerini 2003
Belgian Artist/Designer Marcel Wanders, used motion capture to record the 3D form or airborne
mucus. The forms obtain where digitally manufactured as a series of vases, Figure 12.
Figure 12
Snot Vases, Marcel Wanders 2001
37
1.6 Commercial Examples of Digital/Direct/Rapid Manufacturing
At the inception of this project, direct digital manufacture was limited to a few niche examples. RP
processes were expensive and not widely available. The materials employed by these processes
were intended for short term visual representation and limited in terms of functionality. There were
examples of the Direct Manufacture of high value/ultra low volume components used in specialised
industries such as aerospace and motorsport. There were also examples of mass customisation in
the medical field, such as hearing aids built to precisely fit a particular patient’s inner ear.
Over
the period of this study, developments in this area have seen digitally manufactured products
available in the high street, to the extent that there is now scarcely a need to present a case for the
viability of direct manufacturing itself in a retail context.
The processes themselves remain
relatively expensive compared to conventional mass-manufacturing. However, the increasing
availability is producing more competitive pricing and within the RP bureau industry there is talk of
adopting ‘production’ rather than ‘prototyping’ pricing structures. It is a growing awareness of the
capabilities of the technology and the value that can be added to a product, in terms of complexity
and functionality, that provides a justification for higher production costs. The emphasis placed on
speed by service bureaus, which is such an asset to prototyping, often hinders the development of
Rapid Manufacture. The bureau industry is often characterised by feast/famine workloads and
large amounts of overtime. This does not lend itself to rapid manufacture where efficiency is vital
and profits are made over a longer term. Exotic new materials are being developed all the time; the
reality is however that these take time to filter though to the mainstream bureaus. They are often
expensive, having been developed for engineering performance, rather than to be cost effective.
Switching materials can be extremely costly often necessitating the replacement an entire reservoir
of material.
1.6.2
Materialise
Materialise was founded in 1990 as a joint venture with the University of Leuven, Belgium, and
became one of the first European rapid prototyping service bureaus. Materialise now has four
divisions. Three established divisions cater for software development, SLA technology, and medical
applications.
The fourth and youngest division, Materialise.mgx, is engaged in direct digital
manufacture.
The manufacturing capacity of Materialise places them at the forefront of the bureau industry.
Materialise has developed its own SLA machines, the Mammoth, with a build volume of
2100x650x780mm.
They have a unique, competitive, advantage amongst service bureaus
therefore, in that they use their own technology. This applies only to SLA production; however, for
38
other process such as SLS, third party machines are used. Currently SLS technologies tend to be
favoured over SLA for Direct Manufacture, as it uses more functional, durable materials and does
not require a support structure.
In 2001 Materialise began participating in an annual Dutch initiative, “Young Designers and
Industry”, forging links between design students and manufacturing. Students were encouraged to
explore the potential of Rapid Prototyping techniques with the offer of Materialise building the most
successful projects. In the summer of 2003 Materialise formed its own design division, MaterialiseMGX, to market a collection of lampshades, manufactured using SLA and SLS techniques: these
designs were the result of the Young Designers and Industry collaboration. Among the designers
were Alex Gabriel, Vince Vijsma, Koen Koevoets, future MGX Art Director Naomi Kaemfer, and
Freedom of Creation duo Janne Kyttanen and Jiri Evenhuis. The collection gained considerable
attention in the design press first at the Milan Furniture Fair, April 2003 and then under the MGX
brand at 100%Design London, October 2003, Figure 13.
Figure 13
Lilly, Janne Kyttanen 2003
Early MGX designs focused the aesthetic possibilities of the process and early designs were simple
lamp shades simply placed over a halogen bulb. Materialise had no experience of engineering
development in an industrial design sense. Several of the designs could even have been made
conventionally with development and really only exploited the cache of an exotic new technology.
Later collections began to exploit the technical possibilities of the process as well as the freedom in
form.
39
1.6.3
Freedom of Creation, FOC
Fin Janne Kyttanen and Dutchman Jiri Evenhuis both graduated from the Rietveld Academy in
2000. During the course of their studies they developed and filed a patent for 3D printable ‘fabric’.
The fabric consists of interlocked loops forming a chain-mail like structure. In 2002 they formed
Freedom of Creation. They have designed and produced artefacts for the Materialise-MGX lighting
collection and fashion items using their ‘fabric’ including handbags, mobile phone poaches, and a
sleeveless lady’s blouse, Figure 14. Their publicity material promotes the notion of mass
customisation with the concept of garments being built to fit, this service has not however been
offered commercially as yet.
Figure 14
Freedom of Creation V-bag 2004
1.6.4
Patrick Jouin
Patrick Jouin is an independent French designer.
In 2004 he presented the ‘Solid’ furniture
collection, a chair and a stool produced by SLA and SLS techniques respectively, Figure 15. This
collection is notable for the sheer scale of the pieces. The full size dining chair was produced in
one piece using Materialise’s Mammoth SLA machines. The chairs are in effect, gallery art objects
being extremely costly and in the case of the SLA chair, too fragile for effective use. The chair
requires a substantial amount of finishing as it is produced by Stereo Lithography (SLA) which
required support structures. It is hand finished and painted with a price tag around 30,000+ Euros
(Milan Furniture Fair 2005). The stool is a more practical proposition for direct manufacture. It is
40
produced by Selective Laser Sintering (SLS). It is robust and requires no post finishing. The largest
laser sintering machines at the current time, limit the build volume to 700 x 380 x 580mm; hence a
stool rather than a chair.
Figure 15
Chairs – Solid Collection, Patrick Jouin 2004
1.6.5
Bathsheba Grossman
Grossman studied sculpture and metalwork with Erwin Hauer and Robert Engman, mathematical
sculptors who were both trained by Josef Albers. In the late nineties, after several years practicing
traditional sculpture, she experimented with CAD/CAM and began designing sculpture digitally for
production by 3D printing. Grossman describes her work as “exploring the region between art and
mathematics” (Grossman 2005), her work features pure, mathematically derived, surfaces achieved
via computational design.
Grossman is an artist rather than a designer. Materialise, however,
manufacture lighting designs produced by her, Figure 16. These tend to be examples of sculpture
that have been lit rather than designed lighting products suggesting that function has been an
afterthought and perhaps indicative of differences in practice between the artist and the more
holistic approach required of the designer. The work is nevertheless extremely significant as an
example of computational design and algorithmically derived form.
41
Figure 16
Quinse, Bathsheba Grossman 2005
42
1.7 Mass Customisation and Individualisation
Mass production itself is a relatively recent concept. Prior to mechanised production artefacts
would be produced by craftsmen whose individual skills would be reflected in the product. Mass
production depends on uniformity. Since the worldwide adoption of the mass production model, the
goal of manufacturing has been accurate repeatability. Mass production has made desirable
objects affordable. The size of the market allows levels of design development and the use of
sophisticated processes not possible at lower volumes.
There is however a perception that
something has been lost. In today’s consumer world we are surrounded by every conceivable
product for every possible application, all at affordable prices. This availability, and the
omnipresence of mass-merchandise, fosters within us a desire for something personal and unique.
The term, 'Mass Customization’, was coined by Stan Davies, in his book ‘Future Perfect’ (Davies
1987). The term is deliberately paradoxical: 'Mass Customisation' can be defined as “a delivery
process through which mass-market goods and services are individualised to satisfy a very specific
customer need at an affordable price” (Davies 1987). Based on the, “Public's growing desire for
product personalisation, it serves as the ultimate combination of custom made and mass produced"
(Fu 2002). There are many different models for mass customisation; suiting different products and
market sectors. They are all, by definition, consumer driven. Customisation may be achieved
through the combination of options; for example, selection from an extensive but finite range of
colours and finishes. Alternatively, the consumers may provide data on personal preferences or
accurate measurements of body parts, to enable the production of a ‘tailor made’ product.
Consequently, examples of mass customized products range from genuine medical 'needs', such
as perfectly fitting hearing aids (Fu 2002), to desired product differentiation in a kitchen stove or
better-fitting bespoke jeans (Marsh 1997).
1.7.2
Ron Arad
Ron Arad, in collaboration with Geoff Crowther, Yuki Tango, and Elliot Howes, presented, ‘Not
made by hand, Not made in China” in 2000. This collection featured a range of artefacts
manufactured using RP techniques. The project publicity spoke of ‘growing’ products as a new,
fifth way, of manufacture after:subtractive methods, such as carving or machining;
moulding;
forming, such as bending or pressing; and
assembly.
43
One of the projects was “bouncing vases”, using Maya, 3D software aimed at the video animation
industry. A helical vase form was created, that was expanded and compressed in an animation
clip, Figure 17. “We can take a model, throw it in the air, let it drop, animate and calculate the
distortions, thus producing hundreds of frames, each a mature model ready to be “grown” into a
real object” (Arad 2000).
Figure 17
Bouncing Vases, Ron Arad 2000
The impression in the work is of a fixed model that is acted upon by external forces that impose
deformation. The model is ‘thrown up’ and then ‘bounces’: it is being virtually customized. The
subsequent changes are predictable, albeit they may be complex. The bouncing vase itself is
designed in the form of a helical spring, to deform in a particular way as with a spring or bellows.
The range of possibilities is readily apparent and is little different to scaling the overall form.
The
concept of designs generated from an animated model is non the less highly significant.
1.7.3
Professor Celestino Soddu
Professor Celestino Soddu has written extensively on the use of generative design in product
design and architecture, since the 1980’s. From 1998 he has organised the annual international
conference ‘Generative Art’.
In the late 1990’s, Professor Soddu recognised the potential of emergent rapid prototyping
techniques to facilitate the Direct Manufacture of artefacts directly from computer based data. In a
paper given at the European Academy of Design Conference Stockholm, Sweden, in 1997, Soddu
spoke of a return to unique products in a post-industrial era; made possible by a combination of
44
generative computer models and rapid prototyping techniques (Soddu 1997).
In 2001, as
Professorial Research Fellow at Hong Kong Polytechnic University, he claims to have created
software that enabled the direct output of unique objects via Rapid Prototyping equipment. In his
Stockholm paper, Soddu points out that, “Before the industrial era every object was unique,
unrepeatable, and strongly connected to the identity of its maker or user” (Soddu 1997).
He put
forward the notion that in the post-industrial era the values of mass production were no-longer valid
and that mass production would no-longer be more cost effective than individualised computerdriven manufacture. This prediction on cost was, and remains, over optimistic. It is extremely
unlikely that in the near future Direct Manufacturing will be viable in high volume applications.
Although it is already proving viable in low volume niche applications and is likely to become
competitive in medium volume applications in the years to come (Wohlers 2004).
Soddu points out that design optimisation need not necessarily lead to a single solution and that the
‘quality’ of a design is not the final result in itself. He discusses the subjectivity of both designers
and consumers in regard to what is considered ‘necessary’ and ‘optimal’ and the impoverishment
that has resulted from a disregard for the quality of uniqueness. Soddu describes how ‘ideaproducts’ could create a new market with industry buying into an ‘endless sequence of
automatically generated 3D models’ (Soddu in Bentley & Corne (Eds) 2002).
The work is fascinating and ground-breaking. From a product design perspective however it is only
the start. The computer generated ‘product’ forms that illustrate Soddu’s publications are figurative,
Figure 18. They collections of primitive volumes representing for example, the lid, body, handle
and spout of a coffee pot. These elements are then transformed and configured in different ways.
Soddu’s stated goal of a ‘recognisable design’ does not seem to be achieved in product design
terms.
Figure 18
Computer Generated Chair Designs, Celestino Soddu 2001
45
There would not appear to be little difference between the results shown and those that would be
expected from random configurations. The chair examples shown are recognisable as a ‘species’,
only by virtue of a common chaotic make-up of random primitives. There appear to be few technical
rules and little guidance or targeting. Functionality beyond the conceptual does not appear to have
been a consideration in the work. In an example of coffee pots, there is little regard to function
beyond the provision of an appropriate volume and spout.
These are virtual designs. They
potentially could be translated into real-world artefact but this would require substantial postprocessing prior to any build. These are virtual designs The digitally manufactured output produced
by Soddu in 2001, are small scale, appearance models, Figure 19. It is of interest and significance
that theses pieces were generated automatically. They are however, ‘proof of concept’ rather than
any form of prototype and are considerably less sophisticated than the vases and lamps presented
by Arad in 2000 (Arad et al 2000). The lack of design in perhaps borne out by the absence of any
functional outputs in the intervening years. The technical resolution of the output however does not
detract from the significance and prescience of Soddu’s work at a concept level which is clearly
ahead of its time.
Figure 19
Digitally Manufactured Chair Models, Celestino Soddu 2001
46
1.7.4
Fluidforms
Austrian Fluidforms describe themselves as a fusion between designers, artists and programmers.
Fluidforms offer the consumer the opportunity to create household items (vase or salt/pepper
grinder), “according to his own visions” (Fluidforms 2005). This is achieved via an interactive web
based experience. The user is able to adjust, via on-screen click and drag, the cross section of an
axi-symmetric form. Control points on a cross section curve can be displaced in 2D modifying the
form. Figure 20 shows two iterations, the pink control points may be dragged in both X and Y, the
green in x only. The limited number of control parameters and their restricted range (14mm on the
radius) limits creative expression. The artefacts are built in CNC cut wood laminate and hand
finished. The customised element is essentially a decorative sleeve that houses a proprietary
grinder mechanism or glass vase. This limits, to a large extent, the scope of what can be achieved.
Despite these limitations, Fluidforms is an example of true mass customisation. The design can be
customised and ordered over the internet for a cost of 90 Euro (2008). The company passes the
creative role to the consumer, “Fluidforms hands this role over to the customer. Thus, everyone
can become a designer” (FluidForms 2005). There are set limits to the individual parameters but
no relationships between them and hence no design intent is maintained beyond overall
dimensions and a sacrosanct inner volume.
This is an intriguing step forward in mass
customisation as it is web based and allows interactive adjustment of form, rather than merely
selecting options. From a design perspective however, it is reliant on the consumer for creative
decision making. It is only the limited scope of adjustment that prevents ‘undesirable’ outcomes.
Figure 20
Serene Pepper Grinder, Fluidforms,
Customisation Screen Shots, 2008
47
1.7.5
Front
An extreme example of Customisation is the Sketch Furniture project by Front, a Swedish Design
Collective formed in 2005. Front have a reputation for subverting the norms of the design process
with ‘products’, like wall paper, made by the gnawing of rats, Figure 21.
Figure 21
Rat Wallpaper, Front, 2006
At Tokyo Design week in 2006 the group showed the Sketch Furniture Project with furniture ‘drawn’
by motion of the hand in ‘thin air’, Figure 22.
Figure 22
“Sketch” Furniture Creation Video, Front, 2005
48
The movement is captured digitally and used as geometry to define an extruded surface. The virtual
form created can then be built by RP technology, Figure 23.
Figure 23
Chair - Sketch Furniture Collection, Front, 2005
The video is extremely engaging; the 3D output however is extremely crude. There is no control:
the ‘designers’ work blind, unable to see where their hand is in relation to what has gone before.
The work is in essence similar to the work of glass and ceramic artist Tavs Jørgensen. In his
Motion in Form project Jørgensen uses a data glove, Figure 24, to describe artefact features, such
as the lip of a glass vessel, in space.
Figure 24
49
Jørgensen’s Data Glove
The results have been produced in glass and ceramic via CNC (Computer Numerical Control)
milled tooling. Ironically, the ‘artist’ in this case puts considerably more control and constraint into
his work, Figure 25 , than do the Front ‘Designers’ whose work is essentially performance. It might
be expected that a design process would put greater emphasis on functionality and allow less
freedom. Jørgensen however is using digital technology to capture the authentic momentary
gesture of the artist rather than as a development tool to refine and perfect his ideas towards some
notional ideal. The significance is that the vessel rims bear witness to discreet acts of the maker.
The usual meaning of the word “craft” opposes high-technology processes in which the hand plays
a diminished role (McCullough 1996). Digital technology is to the fore in this project yet the ‘hand of
the maker’ is evident and celebrated despite the physical separation of action from the
manufacture.
Figure 25
One-Liner Glass Bowls, Tavs Jørgensen, 2008
50
Sketch furniture, with its crude extrusions, is also reminiscent of artist Roxy Paine’s Scumak No.2.
In this experimental automatically generated art, plastic was extruded via a randomly controlled
CNC (computer numerically controlled) nozzle. The extrusion would be deposited onto a conveyer,
which periodically would index along to begin the next piece, Figure 26.
Figure 26
Scumak No.2, Roxy Paine, 1998
51
1.8
Software Developed for Generative Design
1.8.1
Genometri – Generative Design Software
Genometri is a Singapore based design technology company which markets generative design
tools to the design industry, principally product design and architecture. Genometri was a start up
company in 2005, a commercial technology spin-off from the National University of Singapore,
which received funding from the Singapore Government.
Genometri developed a plug-in for the SolidWorks solid modelling package called Genovate. The
idea behind Genovate is that it generates any number of random variants from a given parametric
CAD model. The premise being that a busy design studio does not have the time to consider every
possible permutation. According to the founder, Dr Sivam Krish, “It allows the rapid generation of a
vast number of designs based on a generic model.
It is able to explore a larger set of design
possibilities than what is manually possible today” (Krish 2005). Significantly, Genometri uses the
kernel of a parametric CAD package to control the geometry of the model, as has been proposed
with FutureFactories (Unver et al 2003 Appendix 1) (Dean et al 2005 Appendix 2). The Genometric
software has been developed as a plug in specifically for Solidworks CAD software. Theoretically it
could be developed to suit any parametric package. The fact that it is only available on a single
platform currently is something of a limitation. While it is common practice to export geometry from
one package to another, with varying degrees of success, Genometry works with variables
contained in a construction history. This effectively requires that geometry be built in Solidworks
rather than use imported elements.
A reasonable level of competence in a specific software
package is therefore required. The idea is that a Solidworks file can be imported directly into
Genovation and ‘Genovation’ can begin immediately.
All named dimensions are reassigned
random values. The range of variance can be controlled via a slider bar that specifies a maximum
percentage change (relative to the original parameter value) across all variables or by manually
setting limits for individual dimensions. The iterations are wild, random jumps contrasting sharply
with the subtle changes over time envisaged by the researcher. Seismic changes bring issues of
control and stability which will be discussed, 4.3.4 p81. There is also an issue of presentation.
Genovate is aimed at generating design options in the commercial context. Output is all important
and there is little need to engage the user with the generative process or to reassure them that
control is in place since they themselves have established the parameters. It is interesting to note
that the pop-up screen for setting limits manual is named ‘live-DJ’ with the connotation of ‘remixing’
creative output.
52
Genovate avoids issues of functionality: it is a plug-in which adjusts parameter values within a predefined model. There is nothing to constrain the form beyond the geometry limitations of the input
Solidworks model which when exceeded, as frequently happens on a model of any complexity,
causes the iteration to fail. The generative software operates purely on the CAD model’s data set
and is blind to the significance of particular features. Each parameter is treated independently.
There is no provision for linking features as has been considered necessary in this project to
maintain control over the form (Atkinson et al. 2003, Appendix 3). The character of a design often
centres on proportional relationships. If for instance, a feature ‘X’ is small, the possibilities for
feature Y might be restricted. Should this feature X be larger this in turn might offer more freedom
in feature Y.
Genovate proves the virtue of using a parametric CAD kernel to control an animated 3D model and
proves that applying mutation to such a model can be relatively straight forward. The only control
the designer has over the character of the design is via range limits. The greater the range, the
less likely a random change will be acceptable. It is interesting to note that the extreme end of the
range slider bar is labelled ‘creative’, this implies that cruder adjustments are equated with
increased creativity. Provision for setting relationships between parameters would enable greater
control of the design’s character.
In 2005, the researcher established links with Genometri and undertook beta-testing. As a result
Genometri developed a dedicated version of the software, that allowed the storing and modification
of parameter ranges in a Microsoft Excel file. Cornuta (8.1) was produced using this software: in
this work relationships between parameters were introduced via formulas set in the CAD package.
1.8.2 Bentley Systems, Generative Components, and the Smart Geometry Group
Bentley Systems is a software developer. The company has developed Generative Components,
released for beta testing in 2005, a software package for exploratory architecture that combines
CAD (Bentley’s own Microstation) with computer scripting. Large numbers of architectural firms are
experimenting with scripting options to embed mathematical models within CAD packages, to
digitally re-shape a form in response to loads for example (Hesselgren 2004). The Smart Geometry
Group is a non-profit organisation promoting computational and parametric approaches to design,
principally in architecture.
Lars Hesselgren architect, researcher and founding member of the
53
Smart Geometry Group comments that, “It moves the decision as to what is “architecture” versus
what is “engineering” from the software vendor to the user. Just as with the adoption of
spreadsheets 20 years ago, it is again the user who creates meaning, not the software.”
(Hesselgren, 2004). In a presentation to the SmartGeometry Conference, January 2006, Lars
Hesselgren spoke of Computational Design, design practice driven by rules and relationships, as a
development of Digital Design.
Digital design was compared to digital word processing,
computational design to the spreadsheet such as Microsoft Excel. In the case of the spreadsheet,
the importance is not the initial values entered, but the relationships into which they feed. The
SmartGeometry group, founded by four key industry experts; Robert Aish (Bentley Systems), Hugh
Whitehead (Foster and Partners), Lars Hesselgren (KPF) and Jay Parish (Arup Sport) is in the
process of registering itself as an educational charity, sponsored by Bentley Systems, with the aim
to further advance education and research in the area of advanced 3D CAD applications.
Generative components aim to offer parametric modelling at a much more sophisticated level than
is found in conventional parametric CAD packages. Conventional CAD functions are combined with
scripting, to achieve more complex parametric relationships in what Bentley Systems terms
‘Programmatic Design’. Robert Aish, Director of Research, for Bentley states that the software can
be used to capture design rules in a logical form and that, where programming skills are available,
these can be harnessed to create additional programmatic components, without requiring a
complete application to be written (Aish 2004). As with Genometri, much emphasis is placed on the
rapid generation of variants. “In real-terms that could provide a practice with the ability to come up
with 20 or 30 designs simply by moving sliders within a Generative Component model” (Day 2005).
The suggestion is that project specific parametric rules can be set down to define architectural
details and that these details can be rapidly applied in widely differing formats all bearing the same
‘character’. This is maintaining the designers’ intent as discussed in this thesis. It is easy to see
that, disregarding focus on architectural applications, rules could be set to carry the visual language
or the ‘genes’ of a product ‘species’. It is interesting to note the importance attributed to the
interface. The need for an intuitive set-up interface was identified in this project (Unver et al 2003
Appendix 1)
54
1.9 Literature Review Summary
The Literature review examines work in generative art and design and direct digital manufacturing.
The generative artwork of Todd and Latham includes speculation on the automated generation of
3D designs. Soddu goes further predicting emphatically the concept of individualised production
brought about through the combination of generative computation and direct digital manufacturing
He demonstrates extensive families of computer generated virtual products albeit that they are
rather random and technically unresolved. If Soddu’s output lacks functionality this addressed in
Arads ‘Not made by Hand, Not Made in China’ which yields functional product iterations from CAD
animations.
Different levels of consumer interaction are explored. Soddu states that his works are ‘not tools for
designers’ (Soddu in Bentley & Corne (Eds) 2002) where as at the other extreme Fluidforms would
like the user to create products ‘according to his own visions’ (Fluidforms 2005)
The concepts and predictions of the generative research considered are given weight by the
emergence of digitally manufactured retail products documented in this section. These products
produced in respectable volumes for retail sale indicates the commercial viability of such scenarios.
Examples of work from artists and designers demonstrate how virtual artefacts can be manipulated
virtually and physical and 3D artefacts generated from the data. These works highlight to
possibilities for ‘living’ products free from rigid specification.
55
56
2.0
Project Overview
In 2002, The School of Design Technology at the University of Huddersfield, took the decision to
allocate an amount of research funding to provide an ‘Artist-in-Residence’ to work alongside Fine
Art students, and a ‘Designer-in-Residence’ to work alongside Product and Transport Design
students for a period of one year. Project proposals were invited, in April 2002, for the post of
Designer in Residence. The researcher presented an embryonic proposal ‘FutureFactories’ for
product manufacturing, using the then relatively new technology of Rapid Prototyping, and was
subsequently appointed Designer is Residence for the Academic Year 2002/2003.
2.1
Design Residency Program
The concept of an ‘Artist-in-Residence’ has been part of the international art world for over a
century. Residency programs may serve as a kind of patronage, offer a utopian seclusion, or
facilitate an engagement with a possibly aesthetically impoverished public or business. Their
presence in institutions, business and the arts is familiar. In art education the benefits of following
an experienced practitioner is recognised and valued. A Designer in Residence working within
design education is however somewhat unusual.
Fine art and craft traditions often feature specific manual skills which are accessible and can be
readily observed even if more profound aspects of the practice remain elusive. Craft practice is
usually workshop based and therefore placing a working professional practitioner alongside
students is relatively straight forward. In design for industrial manufacture, the practitioner’s skills
tend to be wide ranging and encompass managerial activity as conflicting design criteria, for
example mechanical function, user interaction, industrial production, environmental impact and
aesthetics, are balanced against each other. Product design can involve a variety of activities
from manual skills such as sketching and model making, to knowledge based activities such as
CAD and calculations. Whilst the design is progressively refined in the course of a project, this is
only visually evidenced at key presentation stages: there is rarely the continuous refinement of a
physical piece. Development issues and decision making may be opaque requiring detailed
explanation, rather than being apparent from observation alone. Accommodating a practicing
designer is therefore potentially more of a challenge. Digital design practice is particularly difficult
as the majority of the work is screen based. There may be no manually based practice beyond
the use of a computer mouse. There may be test-rigs, printouts and diagrams but there need not
be and in the researcher’s practice such physical evidence of the creative process is increasingly
unnecessary and rare.
57
The potential impacts of digital manufacturing, and the opportunities it presents, were considered
to have sufficient significance to outweigh presentational barriers to the work. Paul Atkinson
commented that, “The (FutureFactories) project was fresh, exciting and potentially stimulating for
students to see unfolding, the approach to design was particularly suited to the school as it
combined theory and practice in a balanced way” (Atkinson 2006).
For the academic year
commencing Autumn 2002 the researcher was based part-time at the University of Huddersfield,
working alongside Product and Transport Design students in an open plan studio space. An
open sided booth was provided with CAD workstation and wall space for flatwork. In addition to
the day-to-day presence of the work, a series of presentations were arranged to allow a fuller
communication of project thinking. One aim of the project was to encourage students to consider
the implications of digital manufacturing within design practice and education. In addition to
stimulating this theoretical and philosophical debate, the residency project aimed to benefit
students at a more pragmatic level, by providing an insight into contemporary professional
practice, highlighting the project management and time planning required.
The Design Residency program was for a fixed period of one year. Target outputs for the project
were outlined at the proposal stage; these included culminating the residency with a public
exhibition. This exhibition would include digital displays of the ‘system’ in action and a collection
of physical products generated from it. It was considered important that actual artefacts be
displayed to distinguish the work from virtual computer art and to engage a wider audience
beyond the digitally aware. The focus of the project was the creation of design systems rather
than discreet product design types. The exhibition would therefore feature a range of different
consumer products. Presenting a range of designs would also enable exploration of alternate
approaches, geometries, processes and materials. Early in the residency period it was decided
that the concept was generating sufficient interest to merit a small touring exhibiting; taking the
material beyond the University and opening up debate to a wider Art and Design community.
As well as the practice based elements of the project, the researcher was involved with research
activity in collaboration with Huddersfield academic staff.
During the residency period this
academic activity was reported on in a paper to the 5th European Academy of Design
Conference, Barcelona, April 2003, which considered cultural and pedagogic aspects of the
project (Atkinson et al
2003 Appendix 3).
A second paper presented to the International
Conference on Advanced Engineering Design, Prague, May 2003, considered the technical CAD
modelling and programming elements (Unver et al 2003 Appendix 1).
58
2.2
Expansion of the Project into a PhD Study
As the residency program developed it became clear that the work was of a significance that both
merited and required investigation beyond the one year program. The project’s first presentation
outside the University, at the European Academy of Design Conference, Barcelona, April 2003,
demonstrated a significant level of academic interest in the work. In the same month the Belgian
Rapid Prototyping service provider, Materialise (1.6.2), unveiled a collection of lighting design
prototypes. In September 2003, these designs where launched as commercial retail products
under the company’s own brand Materialise.mgx. This proof of the economic viability increased
the relevance of the work and created an imperative to complete and disseminate the study. In
Spring 2004 the researcher formally enrolled on a practiced based PhD study. Project outputs
have been exhibited regularly, both nationally and internationally, since the conception to-date.
There has been increasing overlap with the researcher’s professional design practice. A number
of commercial retail products have directly or indirectly stemmed from the work and limited edition
gallery pieces have been created under the FutureFactories ‘label’. The researcher’s work is now
focussed exclusively on Direct Digital Manufacturing.
2.3
Justification of the Practice Based Elements of the Project
Contemporary consumer products are almost invariably mass produced, and are visually
identifiable as such through consistent detailing, regular surfaces and high quality surface
finishes. Craft practice has an altogether different aesthetic. It is perhaps easy to imagine
individualised craft output, as variations are often inherent in hand making: industrial processes
are required to achieve standardisation and uniformity. Mass produced articles, by contrast, are
expected to be exact facsimiles with variation considered a flaw. The broader ‘systems design’
element of the study can be illustrated and discussed through comparison and analogy.
Comparisons can be made with the natural world, pre-industrial production and contemporary
craft. It would remain difficult however to imagine the impact of variation on artefacts hitherto
defined by their uniformity. For instance, when the project considers ‘variation in form’, what
degree of difference would be involved? High-end Rapid Prototyping equipment can achieve a
resolution of 0.05mm. Minute variations could be introduced that would not be visually apparent.
This would technically be in line with the concept but would not be in the spirit of it. Artefacts and
a practice-based element to the work is required to fully explore the implications of the concept.
Additive manufacture will remain an expensive process compared to the conventional moulding
processes of mass manufacture for the foreseeable future. The commercial viability of
individualisation using this technology depends on the premise that variation brings some form of
59
added value. This is likely to require more than a technical difference; at some level even
‘identical’ mass-produced artefacts will exhibit differences as a result of manufacturing tolerance.
Would differences in chemical composition, for example, hold any cache? It would seem unlikely.
What degree of change then would be required to capture consumer interest and can this need
for obvious change be accommodated while retaining a brand or design identity? Value and
appeal are abstract concepts that can only be effectively examined through experimentation and
dissemination. Technically a system can be defined on paper: the practicality, desirability and
likely impact of such a production model can only be effectively debated through practical testing.
2.4
The Researcher’s Established Practice
The researcher had 15 years of experience as a professional industrial designer prior to this
thesis. He trained and practiced as an automotive designer before developing a specialism in the
design of lighting objects.
Computer Aided Design was not widely available during the
researcher’s training although systems were demonstrated.
An undergraduate training in
automotive engineering however and in particular the Cartesian referencing practiced in the
automobile industry, made for an east transition to desktop CAD when it arrived.
The researcher’s practice, in common with standard practice in the Industrial Design industry,
would be to commence a design project with concept sketchwork.
Loose drawings would
become increasingly defined to the point where dimensionally accurate orthographic projections
could be drawn and scale models made. The researcher’s technical background always favoured
accurate definition and complex curvature would habitually be defined by orthographic cross
sections: a practice that has relevance in the CAD techniques developed in the thesis. In general
terms the methodology would be to set out loose ideas that captured a desired aesthetic and then
to iteratively refine the concept making it increasingly viable.
60
3.0
Mass Individualisation: Industrial Scale Production of One-off Artefacts
3.1
Why Individualise?
Mass production is a relatively recent concept characterised by the production of large quantities
at low cost per unit. Rather than dictating low budget inferior products, mass production can
allow precision manufacture and quality control. The size of potential markets allows intense
design development and the use of sophisticated processes not economically viable for lower
volumes. Mass production has made desirable objects affordable.
Standardisation,
rationalisation and uniformity are used to achieve a level of repetition that allows both affordable
pricing through economies of scale and quality through technical refinement.
The economics of mass production is dependent on high volume manufacture, supported by
mass-consumption. To be appropriate for mass-production therefore, products must have mass
market appeal.
Since the worldwide adoption of the mass production model, the goal of
manufacturing has been for accurate repeatability. By definition, mass-market designs become
commonplace and unexceptional.
Pre-industrialisation artefacts would be produced by craftsmen whose individual skills would be
reflected in the products. Products would have a strong connection with the user and maker.
The artefacts produced would be bespoke interpretations of a design ‘type’, recognisable yet not
facsimiles. Each artefact produced could be more or less faithful to this original ‘specification’. A
level of variation would be inherent in the process with makers often working ‘by eye’. The design
specification itself might be ill defined and open to interpretation, organic, developing and
mutating over time.
The material stock employed in the design might be non uniform and
adaptation might be required to work around a flaw or blemish. The manufacturing process may
not be consistent with many craft processes being a balance between the demands made of the
process and the control of it, for example, hand-blown glass.
Such discrepancies in interpretation, material or process, rather than resulting in scrap, might
produce an interesting variation on a familiar theme. This lack of uniformity, this uniqueness and
un-repeatability, far from being seen as a negative by the consumer, is often valued. Mass
production has made desirable objects affordable. The size of the market allows levels of design
development quality and technical sophisticated processes not possible at lower volumes. There
is however a perception that something has been lost.
In today’s consumer world, we are
surrounded by every conceivable product for every possible application, all at affordable prices.
61
The availability and omnipresence of mass-merchandise fosters within us a desire for something
personal and unique, something we can imbue with a soul or character of its own.
An aim of this thesis is to deliver the technical resolution and affordability of mass market
products, combined with the idiosyncrasies of craft production: to reintroduce the individuality
inherent in hand manufacture whilst still exploiting the economic and technical benefits of a massproduction.
3.2
Industrial Production Versus Craft Making – The Need for Automation
The distinctions between craft and design can be complex. Perhaps simplistically and for the
purposes of this discussion, ‘Craft’ will be defined as production dependent on the skills of an
individual and ‘Industrial Design’ as a process in which artefacts are defined for production by any
“appropriately skilled workers”, usually employing automated equipment.
CAD software is a production tool that can be used to effectively hand-craft one-off artefacts,
albeit indirectly with a mouse or similar input device.
manipulated and edited with relative ease.
Virtual three dimensional models can be
Systems can be envisaged in which a creative
practitioner would manually adjust a virtual model in response to each and every order received.
The aim of this project however is an automated design and manufacturing system. If a designer
is required at the point of production, then the outcome becomes dependant on a particular
individual’s skill and the production capacity limited by the time they have available. This is in
contrast to the industrial design model, in which the designer’s skills are required only to define
an article and its manufacture. Once the design is complete, manufacture becomes mechanistic,
perhaps requiring skills and training, but not dependant on an individual’s creativity. If the CAD
manipulation can be automated, production becomes viable on an industrial scale, and therefore
not dependant on individual skills. The aim of this project is a design process with the capacity to
run on an industrial scale and an automated process is therefore essential. Within the context of
this project, mass-individualisation is defined as the automated industrial scale production of oneoff artefacts.
62
3.3
Mass Individualisation Distinct From Mass-Customisation
The term 'Mass Customisation’ was coined by Stan Davies, in his book ‘Future Perfect’ (Davies
1987).
Mass customisation offers personalisation though the use of flexible manufacturing
systems without losing the economic advantages of mass production; “Producing goods and
services to meet individual customers’ needs with near mass production efficiency” (Tseng and
Jiao 2001). Kaplan and Haenlein define mass customisation as, “A strategy that creates value by
some form of company-customer interaction at the fabrication/assembly stage” (Kaplan and
Haelein 2006). This interaction normally involves a level of specification by the consumer; either
passively through the use of body scan data for example, or actively through some form of
configuration.
In contrast, FutureFactories derives only limited input from the consumer. The user can merely
arrest an ongoing development, or initiate a new one. Crucially, the nature of the form generated
is independent of any consumer interaction other than this crude stop/start. There is no attempt
to configure an artefact to specific requirements or taste. The aim is merely to generate, on an
industrial scale, a series of unique design iterations and thereby to give each customer the
satisfaction of owning a one-off artefact. The FutureFactories model is therefore more correctly
termed ‘mass individualisation,’ defined in this context as, “the production of unique product
variants with near mass-production efficiency.”
3.4
Consumer Input to the Individualisation
FutureFactories achieves individualisation by introducing elements of random variance at certain
stages in the design process. Crudely, certain parameters and sets of parameters are allowed to
‘float’ within their own predetermined range. It might seem logical to allow the consumer to make
these adjustments via a ‘slider-bar’ style interface? The researcher, however, has sought to
avoid this. The desire for customised merchandise can be a search for something unique, in a
mass market culture.
product design.
It does not necessarily result from dissatisfaction with contemporary
Indeed it is the very success of particular designs, their proliferation and
omnipresence that fosters the desire for something more personal. The desire is not necessarily
for something designed by oneself personally, dependant on whatever skills and experience one
has, but for something individual and personal. The intention of this research is to introduce
individuality into “off the shelf” mass produced artefacts rather than to empower the consumer as
a designer/maker themselves. Such an empowerment would be worthy of study and there are
several projects and companies working in this direction, for example Genometri (1.8.1) and
Fluidforms (1.7.4). FutureFactories is concerned with the boundaries and nether ground between
63
industrial repetition and the bespoke. Practitioners are often commissioned to work to a client’s
individual instruction and a case could be made for some element of co-design. However, for the
sake of clarity, and to draw a clear distinction between FutureFactories and mass customisation,
the decision was made not to allow consumers into the design process, other than to initiate or
halt its development.
64
3.5
Random Morphing,
M
Co
ontrol or Hap
ppenstance
eFactories re
equires adjus
stments to th
he design sp
pecification fo
or each and every artefact
Future
produ
uced. Individ
dual variants cannot be created by the
e designer a
as an automa
ated, industrial
system
m is required
d. There is also
a
is no des
sire to assign
n a creative rrole to the co
onsumer. Th
he
solutio
on is to introd
duce an elem
ment of compu
uter generate
ed variance; a random ele
ement. Centrral
to the
e project is that
t
this com
mputer genera
ation remain s only an ellement within
n a controlled
d,
recog
gnisable, desig
gn form. The
e aim is coherent, identifia
able designs rrather than ra
andom objectts;
artefa
acts compara
able in style and design to mass pro
oduced consu
umer productts.
Rules arre
requirred to ensure
e that each and
a
every solution maintaiins a desired
d aesthetic an
nd is thereforre
designed, rather th
han randomly generated.
n
of there
e being a sing
gle optimal so
olution and un
nique physica
al specification
n for any give
en
The notion
design concept is something
s
off a myth. Ofte
en design is tthe managem
ment of comprromise, tradin
ng
s against eac
ch other. This is particula
arly true wherre product ae
esthetics are a
off diffferent factors
driving factor and there
t
are arb
bitrary decorattive elementss; taking the F
FutureFactoriies Holy Ghost
chair as an example, Figure 27..
Figure
F
27
Holy Gh
host chairs, 20
006
H
Ghost ch
hair back is made
m
up of a number of ‘b
e weight of th
he occupant is
buttons’. The
The Holy
share
ed between these buttons and
a there has
s to be a sufficcient numberr to share the loading from a
structtural perspecttive. From a visual perspe
ective, howevver, the aesth
hetic is not de
ependant on a
65
fixed number of buttons although clearly there would be limits. The exact number of buttons is an
arbitrary design decision. Accordingly the computer script underpinning the Holy Ghost design
has been developed to generate a range of between 20 and 26 buttons.
As designer, the
researcher considered that the character of the design was effectively maintained throughout this
range.
3.6 Meta-designs
In the FutureFactories model the designer creates what has been termed a design template
(Atkinson et al 2003 Appendix 3). Rather than specify a discrete design solution, the designer
sets up a series of rules and relationships that define desired aesthetic and functional criteria over
a potentially infinite range of outcomes. This meta-design is equivalent to the organic design
specification of artisan making. It allows a certain freedom in form whilst maintaining an overall
design intent.
The template should allow a balance between freedom and control.
Whilst
maintaining a coherent design identity there should be obvious difference between iterations and
even the potential for a surprising twist.
66
4.0
Computer Generation of Variants
FutureFactories began as a one year residency, which was to culminate in a series of exhibitions
in the Yorkshire region. One of the key demands in disseminating project thinking was the
creation of real-world consumer products to illustrate what could be achieved. There were, and
are, few close precedents in the commercial realm. Creating 3D outputs from the project required
both product designs and software solutions to drive and control the generation of variants. The
need to exhibit 3D outputs after a comparatively short period of development meant little time was
available to develop the underpinning computer scripts. As the potential of the project to develop
beyond the blue-skies residency became clear the demands for dissemination increased. In
order to exploit interest in the work, communication was considered key and prioritised producing
presentable artefacts over writing lines of code. As the project has become established the
software development has gradually caught up. In the later design examples the project aim of
real-time individualisation is achieved and developed.
In FutureFactories, meta-design templates maintain design criteria over potentially infinite
numbers of outcomes. This is achieved using a computational design approach that combines
parametric CAD software with computer scripting. Three dimensional CAD models are defined
by geometry, usually a set of curves, and relationships between them.
Taking a tuboid form similar to the ‘limbs’ that make up the Tuber lamp, the form is defined by a
series of circles. Three operations define each circle, a translation from the origin (in Cartesian
space), a rotation relative to global axes and a scale. Each of these three operations is then subdivided into X, Y and Z components. Each element in the CAD geometry therefore has nine
discrete parameters that can be adjusted, Figure 28.
The overall model is defined by a string of listings, nine for each geometric entity. Such listings
defining 3D form have been compared to the DNA genotype in nature. Richard Dawkins uses
geometric parameters to represent genes in a graphic growth structure illustrating evolution
(Dawkins 1986). The virtual CAD model is the phenotype or observable morphology, the
embodiment of the genotype data. The genotype listing carries the mathematical instructions
required to build the phenotype form. Parameters within the list can be modified and, providing
values remain within certain bounds allowed by the geometry, a new variant of the same design
will result.
Each FutureFactories model is defined by a list of geometric features and associated parameter
values. It is by operating on this genotype list of parameter values, that the morphing of design
67
variants is achieved. The CAD meta-design describes the features: the computer script, via the
genotype parameters, defines a particular configuration of those features.
Figure 28
Phenotype Form and Genotype Data List
4.1
Key Frame Animation
Design variants in the work discussed so far have been achieved by modifying parameters. A
project aim was to avoid a series of staccato jumps, as one random value is replaced by another.
Instead the model should appear to ‘grow’ with one mutation flowing seamlessly into the next. In
the early work this was achieved using CAD package based key-frame animation. In key-frame
animation an entity is created along with a series of developmental stages for that entity between
the start and end states. CAD software then extrapolates between these key stages creating a
series of transitional models between them. This results in an extended sequence of model
states each differing only slightly from its neighbour. Rendered images of these model states can
played sequentially to produce a seamless video animation.
Key-frame animation was employed to create the design variants of the ‘First Collection’
Lampadina Mutanta, Nautilus, Twist, Let’s Twist Again and Tuber exhibited at the end of the
residency period. At that point the generative designs created were limited to test sections like
the tubular volume, in Figure 28. In the more complex design examples produced, the key frame
states were derived manually.
Creating a key-frame animation can be an intuitive process with the development assessed at
regular intervals. The end state is fixed and pre-defined. In extrapolating between the key
68
stages, the software generates a discreet model at every frame. A vast number of models can be
created from even a short clip of animation (30 frames a second is typical). The collection of
Tuber variants was derived from a two minute animation clip offering a potential 3600 discrete
forms, one for every frame of the animation.
The difference between neighboring frames
however is almost imperceptible, a level of subtlety necessary to achieve a smooth animation.
Five distinct Tuber variants were created from the animation, Figure 29. If a greater number had
been produced from the clip, the separation of the frames produced would be reduced and there
would be greater similarity between the variants.
Pre-defined key-frame animation offers potentially large numbers of unique variants for relatively
few key-frame inputs. The scope is nevertheless limited however, and falls short of the project’s
fully automated production aims in which there should be no limit to the potential variants.
Figure 29
Tuber variants
4.2
Procedural Animation: Rules, Ranges and Relationships
Each solution generated is intended to be unique and not repeated in a cycle, however long. The
second stage of software research was to develop procedural animations.
In a procedural
animation, entities are modified by a procedure or algorithm to create successive keyframe
states.
A set of developmental rules and relationships are established along with an initial
condition for the entity. Solutions are then generated automatically. In contrast with the standard
key-frame approach, procedural animation is abstract. ‘Control’ of the development is attempted
via indirect inputs which can be multi-layered and interrelated.
The results, while being
determined by the algorithms (as apposed to truly random), can be unpredictable, with
69
experimentation required to achieve the desired results. Once created, however, a procedural
animation can yield a potentially infinite series of solutions, given an appropriate script.
4.2.1
Procedural Animation Principles as Applied in FutureFactories
The procedural animation work began with setting variables, to cycle independently through
specified ranges. Different variables were set to cycle at different rates. The fact that variables
changed out of phase provided the computer generated random element.
Figure 30
A Solid Formed by Lofted Square
To understand the principles a simple box can be considered, Figure 30: this could be, for
example, the leg of a piece of furniture. A simple solid model is created by ‘lofting’ three square
sections. ‘Lofting’ is the creation of a 3D transitional form between 2D profiles (usually) and is a
common feature in high-end CAD. The form defining 2D profiles are termed control curves. The
size of each square control curve, defined by dimensions D1, D2 and D3 respectively, is allowed
to cycle independently 100% - 30%. Figure 31, shows the effect of this applied to the uppermost
control curve only.
70
Figure 31
The Effect of Scale Variance
Now all the scale of all three control curves is allowed to cycle through the range but at different
rates. The cycling of the variable D3 is set at a given rate. The other two variables, D1 and D2
are now also set to cycle though the same value range but at different rates. The effect of this is
illustrated in Figure 32. In this example D1 cycles at twice the rate of D2 and D3 at five times the
rate of D2.
Figure 32
The Effect of Differing Scale Variance on all Three Lofted Profiles
71
With only three variables that cycle through relatively narrow ranges, the chances of similarity are
high and similar pairs are evident. Another element of variance that could be considered is the
addition of a twist about a vertical axis formed by rotating the horizontal profiles in Z, as illustrated
in Figure 33.
Figure 33
The Effect of Rotating One Lofted Profile
So far the three control curves have been evenly spaced with the mid-profile of the loft
construction located mid-way up the form. This control curve can be allowed to rise or fall, Figure
34.
Figure 34
The Effect of Altering the Height of the Central Profile
72
The rotation about the vertical axis and the asymmetric placement of the mid profile are assigned
ranges and independent rates of change. These transformations are overlaid on the earlier
Figure 32 model, and the resulting forms illustrated in Figure 35. There are now five variables
and, in spite of the similarities in the scale transformations, there is a reasonable difference
between all iterations.
Figure 35
Effects of combined transformations
The Twist candlestick, Figure 36, was developed using these principles. The design has three
leg volumes which meet at the top. Each of the legs has elements of variance similar to those
used in the box example. The three legs morph independently, but with a constraint to ensure
the upper control curves of each leg volume match (Section 6.1.5).
Figure 36
Twist Candlestick
73
4.2.2
The Nature of Morphing: Micro Changes, Macro Changes and Alterations of the
Geometrical Structure
In the process described in 4.2.1 parameters are considered in isolation. In design terms this is
simplistic. When manipulating a model a designer will often choose to apply the same operation
to a group of features, moving an entire leg for example. It may be desirable therefore to group
features in order to apply operations to the set as a whole.
Another manipulation strategy common in digital animation is to spread an operation over a
number of features, blending out its effects progressively. Taking the control vertices (CVs) on a
surface for example, Figure 37: a single central CV could be displaced deforming the surface
locally (left) or the displacement could be spread over neighbouring CVs with diminishing
amplitude (right). An operation can be given a ‘sphere of influence’, such that it affects all entities
within spherical bounds: this has diminishing intensity with increasing distance from its centre.
Figure 37
Sphere of Influence
The table leg in Figure 38a may have a translation applied locally, Figure 38b, creating a kink in
the form. Using a sphere of influence, the effects of the same translation can be spread over the
entire leg producing a flowing bend, Figure 38c.
74
Figure 38
Modes of Mutation
4.2.3
Structural Changes to Geometry
Sometimes the desire for change goes beyond the mere adjustment of existing geometry. Some
modifications, the addition of a whole new leg in Figure 38d for example, requires changes to the
fundamental geometric structure of the CAD model. This type of change proved difficult to
accommodate in the early FutureFactories models, due to their surface based geometry and the
requirement that each iteration produces a potentially viable product. The parametric models
could not support the creation of new geometry or the removal of existing features whilst
maintaining construction history.
In the natural world complex systems have evolved from crude beginnings. The human eye
perhaps evolved from something akin to the light sensitive spots processed by some single celled
animals (Dawkins 1986).
With a pre-defined design template the design remains constrained,
however long the development period. An LED, for example, has fixed physical parameters and
is either there or it is not. It cannot evolve for example from, in a parallel to the natural world, a
slightly glowing protuberance.
A degree of structural evolution is desirable, if not essential. Structural changes to the geometry
can be achieved by breaking the design down into an assembly of separate sub-models. Tuber
consists of four limbs that intersect.
These can be separate CAD models joined in an
amalgamating Boolean operation (Technical Glossary page 267). Each of the four volumes has
its own geometry whose integrity must be maintained; the links however, need not be predefined. The links are created indirectly by physical interference between the limbs and are only
75
created if a particular iteration is manufactured. As the links are not pre-defined the format of the
assembly can change during the evolution as long as all four limbs remain linked. A link can pass
from one limb to another, in the manner of a baton being passed in a relay race, Figure 39. A
direct link between two volumes can be broken as long as they remain indirectly linked and all
four volumes remain joined.
Figure 39
Changes in Model Structure
76
4.3
The Use of Evolutionary and Genetic Algorithms: The Introduction of Selection
To this point it had been considered necessary to define a complete envelope of parameters.
The envelope defined a ‘solution space’ covering every possible mutation of the form. Each
individual parameter required specified ranges that considered both the effects of that parameter
alone and its effects in combination with others. If there are more than a handful of parameters
and their effects interrelate to any significant degree, then the task of specifying such an envelope
becomes extremely long and complex. An aim of FutureFactories is to explore generic systems
of commercial potential. For commercial viability, it should be possible to introduce new designs
with reasonable ease. Ideally one would be able to apply mutation rules to a conventional 3D
model via an intuitive on-screen process: the system was initially seen as a ‘plug-in’ addition to
high-end parametric CAD systems. The complexity of specifying a parameter envelope could be
reduced by severe restriction of the permissible parameter ranges and by isolating their effects
where possible.
However this would lead to uninspiring, predictable changes in the form,
repeated oscillations, for example.
A way of simplifying the generative rules had to be found. Evolutionary design principles offered
a potential solution. Genetic algorithms permit virtual entities to be created, without requiring a
full understanding of the procedures or parameters used to generate them.
Instead of
incorporating expert systems of technical knowledge into the programming, evolutionary design
systems rely on utilitarian assessments of feasibility and functionality (Sims 1999, Funes and
Pollack 1997). The parameter envelope specified in this research is designed to perform two
functions; to ensure that manufacturability be maintained and that the mutated form retains the
‘designer’s intent’. If the degree of success in meeting these requirements can be assessed and
quantified, then selection can potentially guide the design towards acceptable solutions rather
than having to identify every possibility.
4.3.1
A Model for Mutation and Selection
The procedural animations created had been driven by a series of algorithmic steps. Introducing
evolutionary algorithms, each key-frame step becomes a generation with a single parent and an
arbitrary number of randomly mutated offspring. Each of these offspring would have a single
randomly selected parameter modified by a predefined step. The iterations would then be ranked
for ‘fitness’ and the most successful selected as the parent of the next generation. The scoring
for fitness is based on the ‘desirability’ of the last transformation with reference to the designer’s
intent. Before becoming the parent of the next generation the selected iteration is tested against
functional criteria such as fitting the production machine volume. This ensures that after mutation
77
the design is no less manufacturable. If the parent fails this assessment the next best offspring is
selected. Only selected offspring are tested against the failure criteria to reduce computation.
Animation is then employed to provide a flowing transition between one generation key frame
step and the next, as illustrated in Figure 40.
Figure 40
The Mutation, Selection and Animation Process
78
4.3.2
The Introduction of an Evolutionary Pressure: Aesthetic Evolutionary Design
Evolutionary principles were introduced to simplify control; the aim is not the functional
optimisation of the designs through evolutionary computation. The suitability and functionality of
the design are present in the initial seeded product form. These qualities are then maintained by
the selection process, rather than their being improved. It is hard, however, to avoid creating
evolutionary pressures by virtue of the ‘fitness’ scoring. Public reaction to the early stages of the
project also pointed to the inclusion of an evolutionary element. There was a desire to see the
form evolve in a direction. Creating this type of intrigue is obviously important from a marketing
perspective. A level of evolutionary development is seen as a way of stimulating interest and
engagement. One possibility is that designs would be available and evolve for a limited period.
Different periods of the evolution process may achieve different levels of desirability. The value
of an artefact would vary according to its position in the evolution. There may be ‘good’ and ‘bad’
evolutionary periods as there are good and bad vine harvests. The introduction of evolutionary
pressure creates what has been described as Aesthetic Evolutionary Design (Bentley 1999), an
area that borrows from Evolutionary Design Optimisation and Evolutionary Art, Figure 41.
Figure 41
Aspects of Evolutionary Design by Computers, Bentley 1999
79
4.3.3
Ranking
Evolutionary pressure is introduced to evolve increasingly visually interesting designs. The
designer creates both an initial form and the evolutionary pressure that will guide changes in that
form over its evolutionary lifespan. The use of digital manufacturing favours forms of geometric
complexity as these justify the costs of the processes.
It made sense therefore to equate
desirability with geometric complexity.
In order to score geometric complexity, one option considered was to measure surface area
divided by volume.
Dividing by volume prevents simple expansion.
Experiments were carried
out on a simple pendant lamp form based on one of the Tuber volumes. The effect after 200
generations can be seen in Figure 42, the initial form is on the left, the form after 200 generations
is on the right.
Figure 42
Initial State and the Form after 200 Generations
80
4.3.4
Micro-Mutation, Macro-Mutation and Achieving a Balance in Transformation Step
Size
Evolution in the natural world is the result of accumulated small changes. Dawkins points out that
in nature small change is crucial, “Even a small random jump in genetic space is likely to end in
death. But the smaller the jump the less likely death is, and the more likely is it that the jump will
be in improvement…….The chance of improvement resulting from a transformation tends to zero
with increasing step size and to 50% as it decreases” (Dawkins 1986). Similarly, if the parameter
values of a CAD model were re-assigned at random there would be a strong possibility that the
geometry would fail.
The larger differential between old and new values, the greater this
possibility will be. In addition, the more operations that are occurring simultaneously, the lower
the probability that any particular one will be successful as effects overlap. For this reason each
‘child’ generated from the ‘parent’ form has only one parameter adjusted at random +/- one ‘small’
step. The step size is a value arrived at through experimentation and is relative to the operation
and feature in question. The steps must be balanced so that they each achieve a comparable
degree of change to the form.
4.3.5
Assessing Functionality and Manufacturability
Assessing the children for manufacturability is relatively straight forward to envisage, even if
difficult to implement. The validity of the CAD models created can be assessed through the
ability to export a suitable digital file for manufacture. Problems, such as overlapping surfaces,
either prevent successful export, or are flagged up by error messages.
The manufacturing
limitations of the intended digital manufacture process can be imposed, minimum material section
thickness and the machine build envelope for example. Functionality may be harder to evaluate.
Stability may be assessed via the centre of gravity. The space to house internal components can
be examined with an interference check. Practical assessments of this type are used to impose
absolute limits rather than for relative scoring and are a separate check in the generative cycle,
Figure 43.
The aim is not technical refinement. It is not the intention to select the quickest to
manufacture or the most stable; merely to assure that each generation conforms to a minimum
functional standard.
81
Figure 43
Scoring and Functionality Check
4.3.6
Scoring the Aesthetic
Maintaining the designer’s intent requires a more abstract, relativist approach than does
assessing functionality and manufacturability. Selecting designs based on a relative scoring of
specific geometric criteria, allows the designer to express a general image for the design rather
than absolute limits.
The notion of ‘slightly twisted’, for example, might translate to a high
probability of rotation, followed by an exponential decrease in the probability of further rotation
following each rotation selected. As in evolutionary ‘survival of the fittest’, the designs with the
highest ‘fittest’ scoring will prevail. This less rigid form of definition simplifies the set up of the
model and also allows the possibility of new unexpected forms (although the possibility of
dramatic change has to be balanced against maintaining a coherent, identifiable design). The
effects of the rules are ‘softened’ by the use of probability: a high fitness score can be allocated a
higher probability of selection rather than assured success. This again broadens the possibilities,
allowing the occasional success of a less ‘optimal’ and therefore potentially surprising solution.
FutureFactories focuses on a single mutating parent. Evolutionary algorithms are usually much
more sophisticated. They usually feature populations of solutions, and two parents, both of whom
contribute to the offspring’s ‘genetic’ make up. This ‘crossover’ combines characteristics and
provides stability.
82
4.3.7
The Constructive Solid Geometry, CSG ‘Building Block’ Approach
Exhibitions following the initial residency period yielded valuable feedback on the concept.
Would-be consumers, in the form of exhibition attendees, expressed a desire for more dramatic,
fundamental changes than the gently writhing designs presented. At the same time, it became
clear that the number of variables involved and the need to achieve a reasonable balance
between freedom and control made scripting an onerous task; a design investment that would
mitigate against widespread uptake of any system. The adoption of an evolutionary algorithm
approach had helped simplify set-up to a degree, this however works best when limited to gradual
subtle changes. Simplifying the designs and achieving stability in morphing was proving difficult;
it was also at odds with the demand for more visual excitement. A fundamental change of
methodology was needed. The approach to this point had been to take complex CAD models
made up of mathematically defined surfaces and to apply modifying scripts. An alternate, more
script centred, building block methodology was now considered. Rather than the script redefining
geometry, it would more simply re-configure pre-defined geometric building-blocks or primitives
(Technical Glossary).
The building-block approach was explored using Virtools, a software development tool aimed
primarily at video game creation. Virtools was selected for a number of reasons:
Scripts developed using this software can run on a freely available web browser plug-in,

the ‘3D life’ player;

hard scripting;

Designed for web applications, Virtools allows the easy creation of flexible interfaces;

environment;

3DStudiomax;

and

Software is created within Virtools using an intuitive block diagram method rather than
Custom software building blocks can be written if required;
Virtual 3D entities can be rendered with realistic materials comparable to the CAD
3D mesh data can be imported into Virtools from a range of CAD software, including
It has a strong user community with active forums providing problem/solution sharing;
It supports the management of data and attributes, via 2D arrays, with the capacity to
import and export data files.
A simple example of a Virtools code is illustrated in Figure 44. In this script an ‘Iterator’ building
block cycles though each entity in a data set. Each entity in turn is given a translation, rotation
83
and scale operation by respective blocks. These operations loop back to the Iterator. When the
data set cycle is complete, the ‘finish’ output is triggered. The trigger path can be seen in black.
The dotted green lines represent the input of information, in this case the selected entity to
operate upon. This is a simple example and in practice the schematics become significantly
more complex.
The principles, however, remain intuitive with actions tracked through a
schematic path. This intuitive approach, coupled with the software’s suitability for web based 3D
operations, made Virtools an effective tool for the project.
Figure 44
Example of Virtools’ Script
The Grasshopper plug-in for Rhino CAD software, RhinoScript, a more sophisticated Virtual Basic
scripting option for the same package, and Generative Components that works with MicroStation
architectural CAD were considered as alternatives. Being CAD system based, these options all
offered more flexible links with the model geometry. It was freely available web player however
that made Virtools the preferred solution. This would allow ‘customers’ to run scripts on their own
PCs in the manner envisaged for a commercial system without the need for specialist software.
84
4.4.1
DNA: Design Case Study of a Constructive Solid Geometry ‘Building-Block’
Approach
A simple, modular design was considered comprising a network of multi-coloured lenses
arranged around a standard GLS incandescent light bulb, Figure 45.
Figure 45
DNA Luminaire
This design was titled ‘DNA’. In this design a series of linked rims, rather like spectacle frames,
are built one after another around the light bulb starting from the bulb holder. The sequence
begins with a simple open rim which attaches to the bulb holder, via the industry standard
threaded clamp provided for lampshades. There are three different lens sizes: small, medium
and large, 20, 25, and 30mm respectively, and six colour options; amber, blue, green, purple, red
and yellow.
The linked rims are built into a self supporting framework, which is digitally
manufactured in a single piece. The lenses are punched from polypropylene sheet and are
clipped into the rims as a post production process.
Each rim may have three branches, evenly spaced 120 degrees apart, around its circumference.
Links between rims may be straight or twisted with rotations in two axes, Figure 46.
85
Figure 46
DNA Components
In Constructive Solid Geometry, CSG (Technical Glossary), geometric primitives are combined
using the Boolean operations of union, subtraction and intersection, to form a more complex
whole. In the DNA design, rim elements pre-exist as a library of CAD models that are assembled
via Boolean union, Figure 47. Sets of angled links are pre-created and assembled as required
rather than geometry being manipulated to the requisite angles. This ‘building-block’ approach
simplifies the Virtools script, avoiding the manipulation of geometry within the package.
Translation of the virtual representation, through to a manufacturable model, is also simplified.
The design is defined by the configuration of sub-models within it. As the virtual model is made
up of pre existing CAD entities, the ‘production’ CAD model is likewise a configuration of predefined CAD sub-models. Hence computationally ‘light’ visualisation models can be used within
the script, these are then mirrored in a set of technically robust equivalents used for production.
Figure 47
Boolean Union
86
DNA was created to explore a simpler approach to generative design, with as few rules as
possible. Design rules were created to:


Maintain sufficient clearance around the bulb as a thermal constraint;

Restrict the number of lenses in the assembly to define the ‘size’ of the design; and

Restrict the overall form to a practical, easy to build and saleable size;
Prevent physical interference between the lenses.
To resolve the size and clearance issues in the structure as a whole, inner and outer boundary
spheres were created with the design allowed to grow in the intermediary space. The structure
builds in steps with subsequent rims, their positions and orientations selected at random.
Determining the step’s ‘success’ is comparatively simple; if it does not clash with boundary
volumes (inner and outer) or its lens does not clash with any others present in the structure, the
step is allowed. Clashes between the post production fitted lenses are not allowed as this would
prevent assembly. Interference between the surrounding rims is permitted however, as it is
visually interesting and adds strength. A degree of interference between the rims is judged
essential; the minimal nature of the rims means loads must be spread around the structure.
Where rims overlap in the virtual design, they are built structurally united. If DNA were to build
itself as one long coil with no inter-linkage, the resulting structure would potentially not be selfsupporting. In spite of this imperative, no assessment of structural cross-linkage was attempted
in the programming. Structural integrity would be extremely difficult to assess across a randomly
generated network in which intersections have varying degrees of structural merit. In practice
intersections proved plentiful without any scripted provision for them and the issue was set aside.
The structure can in practice be assessed with a quick visual check of the virtual design. The
DNA I Virtools’ script was developed in collaboration with supervisor Ertu Unver. A schematic of
this code can be seen in Figure 48. Subsequent scripts are the work of the researcher alone.
87
Figure 48
DNA Schematic
88
The structure grows until it reached a predetermined number of elements. After this point the
addition continues in the same manner but for every addition an existing rim is removed. The rim
to be removed is selected at random from the set of all redundant rim positions in the framework;
those without child branches, whose removal has no effect on the remainder of the structure.
Once built up to the preset rim density, the design will theoretically continue to modify itself ad
infinitum.
Once the basic build methodology of the concept was established, the design could be refined.
In order to influence the character of the design, restrictions were placed on particular features.
The smallest, 20mm lens would be a dead end from which no further branching would be
possible. The most prevalent lens in the structure would be the medium 25mm lens followed by
the large 30mm with the small lens being the rarest. The probability of selecting each lens size is
as follows:-
20mm lens – 10%
25mm lens – 60%
30mm lens – 30%
This probability is established using Random Switch building-blocks within Virtools, Figure 49
Figure 49
Lens Size Section within the Virtools Script
89
Figure 50
DNA II iterations
90
4.4.2
DNA II
Unrestricted by boundary interference rules, the DNA script would build structure randomly in all
directions out from a point. The ‘natural’ form of the growth in clusters is very different from the
hollow shade form required. Forcing the structure to grow in a spherical skin around the light
source requires considerable computation with many rejected steps for each successful one. As
the structure grows the steps become progressively slower. It became evident that the script
created would lend itself better, both computationally and structurally, to creating a ‘solid’ rather
than a hollow form. In a separate issue, the GLS light bulb on which the design is based is
becoming obsolete for environmental reasons. A voluntary UK retail phase out began in 2008.
To address these issues an evolution version of the design, DNA II, was developed with the GLS
light bulb replaced by six high intensity LEDs, Figure 50. The LEDs are attached to 30mm lenses
within the structure and do not interfere with the branching, Figure 51. Side-emitter LEDs are
used which spread light out into the surrounding structure rather than producing a focused beam.
Figure 51
Side emitter LED fitted to 30mm rim
The rim elements were re-modeled with the links formed in two halves. This allows the X rotation
to be achieved via a rotation operation between the halves rather than by the selection of a premodelled angle. Added to this flexibility, the Y rotation range was extended to cover 5 degree
intervals between 0 and 120 degrees. The ‘Link type’ selection operation of the DNA script,
schematic Figure 47, was therefore replaced by two operations, Figure 52:
1 - Select Y rotation +/- 0-120 degrees (nearest 5 degree interval); and
2 - Select X rotation +/- 0-90 degrees (any whole number).
91
Figure 52
Rim Link Rotations
The structure builds in the similar manner to DNA I but without the inner boundary sphere. The
difference is the six LEDs. The first rim, which is always 30mm, houses an LED.
Each 30mm
lens selected thereafter has a percentage probability, initially 20%, of housing an LED until the six
LEDs of the design have been allocated. The exterior boundary is the maximum permissible
build volume. This is the XY plane area of the machine by a proportion of the chamber height
(determined by permissible cost), Figure 53.
Figure 53
DNA II Build Volume Restriction
92
Iterations #1 and #2 of DNA were built to 100 lenses. The aim would be that after this point
development would continue with subtraction and addition. In practice this addition/subtraction
phase has not yet been implemented. The component makeup of #1 and #2 can be seen in
Figure 54.
Figure 54
Component make up of DNA II iterations
The development of DNA II #1 can be seen in Figures 55 and 56.
93
Figure 55
DNA II #1 ‘Growth’ iterations 1 - 10
94
Figure 56
DNA II #1 ‘Growth’ iterations 20 – 100
95
Figure 57
Holy Ghost
96
4.4.3
Holy Ghost: Design Case Study - Combining the Building Block Approach with
Morphing
DNA is fairly basic in product design terms. The next step was to apply the building block strategy
to a more demanding form closer to the designs of the First Collection, Lampadina Mutanta,
Nautilus, Tuber and Twist.
A chair was selected as the subject as this offered significant
opportunities for a large scale sculptural piece whilst having a simple, well understood function.
Given the cost of Rapid Prototyping (RP) services and the restricted build volume of available
equipment, it was decided to build only the back and arms of the chair, taking the rest from the
iconic Stark/Kartell Louis Ghost Chair as unwitting collaborator.
The ‘Holy Ghost’ generative script was again created in Virtools, only this time the build block
approach is combined with the morphing strategy of earlier works. The form would be assembled
from pre-existing geometric entities but, rather than these pre-defined entities remaining fixed,
once placed they would be manipulated by the script.
In the Holy Ghost design, the back support of the standard Kartell polycarbonate Louis Ghost
chair is cut away and replaced by a laser sintered nylon component. This SLS piece is an
assembly of standard building block elements termed buttons (as introduced in Section 3.5).
Collectively, the faces of these buttons form the back support of the chair. Development takes
place in three phases, Figures 58 to 60.
a) In the first phase, the number of buttons that will form the back support is determined.
This set of units is then placed one at a time on a virtual 3D surface pre-determined by
ergonomics i.e. a back support surface positioned to accommodate an appropriate
anthropometric range.
b) In the second phase, the placed buttons expand in a uniform manner (whilst maintaining
the ergonomic envelope) until they almost touch.
c) In the third and final phase the buttons expand in a non-uniform manner as individual
control vertices (CVs) on the geometry are pulled to close up the gaps in the back form.
97
Figure 58
Virtools Holy Ghost Script: Phase 1, Placing ‘Buttons’
Figure 59
Virtools Holy Ghost Script: Phase 2, Uniform Axial Expansion
Figure 60
Virtools Holy Ghost Script: Phase 3, Irregular Expansion
98
Rather than call up 3D elements from a resource library, the buttons are modelled from a 2D
curve on demand from a custom written building block within the script. This was necessitated by
the requirement to manipulate entities in a controlled fashion. The models are manipulated by
moving mesh vertices.
When geometry created in a separate CAD modelling package is
imported into Virtools, it is difficult to control the numbering of vertices and hence to index them
appropriately. Creating the geometry in the script itself provided a ‘cleaner’, more logical, mesh.
In the expansion phases, the operations take place in small steps across each button in turn, with
an associated collision check.
This step by step methodology increases computation but
prevents the first few buttons dominating as they take up the available space and leave little room
for subsequent expansions. Step by step expansion, compared with a full expansion to the limits,
can be seen in Figures 61 – 63.
Figure 61
Placed buttons
99
Figure 62
Sequential full expansion
Figure 63
Balanced results from step-by-step expansion
100
Expansion is applied to the ‘head’ of the button, while the base (the hole in the centre) remains
unchanged.
The Expansion is ‘blended out’ with a diminishing amplitude applied to vertice
groups down the form, Figure 64.
Figure 64
Blending Out of Expansion Amplitude
The expansion is in two phases. The first is a uniform expansion (x,y,x) in which the forms
remain axi-symmetric. The second phase of expansion is non-uniform, in which the form distorts
to fill the available space, Figure 65.
Figure 65
Uniform Compared with Non-Uniform Expansion
101
To avoid ‘spikes’, a vertex group is not allowed to expand more than two expansion steps beyond
its neighbours, Figure 66.
Figure 66
The prevention of ‘Spikes’
The Holy Ghost Virtools’ script generates only the button forms. In the built product, the buttons
are connected by a matrix of curved links which, built in SLS Polyamide, act as live springs and
allow the whole back to flex like a sprung mattress, Figure 67. The addition of these links is a
manual mapping process. The modelling of the links could have been automated in the software
with further programming investment. A decision had to made concerning the amount of time to
spend on the computer script given that only two were to be produced for the original
commission. As the links had little effect on the individual character of iterations and that the
effectiveness of the script could be judged without their inclusion; these features were omitted
from the programming and modelled manually on the two iterations physically produced. The
perforations around each button are also a manual mapping process though this again could be
automated. Currently, the position and shape of the holes is generated automatically by mapping
a 2D pattern onto the form transposing the 2D X/Y co-ordinates onto the u/v isopharms of the
virtual model. Each hole is then ‘trimmed’ from the forms manually.
102
Figure 67
Holy Ghost Polyamide Link Springs
The mapping approach adopted in Holy Ghost with the computer script working on a simplified
model is one common amongst computational evolutionary systems. The scripts need not in
themselves produce the final phenotype output but merely the genotype code or configuration
which can be transposed, manually or automatically, into the intended outcome. ‘Mapping’
operations are often employed to translate primitive blocks which allow for efficient computation,
into more sophisticated models. A ‘trade off’ found in the fields of artificial life and computer
animation is that of complexity versus control. Sims points out that it is often difficult to build
interesting or realistic virtual entities and still maintain control over them (Simms 1999).
A
balance can be achieved by removing operations from the generative script itself and performing
them only when a ‘finished’ design is required.
The viewer, or in this case consumer, is
presented with a ‘stripped down’ version of the design, whilst some of the detail is lost; the
computational burden of building these features at every iteration is avoided.
example, the animation shows four separate ‘solid’ volumes.
In Tuber, for
The script ensures that each
volume has an intersection with at least one other. There is no attempt in the programming to
unify the four volumes into a single hollow volume and thus avoid intersection surfaces which
cannot be exported additive fabrication. To physically manufacture an iteration, the surfaces of
respective volumes must be intersected, trimmed and filleted in a mapping operation, Figure 68.
With the resulting single volume ‘shelled’ to create a hollow form. A balance must be achieved
between the accuracy of the visualisation and the complexity of computation. An over-simplified
visualisation would be open to misinterpretation. If the mapping operation makes significant
changes to the model, then too much is left to the imagination.
103
Figure 68
‘Mapping’ in the Tuber animation
Including the link elements within the Holy Ghost generative script would required significant
additional development and would have made the computer file larger and slower to run. Its
omission, however, means that a significant manual operation is required after the script has
been run. The most significant factor in balancing level of mapping against scripting investment
is the anticipated volume of sales. Holy Ghost is intended as a gallery edition of ten pieces; there
is little point therefore in huge investments in productivity.
104
4.4.4
SuperKitsch Bangle Case Study
The SuperKitsch bangle takes as its theme, the commonplace, banal, charm bracelet. Charm
bracelet elements are stored as 3D models in a virtual library.
These elements are then
assembled at random to form a rigid loop around the wrist leaving a slight gap for it to be slipped
on, Figure 69. The number of elements that make up the bangle form depends on the particular
elements selected but is in the order of 30 pieces.
The library of models is pre-defined Figure 70. All that is needed to fully define a particular
iteration of the design is a list of elements, their positions in 3D space and their orientation. This
information is automatically stored in an array each time the script is run. When a design is
ordered, therefore, there is no need to up-load the built 3D model back to the server.
The 3D entities stored in the host CAD system and those in the user-end script are not the same.
The user-end models are merely for visualisation and for some relatively crude interference
checks.
These are simple polygon meshes which, although they lack the refinement and
integrity, in RP build terms of their NURBS (Non Uniform Rational Basis Spline) surface model
parents, they have sufficient detail to render convincingly. These meshes are far smaller in file
size than their NURBS counterparts. In essence NURBS surfaces are mathematically defined
and maintain their character at any level of resolution whereas polygon meshes are faceted
approximations.
Figure 69
SuperKitsch Bangle Configuration
105
Figure 70
Superkitsch Library of Charm Elements
106
The SuperKitsch script runs as follows; a charm piece is selected at random from the pre-defined
library of models, it is scaled, orientated and placed on a path curve around the virtual wrist.
Further charm elements are then selected, at random, and swept around the path curve until they
butt up against the existing assembly. The band of elements builds in this way around the wrist
until it forms a complete loop, less a gap for fitment.
The path curve is not static: it translates and scales within certain bounds to create a broad band.
Control is achieved through collision detection, which ensures that there is appropriate contact
between the pieces. There should be enough interference to build a structurally united piece but
visually, the pieces should not overlap unduly. If the requisite level of interference is not achieved
the part is repositioned. An interference check, with start and stop planes, forms the gap in the
loop.
Individual charm elements should not be repeated in the bangle.
Currently there are only
sufficient pieces to complete the loop, with little margin. The library will be added to in the future
providing opportunity to tune the design.
Certain key elements may be ‘seeded’, with an
increased probability of selection. There may be restrictions placed on similar elements, the VW
and Mercedes bonnet badges, for example: these may be prevented from appearing next to each
other or barred from the same bangle iteration. There is potential for users to submit their own
pieces for inclusion. These contributed pieces could be offered to others with an incentive given
back to the creator, in the manner of content sharing websites such as Turbosquid.
A schematic of the SuperKitsch script can be seen in Figure 71 and images of a generated
example in Figures 72-74.
107
Figure 71
SuperKitsch Script Schematic
108
Figure 72
SuperKitsch Iteration
Figure 73
SuperKitsch Iteration
109
Figure 74
SuperKitsch SLS Prototype
110
4.5
Section Summary
In common with conventional industrial design practice there is a need to balance conflicting
demands and criteria within acceptable budgets. Complexity in the design needs to be balanced
against stability and control. A simple form with limited freedom to change over time should be
easy to predict; keeping intricate geometry true to the designer’s intent is a more difficult
proposition. Yet complexity is almost a requirement in order to exploit the free form potential of
additive manufacture and thereby justify the premium costs of the technology.
In addition,
dissemination of the work highlighted a consumer desire for significant step changes over and
above the subtle morphing of Tuber.
The building block approach first adopted with DNA offers the potential of significant step
changes to the geometry with a high degree of stability since its constituent elements are preproven. The aesthetic is limited however, to repeated elements which cannot adapt in form and
perhaps give too much of a random air.
The approach adopted in Holy Ghost of combining the
building block approach with morphing proved successful. The controlled spread of the button
forms provides the look of a ‘designed’ form whilst the building block placement of these elements
provides dramatic change.
Also highlighted in this section is the need to justify computational investment against the manual
preparation of production data. In all commercial practice, design investment, in common with
plant and tooling, has to be balance against sales volumes and amortised over a period of time.
It will sometimes make commercial sense in a scripted design to leave aspects of the computer
modelling out of the programming and to add these manually on each CAD model prepared for
production. This simplification of the script can also increase its speed and stability.
111
112
5.0
Summary of Design Work and Exhibitions 2003-2006
An essential proof of the individualisation concept is that it responds to commercial and cultural
pressures to create desirable artefacts.
Producing random, computer generated shapes,
regardless of their aesthetics, would be straight forward and has been done in computer art since
the early days of the microprocessor. It is product design criteria surrounding the artefacts’
appearance and function that presents the challenge. Assessing functionality is relatively easy in
the laboratory and can usually be measured in an absolute sense.
Subjective aesthetics,
however, can only be judged by public reaction to the actual artefacts created. Aesthetics and
desirability are the driving factors for the work; with the aim that consumer interest will support
relatively expensive production methods. There is no technical benefit to the morphing. It is not
an attempt to breed an optimal solution, via the use of evolutionary and genetic algorithms, and
random variance inevitably brings a level of technical compromise (4.3). The physical production
of artefacts and dissemination of works are central to the thesis in order to justify the concepts’
premises.
It is important to distinguish FutureFactories from digital art and virtual sculpture. There is a huge
practical, developmental gulf between 3D visualisation models that exist only in the virtual world
and technically resolved CAD models that can be exported for manufacture. In the virtual world
simple polygons can be disguised with shaders and textures to resemble all manner of complex
form: flowing hair for example, when in reality, little is defined. A decision was made not to
publish designs unless they had been physically built, at least as a prototype. Functionality
places a huge burden on product designs over artworks and physical factors can only be readily
appreciated in the real world. This section will examine the 3D output of the ‘First Collection’
produced for the residency exhibitions October 2003 – April 2004. The aim in producing this
body of work was to achieve recognisable differences between iterations of a coherent
identifiable design.
These were to be 3D designs that could be modified and controlled by
scripting. The target was to generate unique five iterations of each design; these are identified by
the numbering #1 - #5.
The First Collection focuses on lighting objects, in part because of prior commercial experience in
the field. Allied to this, there are important physical factors that make lighting objects a good
proposition for Direct Digital Manufacture and explain why lighting design has proved popular with
digital practitioners worldwide: Materialise and Freedom of Creation, for example.
113
In its basic form, lighting design can be simply an optical barrier hiding the naked bulb from direct
line-of–sight; such as a lampshade that is placed over the light-source. In a pendant format there
is very little restriction on geometry, almost any shape can be hung. Then there is the material.
RP offers only a limited palette of materials and their finishes are generally inferior to the
consumer product norm. When they are backlit, however, they are arguably more attractive than
moulded plastics and in a light fitting are unlikely to suffer unduly from handling.
5.0.1
Lighting Objects
Lighting (object) design is a field that straddles design disciplines giving a wide ranging of
promotional and marketing opportunities. Lighting designs can be technical and performance
driven or primarily decorative. It is a product genre well catered for in the international design
press being extensively covered in product, furniture, interior design and architecture, events and
publications. It is a market with avant-guard buyers; consumers who value scientific novelty and
who are often prepared to be patient with the shortcomings of an immature technology, a market
in which the designer item, is prized and bought at a premium.
5.0.2
Materials and Processes
There is a hierarchy of RP technologies; from the high-end laser based systems, to budget ‘3D
printing’ systems that employ inkjet printing technology (1.4.2).
Three dimensional printing
machines cost around a tenth that of their laser based technology counterparts, but have
significant performance drawbacks. They are usually aimed at visualisation rather than testing
function. Selective Laser Sintering, SLS, can deliver an acceptable if not decorative surface
finish and a mechanical performance similar, in lamp fittings at least, to that of moulded plastics.
Selective Laser Sintering as a powder based system that requires no temporary support structure
and places little or no restrictions on the designer: this has made SLS the tool of choice for
commercial direct manufacturing.
Understandably, given the difference in capital cost, inkjet based processes tend to be inferior in
both finish and mechanical performance. The resolution is typically cruder, giving a rougher
surface, and the materials weaker, requiring thicker walls than would be expected of a
conventional moulded plastic part.
114
Budgetary constraints restricted early FutureFactories 3D outputs to the economy 3D printing
technologies. This imposed limitations on form, function and surface finish. Designs had to
accommodate heavy wall thicknesses whilst retaining the look of industrial production with
precision details.
In many instances the additive fabrication process was combined with
significant amounts of hand-finishing in order to achieve presentable surfaces. As the scope and
profile of the project developed, industrial support for the work grew and enabled access to some
of the more exotic technologies. The First Collection however, was limited to two 3D printing
processes, 3DSystems Thermojet, which prints in wax and the ZCorp process, which prints in
plaster or starch based powder.
5.0.3
The Driving Computation
As discussed in a scripting context, in Section 4.0, the need to exhibit credible 3D output so early
in the project necessitated a separation and parallel development of scripting and 3D design
development.
Through the residency period various approaches to driving computer based
models via scripting were explored, working initially with relatively basic forms. Alongside the
scripting, 3D designs were created which were based on animations. These designs, driven by
fixed length animation ‘clips’, illustrated the potential of the scripting rather than resulting from
programming running in real-time. The scope of these concepts was limited by the number of
frames in the clip.
They could not offer the potentially infinite stream of unique solutions
demanded by the concept. The intention was for these early pieces to prove the merits of the
concept, garner public interest and encourage debate at a time when the actual scripts created,
while technically promising, were still in their infancy.
As the project progressed beyond the
initial residency period scripted outputs of increasing sophistication began to appear.
115
116
5.1
The ‘First Collection’
A series of local exhibitions was planned from the inception of the design residency.
This
program was scheduled to begin in the autumn of 2004, at the end of the one year residency
period, as the culmination of the project. The FutureFactories ‘First Collection’, comprising of five
designs, was created for these exhibitions. There were three lighting pieces, Lampdina Mutanta,
Nautilus and Tuber and two pieces of tableware, Twist and Let’s Twist Again candlesticks. The
lighting designs were presented from the outset as a series of fully functional pieces.
The
tableware generally remained as non-functional appearance models, built in a Z-Corp
impregnated plaster. One functional example of each design was investment cast and presented
in the later exhibitions.
Each of the five designs in the ‘First Collection’ is detailed below, in terms of:
(i)
The Design;
(ii)
The Morphing Features;
(iii) The Production; and
(iv) Design Issues Highlighted
5.1.1
Lampadina Mutanta
(i) Design
The early months of the residency were spent exploring and communicating the overall concept
without looking toward the production of particular artefacts. When it came to creating design
examples, as demanded for the exhibition program, the task proved far from simple.
The
researcher’s training and experience, along with product design culture itself, all tended towards
the singular outcome and an iterative honing toward the ‘ideal’ solution. A different approach was
needed and, in search of this, the researcher looked to film where content changes over time and
is choreographed. Lampadina Mutanta, Figure 75, or mutant light bulb (the Italian ‘lampadina’
had been used previously by the researcher) was the first design created. It was based on an
imaginary screenplay. The subject of this animation was the iconic GLS Edison Screw light bulb.
The light bulb was pictured, forgotten, in an abandoned warehouse. Over time, like a potato in a
darkened cupboard, the light bulb begins to sprout tentacles, each with its own light source
‘head’. Adopting this fantasy approach, it became possible to visualise the physical manner in
which the design might evolve and mutate, mimicking organic growth of nature. This ‘growth’
would have algorithmic rules to drive it, providing a level of control necessary to appear ‘natural’
and convincing.
117
Figure 75
Lampadina Mutanta
In Lampadina Mutanta the light source is a series of high intensity white Light Emitting Diodes,
LEDs. The LEDs are mounted in the ends of ‘tentacles’ which appear to grow at random from the
bulb form. Lampadina Mutanta is the size and form of a standard GLS light bulb. An irregular
growth of tentacles sprouts from the top and bottom of the bulb, with each tentacle bearing a
white LED. These tentacles are animated, swelling and writhing to mutate the overall form.
Design rules dictate that there are fifteen tentacles, ten facing downwards and five up. This
number and ratio is both an aesthetic decision and a practical one, providing a reasonable light
output balanced between up-light and down-light. There should also be a reasonable spread of
light. If too many LEDs point in the same direction a particular solution is rejected.
In the
examples produced this was a manual visual assessment to determine if light would be seen from
each angle around the hanging bulb. In an automated system virtual cones of light from each
LED would be intersected with a horizontal plane and the area covered assessed against a
minimum value. Overlapping beams would reduce this area of coverage and increase the
possibility of rejection.
The end of each ‘tentacle’ is dimensionally constrained to accept an LED and the direction of the
LED beam is restricted to certain angles from the vertical to avoid glare. The form makes full use
of the flexibility inherent in RP processes. The curling, closely grouped mass of hollow tentacles
is full of re-entrant forms (undercuts) that would be almost impossible to manufacture
conventionally, Figure 76.
118
Figure 76
Lampadina Mutanta cross-section
Three distinct characters of ‘tentacle’ have been created:

‘Drops’ form like stalactites on the lower half of the bulb, tapering as they ‘grow’ downwards,
as if acting under the combined effects of gravity and a surface tension on the bulb, Figure
77;

‘Tentacles’ form from the Drops. These have grown beyond the effects of the surface tension
and have developed sufficient strength to resist gravity to some degree, they have a
tendency to curl and coil, Figure 78; and

‘Risers’ form like stalagmites rising from the upper half of the bulb. As they rise they lean out
from the bulb body and begin to curl down under gravity, Figure 79.
These tentacle types appear in varying proportion and at random positions over the bulb, each
developing according to its type.
119
Figure 77
Lampadina Mutanta ‘Drops’
Figure 78
Lampadina Mutanta ‘Tentacles’
Figure 79
Lampadina Mutanta ‘Risers’
120
(ii) Morphing Features
The morphing features of Lampadina Mutanta can be seen in Figure 80, with the construction of
one tentacle highlighted. The GLS light bulb profile body (green) is fixed while the fifteen tubular
tentacles that emerge from it (pink) are free to mutate. The tentacles are skin or loft surfaces
controlled by circular control curves. Theses circles are manipulated using translate, rotate and
scale operations.
Figure 80
Lampadina Mutanta Morphing Features
(iii) Production
Laser sintering in stainless steel was already commercially available when the project was
conceived and it was the possibility of building in such a durable and, potentially, decorative
material that was one of the drivers for the concept.
The plastics and resins commonly
associated with RP at the time did not appear to offer either sufficient physical strength or surface
finish for use in high-end consumer products.
121
The use of Direct Manufacture in metals early in the project proved impossible: additive
fabrication in metals was expensive, was not widely available and even if it could be afforded, the
surface finishes achieved were crude. These problems were seen as short term issues that did
not diminish the overall viability of the concept. As a short term solution, the decision was made
to produce designs in stainless steel using an indirect form of digital manufacturing whereby
single use wax patterns would be ‘printed’ and artefacts would then be investment cast
conventionally from these masters. The main implication of this was an increase in material
thickness required to ensure an adequate flow of molten metal though the investment casting
mould.
Without this restriction, additive fabrication would only build material as and where
required to achieve the desired physical strength. The intended thickness was 2-3mm depending
on the particular RP process employed. For conventionally casting a component of this size, and
complexity, a minimum thickness 3mm was recommended with 6mm preferred. The design’s
section thicknesses were increased to accommodate the casting process, tapering from the midsection where the sprue would be added, to 3mm at the tentacles, Figure 76. At the tentacle tips,
the design features a thin-walled ‘cup’ in which the LED is bonded. As this wall surrounds the
fixed diameter LED, thickening the section at this point would be difficult. Given that the wall only
extends for approximately 2mm, the decision was made to leave this thickness at the 1mm
planned. The early attempts at casting, iteration #1, highlighted this wall as a problem area and
its thickness was gradually increased from the technically possible to that which can be reliably
cast.
Figure 81, demonstrates problems with thickness at the base of the tentacles and
incomplete forming of the LED mounting cup.
122
Figure 81
Casting Problems at the Tentacle Tips
The need to thicken the outer skin left little room for internal wiring and components.
The
Lampadina Mutanta concept was for a stand alone device that simply screwed into a standard
bulb holder, with an electronic LED driver housed within the bulb body. The substantial material
sections required for the casting process meant that the idea of an internal driver had to be
abandoned. The device would be ‘wired-up’ through the narrow neck of the bulb rather than have
any visible split-lines in the form.
The technology used for printing the investment casting waxes was the 3D Systems’ Thermojet
process. This system requires support structure for any overhanging geometry. The removal of
this material can be difficult, time consuming and marks the surface. Removing support material
from the inside of a closed volume would be virtually impossible. The wax pattern was therefore
built in two halves, split on the centre-line of the bulb, to keep the support material to a minimum
and to ensure accessibility, Figure 82. The intention was to assemble the wax halves and to cast
a single, closed volume. Figure 83, illustrates two early prototypes cast in one piece. The
tendency for the investment material to get caught in the confined spaces between the tentacles
can be noted.
123
Figure 82
Wax halves for casting
Figure 83
The first, single piece, castings, March 2003
In practice casting the artefact in one piece proved difficult. ‘Setting up’ for investment casting is
imprecise: the size and positions of sprues and risers is critical and reliant on the intuitive
experience of foundry technicians. In the early casting attempts the lower tentacles, the mould
extremities, failed to fully form. Rather than waste costly waxes, a decision was made to cast the
halves separately and weld them together afterwards. The chamfered bead around the ‘base’,
124
Figure 82, is to provide material for the welded joint which is subsequently ground off. Casting the
pieces in two halves generally worked well and, in total, five iterations of the design were
produced. A set of four was produced for the local exhibitions and a fifth added for the Milan
International Furniture Fair, in April 2004.
It should be noted from the following production
photographs, Figures 84 to 87, that considerable hand-finishing was required to achieve the
desired, polished surface, particularly bearing in mind the joining weld.
Stainless steel is a
difficult material to finish. Marks left from the support in the close confines of the intertwined
tentacles proved extremely difficult to work.
Figure 84
Thermojet Wax Patterns
Figure 85
Stainless Steel Cast Halves
125
Figure 86
Halves welded and the weld dressed
Figure 87
Lampadina Mutanta #3, Barnsley Design Centre October 2003
Five stainless steel Lampadina Mutanta iterations were produced in stainless steel. The pieces
were fitted with LEDs and were fully functional. For exhibition, they were suspended from matt
black bulb holders on period style, braided twin-core cable. In addition to the stainless steel
pieces, a sample was tried in brass. The brass proved attractive, but the halves had to be joined
by brazing. Whilst the welded joint in the stainless steel pieces had been invisible when polished
out, the brazed brass joint remained visible.
126
An impression of what could be achieved using laser sintering was indicated by a sample built for
the project by The Institute of Technology Tallaght, Dublin, Figure 88. This part was built as a
single component and with the intended fine material thicknesses.
No support structure is
required in the laser sintering of plastics and consequently there was no marking to clean up
post-production.
Figure 88
SLS Lampadina Mutanta
127
Figure 89
Lampadina Mutanta iterations #1 -#5
128
Five Lampadina Mutanta iterations were generated, Figure 89. In common with most of the 3D
output, the set of Lampadina Mutanta iterations were not generated and produced at the same
time. The production was expensive and experimental. It had to be tested and the design
developed to suit. This is perhaps symptomatic of the technology. Freedom from fixed tooling
allows continuous development; by manufacturing in this way, a design can rarely be said to be
definitive and fixed.
The first Lampadina Mutanta iteration produced was very much a test piece. Reflecting upon it,
the design was adjusted technically to improve the production results but also from an aesthetic
standpoint. Design iterations #2 to #4 were produced as a batch, prior to the three-venue local
exhibition. There was then a gap which enabled a further period of reflection and adjustment
before the fifth Lampadina was produced for the first international showing in Milan 2004.
On reflection the tentacles of the first iteration appear as an addition, imposed on the form. In the
second set of designs produced, #2 - #4, there is an increased confidence in the visual language.
The tentacles are still a growth applied to the conventional bulb form, but in these pieces the
tentacles appear to grow from the form rather than being superimposed. The distorting affect of
the tentacle is carried further up into the body of the bulb giving a more natural appearance. This
development came from both improving digital skills and a clearer FutureFactories aesthetic
‘vision’. Two significant factors affected in the integration of the tentacles are their size in relation
to the bulb body and the size of fillet given to their join. Smaller tentacles with little or no fillet give
superimposed bodies with little relation between them, Figure 93. A larger fillet allows a degree
of tangency between the fillet surface and the two bodies allowing a more flowing form. As the
tentacle becomes larger in relation to the bulb there is greater possibility that the curvature of the
tentacle will match that of the bulb surfaces giving rise to a flowing integrated form.
129
Figure 90
Lampadina Mutanta tentacle integration
There was a significant opportunity for the researcher to develop the fifth Lampadina Mutanta
variant based on public feedback and personal reflection from the regional exhibitions. The
length of the bulb body form was increased in #5 stretching the form vertically and reinforcing the
impression of growth and distortion.
The body appears to stretch under the weight of the
developing tentacles. The tentacles themselves flow further still into the body of the bulb. A
significant practical development in iteration #5 was to orientate the upper set of LEDs
downwards, Figure 91. As uplighters these LEDs had been somewhat wasted as they lacked the
intensity to reflect light off the ceiling. Aesthetically however, and in hindsight, this does not
appear to have been a good decision. The increased length of #5 is an undoubted success but
the upper LED should have perhaps remained upward facing. The first four Lampadina Mutanta
finished variants, #1 - #4, can be seen in Figure 92. This is a photograph taken by Arkima, the
exhibition publication producer, in September 2003. The variants are identified on the image.
130
Figure 91
Lamadina Mutanta #5 pictured in Milan 2004
Figure 92
Lampadina Mutanta variants, September 2003
(iv)
Design Issues Highlighted
The physical manufacturing issues and consequent developments have been discussed.
In
terms of its success as a script-driven design, Lampadina Mutanta highlighted important issues.
The three characters of tentacle envisaged for the piece were each intended to develop in their
own characteristic fashion. This meant developing separate scripting approaches for all three. A
better solution would be to have a single computer script generate three different types of output.
In later work a common scripting approach was adopted for all features within a design.
131
As discussed in the comparison of variants, the filleting between tentacle and bulb body is a key
factor determining the character of the design. The aim through the development of the iterations
was a flowing, natural form. From a scripting point of view, the fillet should be a constant radius
allowing an automated generation of the feature:
1. Select surface 1, the bulb body
2. Select surface 2, the tentacle
3. Specify the constant radius to build between them.
The geometric regularity of a constant radius fillet however does not appear ‘natural’ in this
design. An appropriate radius in one area of the join is too large or small in another, Figure 93.
Larger fillets can also be unstable and a radius that is successful in one position may not be in
another depending on the surrounding geometry. A variable radius was built manually in the
examples produced. For a design better suited to scripting, the aesthetic influence of the fillet
should be reduced allowing the automatic generation of a small, stable constant radius fillet
feature.
Figure 93
Constant Fillet Radius Problems
132
5.1.2
Nautilus
(i) Design
Nautilus is a development of the Lampadina Mutanta concept and uses the same tentacle
geometry, Figure 94. In Lampadina Mutanta the form was constrained by the need to preserve
the iconic light bulb form.
As the design progressed from the initial concept, the tentacles
become increasingly integrated into the bulb body from which they protruded distorting the
original volume. This distortion of the original bulb had to be limited in order to preserve the bulb
identity.
Figure 94
Nautilus Cut-Away
A further limitation was the Edison Screw mounting and associated bulbholder. In Nautilus eight
tentacles develop from an abstract volume that may be dominated entirely by the tentacle
‘growth’. A minimal, dual cable suspension system was adopted that imposed little restraint on
the form (provision for two holes at a fixed separation) and allowed the artefact to be visible from
all angles.
These cables would carry power to the lamp, one positive, and one negative.
Lampadina Mutanta and Nautilus both use 5mm, indicator lamp style, LEDs; rather than rely on
these as the primary light source, Nautilus features a low voltage halogen spotlight at its centre
(refer to cut-away, Figure 94). As the LEDs are not the primary light source a decorative blue
coloured LED could be used. The eight tentacles are arranged around the central spotlight
accenting it with decorative light. The effect of this can be seen in Figure 95.
133
Figure 95
Nautilus Lighting Effect
(ii) Morphing Features
The morphing features of Nautilus can be seen in Figure 96; with the construction of one tentacle
highlighted. Eight tentacles connect to a suspended volume. This suspended volume, illustrated
in green, remains fixed. The eight tubular tentacles (pink) are free to mutate. The tentacles are
skin or loft surfaces controlled by circular control curves. These circles are manipulated using
translate, rotate and scale operations.
Figure 96
Nautilus Morphing Features
134
(iii) Production
Nautilus was created after the first Lampadina Mutanta prototype had been produced. It was
designed specifically for investment casting, from digitally manufactured waxes, and was created
with the benefit of previous production experience. Few problems were experienced therefore in
the production of the first piece which was again cast in stainless steel, Figure 97. This first
piece, as with Lampadina Mutanta, was produced by the Leicester based foundry, Lestercast.
Lestercast’s experience was with technical engineering industry components, rather than creative
pieces. To explore a more experimental approach to casting, an artisan foundry was approached
to produce Nautilus in bronze, Figure 98. These were less successful and were replaced with
stainless steel versions in time for the third exhibition at the Media Centre, Huddersfield. A fifth
iteration was added to complete the set for the Milan exhibition in April 2004.
Figure 97
Nautilus #1, September 2003
Figure 98,
Nautilus #4 in bronze
135
Figure 99
Nautilus Iterations
136
Five Nautilus iterations were generated, Figure 99. As with Lampadina Mutanta, there were
developments in the design between the iterations.
There is a marked difference in the
integration of the tentacles into the body between the first two iterations and the ones that came
later.
In the first two iterations and in particular #1, the treatment is similar to Lampadina
Mutanta; there is a body to which tentacles have been applied. They emerge from the form like
the spout of a teapot.
In stages, first with #2, the size and influence of the tentacles was
increased to the point where the separation between tentacles and body is no longer as apparent,
Figure 100. From the design of Lampadina Mutanta, through the iterations of Nautilus, the
tentacle geometry has grown in significance relative to the body it is attached to. In Nautilus’s
iterations #3 - #5, the tentacles almost engulf the body. The form appears as a mass of tentacles
with the connections for the suspension cables appearing as two further limbs. The evolved form
gives the impression of resulting from organic growth and natural forces.
The design
development was fixed for iterations #3 to #5 and these give an indication of how variants can
differ whilst maintaining a coherent design.
Figure 100
Nautilus Tentacle Development
The first iteration featured grooves around two of the tentacles to allow for a locking wire that
would retain the 50mm spotlight. Once a prototype had been built, testing proved the retention
device to be unnecessary. As the groove, with a specific location in 3D space restricted the form
of two out of eight tentacles the feature was abandoned. The grooves remain in #1, but they
have been polished out as far as possible. Figure 101 shows Nautilus exhibited at the Barnsley
Design Centre, October 2003.
137
Figure 101
Nautilus iterations, Barnsley Design Centre, October 2003
(iv)
Design Issues Highlighted
The outer skin of CAD surface models has to be broken down into surface patches similar to the
panels of a tailored garment. A level of surface tangency between the assembled patches is
needed for the overall surface to read as one. This continuity is difficult to maintain in a morphing
model. The most effective solution has proved to be the use of as fewer surfaces as possible. A
natural hanging appearance is achieved in the body of Nautilus by using a single, tubular surface,
running in a loop from one suspension cable to the other, Figure 102.
Figure 102
Nautilus Body Geometry
138
5.1.3
Tuber
(i)
Design
In the design of Lampadina Mutanta and Nautilus, the tentacle geometry to which morphing
would be applied had developed in significance, from mere appendages in the early iterations of
Lampadina, to almost engulfing the host body in Nautilus. A logical progression from this was to
do away with the host body entirely. In Tuber, limbs similar to the tentacle geometry are linked
together to form a pendant structure, Figure 103.
Figure 103
Tuber#5
Two limbs hang from positive and negative power cable respectively, one made higher than the
other to distinguish polarity. At the opposite end of the limb to the cable entry is a 1 watt, high
power LED. Two further limbs intersect with the suspended pair linking them and forming a
united structure. The two linking limbs have an LED at either end making six in total, four facing
down and two up. High power LEDs are typically powered at hundreds of mA (vs. tens of mA for
general purpose LEDs) and can deliver large amounts of light comparable to incadencent bulbs.
They also generate destructive heat which, if not dissipated, will quickly destroy the LED. For this
reason they are mounted on a heat sink. One watt Luxeon LEDs driven at 350mA were selected
for the design. These were mounted on a disc shaped heat sink that would be bonded into the
mouths of the tubular volumes as shown in Figure 104. The links between the limbs are not only
structural, but also provide passageways for wiring. The intersection must be sufficient therefore
to allow the internal voids of each volume to combine.
139
As the artefact is suspended from cables, its centre of gravity is a potential issue and needs to
remain as far as possible mid-way between the two cables for the piece to hang straight. At this
stage of the project the morphing was driven by key frame animation of fixed length. Key frames
were set manually which allowed a visual assessment to ensure that the volume appeared
‘balanced’ at regular points in the animation. This approach which proved adequate for the
numbers produced. In a real-time scripted solution, iterations would be rejected if their centre of
gravity fell outside specific limits. This would be checked automatically.
Figure 104
Tuber HP LED mounting
(ii) Morphing features
The morphing features of Tuber can be seen in Figure 105, in which the construction of one
tubular ‘limb’ volume is highlighted. Each of the four limbs that make up the form is free to morph
and there are no fixed areas. Each of the four tuboid volumes is a ‘skin’ or ‘loft’ surface created
by forming a surface between a series of curves, in this case circles. The form of each volume is
manipulated via the control curves or circles that define its surface. Translation, rotation and scale
operations modify the position, orientation and size of each circle and thereby the cross-section of
the form at those respective points.
Figure 105
Tuber Morphing Features
140
(iii) Production
Given that a significant amount of hand finishing was being used for the indirect digital
manufacture of the cast forms, an option to be considered was using a similar level post-finishing
on the RP piece itself. During design development of the piece, it had been digitally rendered
using a high gloss green shade. In an attempt to replicate this effect, and to obtain a finish close
to that of high gloss moulded plastics, the decision was made to combine budget 3D printing with
an automotive paint finish.
The Tuber pendant luminaires were designed for the Z-Corp 3D printing process (Dean 2004
Appendix 4). In this process layers are built by applying a binder to plaster-based powder. The
resulting porous model is impregnated with either cyanoacrylate (superglue) or wax in a postproduction process: Tuber uses cyanoacrylate for strength. Absolute dimensional accuracy was
not important in this design, only the character of the form. In addition surface finish was not a
significant issue as the surface would require hand finishing in preparation for paint whatever the
process. Removing the loose powder from the somewhat inaccessible internal passageways,
which serve as wire runs, proved a challenge; as did handling the delicate model prior to
impregnation. The first four Tuber iterations produced can be seen in Figure 106.
Figure 106
ZCorp process Tuber iterations
141
Figure 107
Tuber Iterations
142
The five Tuber iterations generated can be seen in Figure 106. A significant development across
the set of five pieces is the scale: the first iteration is significantly smaller than subsequent
variants. Initially the model was orientated upright when considering the build volume. As the
process became familiar, the dimensions were ‘pushed’, orientating the part on a diagonal across
the chamber so that it would just fit in. The dimensions of #5 were increased further still, but this
meant building the part in two halves. The design principle does not change significantly across
the variants. There was an increasing level of confidence in the material. In the later iterations
limbs were allowed to separate further and the tubular form to ‘neck’ more. This went perhaps
too far in #5 which suffered several breakages as a result of the fragile build material.
(iv)
Design Issues Highlighted
Installation issues were identified at the various venues. Hanging the lights required precise
adjustment of the twin cables to ensure that both cables shared the weight. If they were not
balanced correctly the lamp would twist about the one cable carrying the weight. As the lamps
themselves are comparatively light achieving sufficient balance proved difficult; a situation that
worsens with repeated installations as the cable became distorted and less inclined to hang
straight. The current carrying cables were designed to be minimal and un-insulated. In view of the
amount of twisting occurring a length of clear insulation was added to one of the cables to prevent
shorting.
The Z-Corp plaster based composite is a material intended for short term visualisation. This
material has distorted slightly over time. By the second venue in the regional tour the Tubers
were showing signs of cracking. There were subsequently re-painted for the Milan exhibition and
have been re-finished again for every exhibition showing since.
The LEDs fitted to Tuber are bright but very directional.
In the original composite Tubers
effectiveness of the design is diminished by the opacity of the build material. In a reasonably lit
space it proved impossible to ascertain whether LEDs facing away from the viewer were on or off.
This was remedied by the change to translucent Polyamide in Tuber9.
143
Figure 108
Let’s Twist Again iterations
144
5.1.4
‘Let’s Twist Again’ Candle Holder
Chronologically and counter intuitively, Let’s Twist Again came before Twist. Let’s Twist Again
was a name suggested by Curator Paul Atkinson at the time of the First Collection exhibition.
The visuals for the exhibition featured computer generated time lapse imagery of the morphing
and iterations of the tripod form assembled in groups. These images were suggestive of dancing
and the original candlestick project names, Twist and Twist II, did not seem to do the flamboyant
forms justice. The dancing theme suited the elegant tapering legs of the first candlestick design
more than the solid geometric second version and so the original ‘Twist’ was remained ‘Let’s
Twist Again’ and ‘Twist II’ became ‘Twist’.
The renamed Let’s Twist Again iterations are
illustrated in Figure 108 and iteration #1, cast in bronze, is pictured in Figure 109.
Figure 109
Let’s Twist Again #1, cast in bronze
(i) Design
Let’s Twist Again was conceived early in the project and was initially developed in parallel with
Lampadina Mutanta. Difficulties in realising the design, in that the fine tapering legs did not suit
3D printing, resulted in a delay in the production. It was produced after the lighting pieces. The
geometry is simple, with three of the familiar tubular volumes connected to form a tripod structure.
One end of each volume tapers to a foot; at the other there is a mouth to accept a candle. All
three of the legs must have an intersection. This intersection should not be too great. The aim is
for surfaces to just ‘kiss’ rather than run into each other.
145
(ii) Morphing Features
The morphing features of Tuber can be seen in Figure 110; in which the construction of one
tubular ‘leg’ volume is highlighted. Each leg has four control curves to which morphing operations
can be applied:
(a) Scale: The uppermost circle is constrained to accept the candle and to preserve an
acceptable proportion of leg diameter to candle. The bottom circle sets the diameter of
the foot which is a ball radius on the tubular volume; the foot diameter is fixed at 5mm.
The two mid-circles are free to scale.
(b) Translate: The position of all control curves is free as long as the 120 degree
separation between the feet is maintained. Added to this, the upper circles must remain
within the triangular plan footprint that the feet create. These measures are designed to
ensure stability. The height of each leg i.e. the candle mounting may vary, separately,
between a maximum and a minimum value.
(c) Rotation: With the exception of the uppermost circle, which is constrained to the
horizontal plane, all circles are free to rotate.
Figure 110
Let’s Twist Again Morphing Features
146
(iii) Production
As with Lampadina Mutanta and Nautilus, the production material was to be cast metal. The
candle sticks however are two to three times the size of the lighting pieces and it would be costly
to print waxes this size. With budgets already stretched, it was impossible to print waxes for the
piece. The first iteration was prototyped using a desktop CNC machine and chemical wood. The
CAD model was divided into sections which could be produced by this 3-axis milling process and
bonded together. For the exhibition program, two further iterations, #2 and #3 were added.
These were ‘printed’ using the ZCorp powder method. Whilst in Tuber the ZCorp process could
be considered to offer a functional part, the physical demands of the legged structure meant that
in this application, it remained very much an appearance model process. A silicone mould was
taken from the chemical wood model #1. This mould was used to produce a wax for investment
casting allowing #1 to be cast in bronze. It is possible to cast direct from the ZCorp process,
provided the appropriate starch-based rather than plaster-based powder is used: although not all
foundries are prepared to work with this unfamiliar material and a fresh ZCorp model would be
needed for every casting attempt. Creating a mould seemed the better option.
(iv) Design Limitations
Out of the five designs created for the first collection Let’s Twist again proved the least successful
a fact that owes much to the level of finish achieved in the iterations produced. The design is
nevertheless flawed in terms of individualisation and function. The overall form is too governed
by the need to spread the feet and to a lesser extent the candles. The form can be more of less
twisted and the coming together of the volumes can be higher or lower. This however does not
give much scope for significant differences beyond the three iterations produced.
From a
practical point of view the tapering form, while elegant, is volumetrically top heavy which
inevitably presents stability issues. This would not be an issue in direct manufacture the wall
thicknesses can be kept to a minimum; the pieces produced were solid however. In direct
manufacture the artefact would be formed as a hollow volume and the lower reaches of each leg
could be filled to ensure a low centre of gravity and stability. The more twisted the form became,
the worse the resultant stability issues were. As an appearance model in the fragile ZCorp
composite, the problem was exacerbated, as in the more twisted forms the legs would bear sideloading forces. The inability to present effective models led to only three iterations of the design
being produced.
147
148
5.1.5 Twist
(i) Design
As with Let’s Twist Again, Twist is a three legged, tripod candlestick, Figure 111. This design
however, accommodates a single large diameter candle.
The geometry consists of three
independent CAD model volumes which are treated separately. There are no blending surfaces
between the surfaces that would require differing geometry depending on the leg configuration.
This makes the control and adjustment of the model a much simpler proposition than in any of the
previous designs. As the upper portion of each leg is coincident and fixed, the join between the
legs is controlled and the only issue is stability, which is governed by the leg separation. There
need be no post production of the CAD model and the information can be taken straight from the
three animated legs. In practice, the three legs were joined together in a Boolean operation and
the candle hole formed in the united block.
The legs exhibit, once again, the tuboid tentacle
geometry seen in the previous pieces. In this instance, however, the cross-section curves that
define the form, are allowed through the manipulation of CVs, to morph between circles and
squares. Each leg begins and ends with a square cross section, creating a geometric form.
Figure 111
#1Twist candlestick in cast aluminium
(ii) Morphing Features
The morphing features of Twist can be seen in Figure 112, in which the construction of one
tubular ‘leg’ volume is highlighted. Each leg has four control curves, the top two of which are
fixed, anchoring the leg at a common start point at the base of the candle. The three control
curves below this section are free to morph, provided that the foot of each leg remains horizontal
and maintains a minimum separation from the other two feet. In addition to the scale, rotation
and translation operations; the degree to which the cross section is square is also adjusted,
Figure 113.
149
Figure 112
Twist Morphing Features
Figure 113
Twist Cross-Section Morphing, Square to Circular
(iii) Production
As with Let’s Twist Again, the first model, #1, was made from chemical wood and subsequent
iterations #2 - #5 manufactured as appearance models using the ZCorp process. A silicone
mould was made from #1 to enable waxes to be produced for investment casting. Using these
waxes, #1 models were cast in bronze and aluminium. The bronze was not of very high quality
and contained too many imperfections to allow a highly polished finish. This piece was available
for all of the exhibitions. The aluminium piece was much more successful and polished well, in
spite of some degree of porosity. Unfortunately it was not cast in an aluminium grade that could
150
be anodised. This piece was produced in time for the Milan exhibition, April 2004. Figure 114,
shows the Twist iterations as displayed in Milan
Figure 114
Twist Iterations, Photographed in Milan, April 2004
The five Twist iterations generated are shown in Figure 115. Unlike previous designs there was
comparatively little adjustment to the design between the iterations.
The first iteration is
somewhat conservative, reflecting the need to fabricate the model; the later iterations are more
flamboyant by comparison. Once the animation had been created, frames were selected from it
for building. Due to the cost implications of build failures, it was the more conservative forms that
were initially selected. Confidence grew with each build, allowing the building of more marginal
geometry. Iterations #1 and #2 are visibly more robust than the later pieces.
Attempting to maintain stability, via the feet positions alone, is somewhat ineffective as it takes no
account of the mass above and the physical balance of the form.
Assessing the balance
computationally, however, would be complex. A visual assessment would be possible in volume
production. Small corrections would be relatively simple to apply and aesthetic judgement would
not be required.
151
Figure 115
Twist Iterations
152
(iv)
Design Factors Highlighted
Unlike Let’s Twist Again, the geometry of Twist offers the potential for substantial feet. While
stability remains an issue it does not threaten the viability of the design.
The scope for change and difference which was somewhat limited in Let’s Twist Again is
dramatically increased by the changes in cross section from round to geometric. This combines
soft natural forms with hard edged ‘manufactured’ surfaces which are still curvaceous and
flowing. This generates an interesting aesthetic that is not seen elsewhere in the work.
153
154
5.2
The Local Exhibitions
Three exhibitions venues in the Yorkshire region, The Barnsley Design Centre, Dean Clough,
Halifax and The Media Centre, Huddersfield were booked to disseminate the residency work
locally.
The five designs of the first collection, Lampadina Mutanta, Nautilus, Tuber, Twist and Let’s Twist
Again, were displayed at each venue; alongside video projections of the virtual designs in
metamorphosis. The exhibition space for all three venues was approximately 30 square metres.
An exhibition system was designed that featured a series of long plinths, one for each design
type. The plinths were colour coded with a high quality automotive paint finish. A 37 page A6
colour booklet was produced to accompany the exhibition and the plinth colours were matched
with the digital render backgrounds from this publication. The design iterations were presented
sequentially, on or above these plinths.
Coming from the Design Industry, it was perhaps easy to underestimate the depth to which the
technical issues surrounding the work required communication. The majority of visitors had not
come across Rapid Prototyping. Before the individualisation concept could be explained, the
visitor had to first understand what additive digital manufacture was and that it was available.
Secondly, they had to understand that these prototyping technologies could be turned to volume
manufacturing, which they were not at the time; that they would become cheaper and
commonplace.
Only when the technology was understood, could consideration be made of
creative uses for it. A series of graphic panels were designed to communicate the project ideas,
in conjunction with the exhibition booklet.
The Tuber and Twist animations were projected on large screens and ran in loops. On separate
PCs there were animations which demonstrated the ordering process. These animations were
driven by scripts written in the Delphi Programming Language (Technical Glossary). The software
enabled users to stop, select, replay, and store selected iterations; but rather than placing an
order, selected design iterations would be printed as an image on a sheet of paper. As designs
were selected the software captured design data and created a database. This database enabled
a search for any correlations between chosen iterations. It was envisaged that in the future, this
facility could provide data on the preferences of different genders and age groups or other criteria
(Unver et al 2004, Appendix 5).
155
5.2.1
Barnsley Design Centre 27/10/03 – 21/11/03
The Barnsley Design Centre was a surprisingly large, glass fronted, venue, in the market town of
Barnsley. In a modernist concrete run of shop fronts, it provided an open, uncluttered dedicated
exhibition space with full time supervision. Sadly the Centre no longer exists.
The FutureFactories exhibition shared the space taking the central floor area of the venue, while
separate exhibitions were arranged around the walls, Figure 116. The work was first partnered
by a photographic exhibition consisting of wall mounted flat work and then by a collection of the
2002 Peugeot Design Award winners. The Design centre provided a large open space and it was
the 3D work of the Peugeot winners that made the best accompaniment; filling the vast floor
space effectively with complimentary pieces of contemporary design.
The 2D printing of iterations proved disappointing. Only with invigilators present were significant
numbers encouraged to participate. This was only possible on the opening day. There was
interest in the projections from the majority of visitors, but the idea of selecting and printing a
particular form attracted few. In hindsight, this is probably due to the fact that no purchase was
being made, there was interest in where the animation was going but which frame got printed
seemed to be of little consequence.
Figure 116
Exhibition Layout, Barnsley Design Centre, 2003
156
5.2.2
Dean Clough, Halifax 01/12/03 – 16/01/04
Dean Clough is a large, ex-carpet mill, arts venue, offering some beautiful contemporary
exhibition spaces. The comparatively modest size of the FutureFactories exhibition, however,
coupled with a lack of supervision and consequent need for security, led to the provision of a
board-room adjacent to an administrative office for the show, Figure 117. Unlike the previous
venue, which was a dedicated exhibition space with a suspended ceiling littered with abandoned
cup hooks, what was permissible in terms of fitments was limited. The arrangement of the plinths
was governed by a suspended lighting track which had to be employed as the only ceiling
mounting point. The exhibition at Dean Clough lacked the professional appearance of the other
two venues. Generally, perhaps due to a time slot straddling the seasonal holiday, it did not
seem to attract comparable attention.
Figure 117
The Dean Clough Board Room, December 2003
5.2.3
The Media Centre, Huddersfield 23/01/04 – 13/02/04
The Media Centre, Huddersfield, offers business spaces to the creative industries, holds regular
creative events and has a thriving café. Audio visual works rather than artefacts are the usual
focus of the Centre. The digital start-up business located there, and the café scene, provides a
ready audience for the digitally aware. The exhibition at the Media Centre, Huddersfield, proved
by far the most successful of the three local exhibitions. Three months further on from the
opening in Barnsley, there had been time and opportunity to refine the exhibits and to replace
some of them, with improved pieces, the bronze Nautili, for example.
157
The Media Centre actively promoted the event by including it in their printed publication and
attracting local press. The press included a magazine article in the Leeds’ Review entitled, ‘The
light at the end of the tunnel is a mutant potato.’
The smaller size of the venue assisted with the exhibits comfortably filling the space, Figure 118.
Added to this was the advantage that the space had been professionally blacked-out for a
previous exhibition. The exhibition space had been partitioned off from the ‘shop front’ windowed
area with stud walling, the walls painted black and all extraneous light sources curtained off.
There was a separate projection booth and quality projection screen. Of the three venues the
audience at the Media Centre seemed the most receptive to the concept. This was unexpected
since out of all the venues considered, this was the one least familiar with 3D work. Immersed in
digital music and graphics culture, however, the audience were possibly more used to the
consideration of digital futures and alive to the creative possibilities of digital manufacturing.
Photographs from the opening night private view can be seen in Figures 119 – 121.
Figure 118
The Media Centre exhibition set-up, Huddersfield, Spring 2004
158
Figure 119
The Media Centre, Huddersfield on the opening night (i)
Figure 120
The Media Centre, Huddersfield on the opening night (ii)
Figure 121
The Media Centre, Huddersfield on the opening night (iii)
159
160
5.3
Designersblock, Milan, 14/04/04 – 19/04/04
Designersblock, which was launched in 1998, curates and produces international design shows.
Typically, these are satellite events to the major corporate design fairs and are held outside of the
fairgrounds in ‘transitional’ architectural spaces. The Milan Furniture Fair is one of the biggest
events in the international design calendar. It has a long standing tradition of satellite events, or
‘Fouri Salone’, championed over the years by the design magazine, Interni, who publish an
annual guide to these events.
FutureFactories was invited to take part in the 2004 Designersblock Exhibition, Milan, to be held
to coincide with the Milan Furniture Fair 13th-19th April 2004. The exhibition was held at Studio
Zeta, a centrally located retail fashion space that is let out for the period of the design fair, Figure
122. The two storey building is spacious and open with a gated court yard. The 2004 Milan show
was a significant one for Designersblock. Alongside their regular selected show they put on a
well funded, touring ‘Scottish Show’. This made for a very full exhibition and helped attract
crowds, in spite of a number of competing festival events in the city.
Figure 122
Studio Zeta, Milan 2004
161
Figure 123
Designersblock, Milan 2004, Ground Floor Plan
The Scottish Show Occupied the First Floor
162
In the Designersblock exhibition the work was seen for the first time alongside other experimental
design projects.
In all there were 35 international exhibitors and a huge diversity of practice.
The exhibition floor plan can be seen in Figure 123.
The Milan FutureFactories show was the same installation that had been used for the three local
exhibitions. The entire exhibition, including the unwieldy 2.4m plinths, was shipped to Italy by
truck. The exhibition space surveyed in a pre-exhibition visit in March 2004 was airy, but with
high ceilings and a limited number of power points, Figures 124 and 125.
Figure 124
Studio Zeta on the March survey visit, 2004
Figure 125
Studio Zeta on the March survey visit, 2004
163
Figure 126
Studio Zeta installation, Milan, April 2004
164
Unlike the previous regional exhibitions, which were held in dedicated exhibition spaces with
assistance from the venue, Studio Zeta provided the space only. Previous venues had each
provided a projection screen for example, whereas a screen had to be constructed for the Studio
Zeta installation. The use of a back projection screen in Barnsley and the separate projection
booth in Huddersfield had proved professional and effective; keeping the exhibition space free of
projection equipment. Studio Zeta were persuaded to allow the use of a closed adjoining office as
a ‘projection booth’ allowing animations to be back projected through a communicating window
onto a large format fabricated screen. This screen, effectively filling the back wall, formed a backdrop to the exhibition showing the Tuber and Twist animations alongside each other, Figure 126.
The 3.5m metre ceiling proved a challenge for installation and required a tower system to be
shipped along with the work.
The exhibits presented were principally those shown at the final local exhibition at the Media
Centre, Huddersfield. The intervening time however allowed a substantial rework of the pieces
with repainting, rewiring and polishing. The visitor numbers were significantly greater than
anything seen at the local venues and, unlike the previous events, the audience were, by virtue of
the time and place, predominantly interested in design and design technology. There was a
steady flow of visitors throughout the first four days of the show, with only the final Monday
slightly quiet. The preview night was filled to capacity for the duration, Figure 127.
Figure 127
Designersblock, Milan 2004, Preview Night
Reaction to the work was strong with the work featuring on several websites including Core77.
The images published by Core77 can be seen in Figures 128 to 130.
165
Figure 128
FutureFactories at Designersblock, Milan 2004 - Photograph Core77 (i)
Figure 129
FutureFactories at Designersblock, Milan 2004 - Photograph Core77 (ii)
Figure 130
FutureFactories at Designersblock, Milan 2004 - Photograph Core77 (iii)
166
5.4
Evolution of the First Collection
By Spring 2004, the project was benefiting from a growing media profile. There had been the
distribution of the exhibition publication, conference papers, internet press reports on the Milan
exhibition and editorials in Icon and NewDesign Magazine, Figure 131. At the same time, the rise
of Materialise-mgx, as a manufacturer, was stimulating great interest in all aspects of digital
manufacturing. As a result of growing awareness of the project, contacts were made with a
number of industrial bodies including EOS, Germany, a manufacturer of laser sintering
equipment.
Figure 131
Newdesign Issue Nineteen Front Cover, 2004
EOS offered to produce a piece using Laser Sintering to demonstrate what the technology could
offer. Rather than reproducing one of the existing designs, there was an opportunity to produce
something more complex that could not be realistically achieved with the 3D printing methods
employed to that point. Tuber was by far the most successful design at the time, and the decision
was made to develop a more intricate SLS version of the form, Tuber9, Figure 132.
167
Figure 132
Tuber9
168
5.4.1
Tuber9
(i) Design
The original Tuber hangs from two separate power cables, one positive and one negative. The
cable lengths are critical in order that the piece hangs correctly (with the design Z axis vertical).
The two cables must share the weight of the relatively light weight luminaire and this makes
installation difficult. In a larger piece, the balance issue would become more acute and the
design moved therefore to 3 suspension cables, one positive, one negative and one for support
only. The cable lengths still need to be adjusted but with more leeway as there is less of a
tendency for the piece to twist.
Moving to SLS manufacture the wall thickness could be reduced, this would give more room for
wiring in the internal voids allowing the use of more LEDs. There would be three tubular volumes
one from each of the cables, each with an LED at the bottom. Added to these would be similar
linking volumes with an LED at either end, Figure 133.
Figure 133
Tuber9 configuration
The possible configurations for Tuber9 were as follows:-
3 suspended members + 1 linking member = 4 downlight LEDs + 1 uplight LEDs
3 suspended members + 2 linking member = 5 downlight LEDs + 2 uplight LEDs
3 suspended members + 3 linking member = 6 downlight LEDs + 3 uplight LEDs
169
3 suspended members + 4 linking member = 7 downlight LEDs + 4 uplight LEDs
A configuration of three linking members, giving nine LEDs, was selected. Compact electronic
constant current drivers were available for up to nine LEDs and this seemed to offer an
appropriate light output for a stand-alone luminaire (the original Tuber were designed for use in
groups).
The majority of issues to this point had been in the digital manufacturing processes themselves.
Moving from limitations of 3D printing to more complex geometries highlighted assembly issues.
Consumer products usually consist of a number of parts, some manufactured specifically for the
product and some ‘standard’ stock components. The product requires assembly either as an
automated process or by hand. The need to pass wires from one Tuber volume to the next was
highlighted in the original design; this issue became more acute in Tuber9. It is possible to check
that the volumes intersect on the virtual model. The degree to which volumes connect can be
assessed by checking if surfaces offset from the originals would still intersect.
These
assessments can be automated. It would be extremely difficult however to access the degree of
difficulty in threading a wire through the assembly. It is not easy to do this ‘manually’ looking
around the virtual design on screen. Blind corners and forking paths coupled with a textured
surface that ‘grips’ the minimal wire can make the task more difficult than it would appear on the
computer. An additional problem is the removal of loose powder, that although un-fused, can
remain in place were the geometry tends to retain it; this material can block the passageways.
The strength and integrity of the build brought design benefits beyond complexity. The material
around the LED could be made fine enough to become translucent.
This allows the whole
structure to glow rather than merely emit beams of light. The stair step striations of the process,
at this fine resolution of the SLS system, become an attractive decorative feature back-lit, Figure
134.
170
Figure 134
SLS Translucency in Tuber9
The strength and elasticity of the polyamide allowed spring-clip features to be designed into the
form. Rather than bond the LEDs into place, spring-clip mountings were incorporated into the
structure, Figure 135. This solution was more minimal than the flange required for bonding,
allowing greater translucency.
Figure 135
Tuber9 LED Clip Fittings
171
(ii) Morphing Features
In terms of the morphing, Tuber9 has the same geometry as the original, Figure 136; only this
time there are six tubular volumes rather than four. Each of the six limbs is free to morph and
there are no fixed areas of the form. The limbs are skin or loft surfaces controlled by circular
control curves. These circles are manipulated using translate, rotate and scale operations.
Figure 136
Tuber9 Morphing Features
(iii) Production
Significantly Tuber9, Figure 137, required no production work beyond the automatic SLS build
and electrical assembly. There was no post-finishing or the significantly skilled operations to
perform (although wiring issues have already been documented). This could therefore have been
an industrial production process independent of the designer’s input.
Figure 137
Tuber9 in Laser Sintered Nylon
172
5.4.2
Lightbikes
The Lightbikes were created for the London Design Festival in 2005. The idea was to create an
independent satellite event outside the main exhibition venues.
A set of three bicycle trailers
were produced; each trailer carried a series of laser sintered lighting designs, Figure 138. The
geometry of these lamps was taken from Tuber. The morphing features are illustrated in Figure
139 and the five iterations generated in Figure 140. The Lightbikes were presented as part of the
London Design Festival 2005 and in various Italian cities through 2006 in collaboration with Italian
cultural group Esterni, Figures 141- 143.
Figure 138
Lightbikes Bicycle Concept
Figure 139
Lightbikes Tuber, Morphing Features
173
Figure 140
Lightbikes Tuber Iterations
174
Figure 141
Lightbikes, Milan 2006
Figure 142
Lightbikes, Rome (i) 2006
Figure 143
Lightbikes, Rome (ii) 2006
175
5.5
Section Summary and Audience Feedback
By April 2004 and the Milan exhibition, FutureFactories’ output had built into a sizable exhibition
with some 23 artefacts. There were five design families and a range of materials and finishes.
The designs presented were functional, highly finished and worthy of exhibition in their own right.
Before the regional exhibition program began there had been little external feedback to the work.
There had been several presentations of the project to Huddersfield staff and undergraduates,
but these were audiences familiar with the concept and they were not seeing finished artefacts.
The exhibitions gave the opportunities to gauge reactions of those new to the concept. The Milan
exhibition was particularly useful in this respect as it had a high footfall and the researcher was
present throughout the event. The researcher was also exhibiting alongside fellow practitioners
and through the event itself, the preparation for it and the break down afterward, there were many
valuable opportunities to discuss the work on show.
The main comment arising was a demand for more dramatic change in the forms.
The
researcher had felt that the morphing was dramatic, given the geometric constraint of the CAD
models. The audience did not have these preconceptions and wanted to see new elements
evolve, features to be subsumed or to wither away. It needs to be remembered that the audience
could ignore physical practicalities and that unlike a manufacturer they have little concern with
building a brand or identity. Nevertheless this pressure for increased drama had to be recognised
and future research adapted accordingly.
Out of the five designs presented, Tuber drew by far the most attention. Feedback from visitors
indicated that they felt this was both the most dramatic design and the most effective consumer
product delivering a powerful lighting effect. Tuber also benefited from the looped animation
presented alongside it with Twist the only other animation shown. The tableware drew the least
attention, these however were the least well finished of the pieces with some of the iterations
presented in the raw Z-Corp finish. In addition to this all of the installations were dimly lit to suit
the lighting exhibits, in spite of spot lighting this did not suit the remainder of the work.
Lampadina Mutanta was the second most popular piece but the public perception of it was more
of an art installation than a collection of functional consumer products.
176
6.0
Commercial Retail Products
By late 2003, Materialise, Belgium, had established themselves as pioneers of RM (Section
1.6.2). The researcher met Materialise.mgx Art Director, Naomi Kaemfer, in May 2004, and
began discussing designs concepts.
While the concept of individualisation did not fit the
company’s marketing plans the aesthetics of FutureFactories were of interest. Two approaches
were considered, adapting an existing FutureFactories design and that of creating a new design
specifically for Materialise.
6.1
RGB.mgx
The Tuber family of lights had received a good public reaction at the Milan Fair in 2004 and
Materialise were keen to see a less complex, more commercial adaptation of this design. The
high intensity white LEDs used in Tuber were also available in red, green and blue for colour
change applications. Using these LEDs with a colour change electronic driver, which varies the
power delivered to three coloured channels individually, a spectrum of colours can be created.
Incorporating colour change in a pendant Tuber would have been difficult, as separate circuits are
required for each of the three colour channels. Tuber pendant lamps are wired in a ‘daisy chain’
with wires passed through often tortuous internal passageways; a single circuit is fed through the
two live suspension cables. Moving to three colour channels would mean three times the wiring
and would be impractical. A simpler wall lamp adaptation was proposed, with short, accessible
wiring runs feeding into a spacious central void. This design would have three LED ‘heads’ as
opposed to the original Tuber’s 6. Each of three ‘heads’ would have a different colour LED, red
green and blue respectively. The LEDs would shine back onto the supporting wall creating a wall
wash of the mixed colour, Figure 144.
Figure 144
RGB detail showing the colour mix
177
Figure 145
RGB iterations
178
Materialise wanted to serially produce designs, rather than pursue any form of individualisation.
However, the researcher was keen to promote the project ideas, in addition to aesthetics, and
created the design with the same morphing strategies seen in the earlier work.
(i) What Morphs?
RGB combines the ideas of Tuber and Twist.
There are three of the familiar Tuber tuboid
volumes, each with a common fixed base. The geometry of RGB can be seen in Figure 146, in
which the construction of one tubular ‘leg’ volume is highlighted. The position and angle of each
head is fixed, relative to the base. The 4 circular control curves between the head and base
sections are free to morph.
Figure 146
RGB morphing features
(ii) Production
A set of three RGB variants were commercially produced, Figure 145. RGB was initially designed
to be installed as a set of at least four, powered by a remote electronic driver that can be
programmed to give a set of preset colours or a sequence of colour change. Materialise later
introduced a stand alone option with three identical fixed colour LEDs, Figure 147.
179
Figure 147
Stand Alone Fixed Colour RGBs
An experimental element introduced on RGB was a delicate ‘eyelash’ feature on the LED
apertures, Figure 148. Fine tendrils would extend from the edge of the aperture into the LED
beam catching intense light to form a decorative feature. The eyelashes blocked the insertion of
the LED to a degree but were flexible enough to be pushed aside. The main design issue at the
time seemed to be balancing a delicate appearance with reliable production. This was achieved
with a diameter of only 0.8mm.
Figure 148
RGB ‘Eyelash’ Feature
180
In practice the issues proved to be packaging and handling. The feature is effectively a series of
fine hooks in a textured material that grips fabric. It was easy to accidentally catch the eyelash on
clothing and break it off. In spite of production success and dramatic aesthetics, Figure 149, the
feature was abandoned for fear of excessive number of returns. This experience highlighted the
fact that there are production limitations beyond the performance of the production systems
themselves.
Figure 149
RGB ‘Eyelash’ Feature Lighting Effect
RGB was selected for the International Conference and Exhibition ‘Colour Days’ Warsaw, Poland
16th October – 16th November 2007. Curated by Anna Siedlecka and Radek Achramowicz, the
exhibition explored the innovative use of colour in architecture and design. A set of six RGBs
were exhibited running a sequenced colour change, Figure 150.
Figure 150
Colour Days Installation, Warsaw 2007
181
Figure 151
Creepers Iterations
182
6.2
Creepers.mgx
RGB was an immediate response to the opportunity of working with Materialise and built on
existing work. Alongside this development, a second proposal was considered; this would be a
new piece designed specifically to exploit the capabilities of RM in SLS nylon. The design would
be a lighting piece designed for LEDs: it would not be the high power LEDs used in the Tuber
designs, but low power 5mm LEDs commonly used as equipment indicator lamps. There is a
‘trade-off’ in these LEDs between light output and beam angle, the brighter LEDs have a
comparatively narrow beam. For an ambient light application, it was essential that the beam
illuminated a reflector or diffuser to make the narrow beam visible from a variety of viewing
angles.
The aim was to create a new form of lighting; a modular concept that uses decorative light to
divide interior space. A Creepers installation would be made up of stems that clip between
vertical low voltage suspension cables: ideally, floor to ceiling, so that there is almost no
supporting structure. These stems would be small enough to be digitally manufactured
economically whilst cumulatively creating a light much larger than anything previously seen in
RM. Clusters of reflectors would appear to float, ‘cloud like’, supported only by the minimal
cables. Stems would be attached, in a seemingly random disorder, clustered in groups, and
reaching out in ‘spore-like’ trails. The first Creepers concept is shown in Figure 152.
Figure 152
Early Creepers Concept
183
Figure 153
Creepers Installation
The suspension wires are spaced 265mm apart and are alternately connected to the positive and
negative of a 12v dc supply.
Some tensioning provision for the suspension wires is
recommended to prevent sagging in the cable. The stems are fitted to span between positive and
negative support cables, Figure 153. The 5mm LEDs typically give a voltage drop of
approximately 3 - 4 volts, allowing three LEDs per stem, with a 12v supply.
(i) What Morphs?
As with RGB, Materialise were looking for a serially produce design rather than to pursue
individualisation. Nevertheless, the design is based on a morphing strategy. The beam of light
from each LED is caught on ‘leaf’ or ‘petal-like’ reflector/diffusers. The arrangement of these
petals gives the impression of random chaos. They are, however, grouped and orientated, 9 or
10 to an LED, to catch the entire cone of light from the LED. The size and position of each
reflector can change. The aim is to fill as much of the light beam as possible, with as little overlap
as possible, as detailed in the pink area, in Figure 154.
184
Figure 154
Creepers’ reflector group
The reflectors are positioned in three dimensional space. As well as a plan view, positioning the
distance of the reflectors from the LED can also change within pre-set boundaries, Figure 155.
The reflectors are wafer thin and therefore semi-translucent. They act both as reflectors and
diffusers. The SLS nylon from which they are made is white: the colour comes from the LED
lighting and Creepers is pure white when switched off. Red LEDs were selected for the design as
they provided the greatest light output.
Figure 155
3D Envelope of a Reflector Group
(ii) Production
Creepers exploits the flexibility of digital design and RM with built-in functionality. The stems
themselves act as flexible conduits into which the pre-wired LED string is sprung, via a slot
running the length of the stem. Spring clips at either end of the stem attach it to the suspension
wires, while two further integral springs push home conductor pins through the support cable
insulation, Figure 156.
185
Figure 156
Integral Spring Clips
In service, users felt uncomfortable with the degree of bending required to fit the conductor and
the spring was later replaced by a separate clip component. In the revised system, the stem
would be clipped into place via the integral spring clips at either end, Figure 157. When the
Creepers composition has been arranged as required, the separate conductor clips, which hang
loose at either ends of the LED wiring string, are clipped into place making the electrical
connection, Figure 158. These clips carry positive and negative markings to ensure correct
connection.
Figure 157
Revised Clip Detail
186
Figure 158
New Conductor Button
Details of a Creepers’ installation can be seen in Figures 159 and 160.
Figure 159
Creepers Installation Detail (i)
187
Figure 160
Creepers’ installation detail (ii)
Creepers was launched at the Milan Furniture Fair in 2005, Figure 161. In 2007, Creepers was
short-listed for the SWELL - Future Friendly Design Awards, Vancouver, Canada, and featured in
the accompanying exhibition.
In 2008, the design was included in a curated exhibition and
publication, “all light-all right”, at the Hangaram Design Museum, Seoul, Korea.
Figure 161
Creepers’ Installation, Milan 2005
188
6.2.1
Tangle
Creepers is an architectural product. The power floor-to-ceiling cables are site-specific, requiring
skilled specification and installation.
A simpler, complimentary design to Creepers was
considered that would be stand-alone and require no installation. The result was ‘Tangle’, a
simple lamp shade, of a similar format to the early Materialise designs (Section 1.6.2).
Materialise, however, felt that Tangle might compromise the commercial positioning of Creepers
and the design was shelved. The Tangle prototype was shown alongside Creepers at Funky, an
exhibition at the Norsu Gallery, Helsinki, Finland, Figure 162.
Figure 162
Tangle Photographed at ‘Funky’, Norsu Gallery, Finland, 2007
Funky, curated by Stephanie Seege, featured the work of six practitioners, whose works were
considered to be, “Finding a balance between the strangely shaped borderline of design and
craft” (Norsu Gallery 2007).
Lin Cheung, (ENG) jewellery
Hans Sandgren Jakobsen, (DEN) furniture
Lionel T Dean, (ENG) FutureFactories
David Taylor, (SWE) silver/metal design
Elina Rebers, (FIN) textile: fabrics
Maria Zitting, (FIN) textile: wall hanging
The FutureFactories work exhibited, comprised Creepers, Tangle, Tuber, and Icon. Creepers
was used for the event publicity, Figure 163.
189
Figure 163
‘Funky’ Event Invitation, Norsu Gallery. Finland 2007
190
6.4
Entropia
Entropia was launched at Light and Building, Frankfurt, April 2006, by Italian lighting and furniture
manufacturer Kundalini. Kundalini celebrated its tenth year, in 2006, and has a reputation for
blending tradition and technology, mixing hand blown glass with 5-axis waterjet cutting, for
example. At the time, Kundalini had no previous experience of Rapid Manufacture or even Rapid
Prototyping. This design was believed to be the first Rapid Manufactured retail product by a
recognised manufacturer i.e. other than an RP service provider (Dean 2006). The significance is
that Kundalini’s primary function is retail manufacturing. Rapid Manufacture could be considered
as a logical step for companies possessing Rapid Prototyping capital equipment.
Unused
machine capacity can be turned to in-house production and the artefacts manufactured serve as
marketing tools, thus promoting the capacity, technology and expertise of the company. The
products effectively become larger scale versions of the samples produced by machine vendors,
only in this instance saleable. Materialise, for example, were one the first RP service bureaus in
Europe and one of the biggest by the time they began their own production.
The principle component in Entropia is a 120mm diameter spherical diffuser, produced in laser
sintered nylon, Figure 164. The design is available in table, suspension, and wall variants, Figure
165. It retails at between 400 and 500 Euro, depending on the model; a price comparable with
traditionally manufactured artefacts, from design-led manufacturers in materials such as hand
blown glass and ceramic.
Kundalini’s founder, Gregorio Spini, wanted to use the geometric freedom of RM to the full in a
product that would baffle; whose conventional production would be inconceivable. A geometry
without pattern or order, without any hint of logic that would suggest how the form had been
created. At the same time, this was to be a commercial retail product with its production price
determined by the market.
191
Figure 164
Entropia
192
Figure 165
Table, Wall and Pendant Entropia Variants
The cost of Rapid Manufacture is to a large degree predictable at the concept stage. It is
generally independent of the geometry and is governed by the number of layers built. It was clear
that the best chance of commercial viability lay in packing the build chamber efficiently with
multiples of a compact form (Dean 2006, Appendix 6). The cost of a full chamber build would be
divided between the number of pieces it contained. Accommodating one or two extra units would
have a significant impact on production price per unit and may prove to be the difference between
viability or otherwise. Early in the concept design stage, a spherical form emerged as the most
likely solution. Externally the design needed to be as compact as possible, internally appropriate
clearance would be required around the hot light source. Working back from a market derived
target production price, gave a maximum diameter of 110mm. After prototyping, however, this
diameter was felt to be a little small and that an increase of diameter to 120mm would have a
marked impact on the products’ received value, outweighing increased production cost.
A
diameter of 120mm was fixed upon therefore as offering the best balance of perceived value and
manufacturing cost. Despite its compact nature, the design would use G9 halogen fittings; these
run at line voltage, eliminating the cost of a transformer. Early pieces were built at Protosystems,
Parma, Italy, who advised on the project along with EOS, Italy. Figure 166, shows a batch of
parts being unpacked from an EOS sintering machine at Protosystems.
193
Figure 166
Entropia Parts at the Break-Out Station, Protosytems, Palma
It was important to the client that the design was taken as far from conventional industrial
manufacture as possible. Complexity was ‘a given’ as any regular form could be produced more
economically by conventional means. The idea was, however, to go beyond awkward geometry.
Although certain forms may be impossible to produce conventionally, undercuts, re-entrant
shapes, and the like, this fact will not necessarily be appreciated by the lay-customer. To the
consumer it makes little difference if a product is produced in one piece via some exotic means or
is made as a well disguised assembly of cheaper components.
It was necessary to convince the buying public that this plastic product was as precious as, for
example, hand blown Murano glass that would sit beside it in the Kundalini Collection. The idea
was to achieve a crafted object that was as far from ‘design-for-manufacture’ as possible. The
idea was to remove all regular pattern and logic from the form; there could be no repeats or
symmetry. At the same time and perhaps in contradiction, there needed to be evidence of
human craft.
194
The solution was to adopt a rule based approach of previous work; but to apply this to elements
within the design rather than to individualising the artefact as a whole.
Parametric design
templates were created for a series of features that would appear in varying numbers throughout
the form.
These templates dictate the underlying style of a specific feature, but allow
considerable flexibility in its particular embodiment.
The Entropia design is made up of
approximately 200 features, consisting of ‘flower’ forms and various ‘leaf forms’, Figure 167.
Figure 167
Flower (top left) and Leaf (bottom right) features
These features are assembled into strings, which are interwoven to create the overall spherical
form. Each time a flower or leaf feature is repeated in the assembly, the parametric template is
modified to give a slightly different outcome, Figure 168.
.
195
Figure 168
Time Lapse Morphing of the Flower Feature
The result is the impression of a natural phenomenon, such as coral. There are clear patterns to
the ‘growth’; the form appears to have evolved rather than to have been constructed, Figures 169
and 170.
Figure 169
Entropia detail (i)
196
Figure 170
Entropia detail (ii)
Although this was not a primary consideration, Rapid Manufacture enabled an extremely short
product development cycle. The concept was agreed, in early January 2006, and the product
launched in April 2006.
While the design task was long and onerous, digital design and
manufacture allowed prototypes to be built, based on sections of the design that were complete.
The design was manufactured in a series of increasingly complete releases, with activities such
as catalogue photography and testing taking place in parallel to an ongoing design development.
Entropia has been in production since 2006 and went on to be featured in the 2008 Publication
Process, edited by Jennifer Hudson and published by Laurence King publishing, Figure 171. A
morphing Entropia wireframe image featured on the publication cover.
Figure 171
The 2008 Publication “Process”
197
198
6.4
Section Summary
The commercial production of Creepers, RGB and Entropia prove the commercial viability of the
concept. Their commissioning and subsequent sales also indicate a desirability in the project
aesthetics.
Although these designs are serially produced they were created with individualisation in mind and
their forms reflect this. Three variants of RGB were produced and the potential is there to create
hundreds more.
Entropia and Creeper perhaps suggest that there is not always a need to
individualise. Their designs are chaotic and irregular making it difficult to identify similarity or
difference. The experience of these projects shows that there are issues such as marketing,
packaging and distribution to deal with beyond the economics of design and production.
199
200
7.0
Later works
7.1
Artenoma
Artenoma is both the name of a design and the name of the commissioning, producing company.
A group of Italian design enthusiasts formed a company called Art Change in 2001. Their idea
was to market Artenoma; collectable sculptures by named artists, designers and architects
produced in a range of styles and materials. The sculptures, no bigger than a “bar of soap”,
would be sold individually and as box-set collections. The first set, ‘The 2003 Collection’, was
launched at the Milan Furniture Fair in 2003. It comprised four pieces by Alejandro Ruiz, Antonia
Campi, Stefano Giovannoni and Ettore Sottsass in bronze, porcelain, aluminium and glass
respectively, Figure 172.
Each piece was numbered and produced in a limited edition, for
example the Campi piece was an edition of 500, the Sottsass an edition of 1500.
Figure 172
Artenoma 2003 Collection
The researcher was invited to submit a design for a second, 2004 Collection, alongside designs
from James Irvine, Ritsue Mishima and Ross Lovegrove, Figure 173. Unlike previous Artenoma,
the FutureFactories pieces would be an individualised set, with the production run set at 500
pieces.
201
Figure 173
Proposed Artenoma 2005 Collection
Unfortunately the company ceased trading before production began, but the design remains
significant to this project. The only physical limitation in the brief was that the artefact should fit
within a 10cm cube presentation box, which was standard across the Artenoma collection. Prior
to the FutureFactories commission, Artenoma artefacts or, Artenome - the Italian plural, had been
limited edition, serially produced, pieces.
The FutureFactories Artenoma was to be an
individualised run, released in stages. As the production run was fixed a key frame animation
approach similar to that used in the Tuber design was employed. An Artenoma would be
produced from every tenth frame of a 5000 frame, approximately 3 minute animation. The design
would consist of four intersecting volumes. These would each run corner to corner, between the
eight nodes of the presentation box. In the initial pieces these volumes would be gently curved.
As the production run progressed the design would become increasingly complex and
interwoven.
The design was to be introduced in stages, with an animation to accompany each
realise. This was a practical device to reduce the amount of design investment and development
time required, ahead of the launch. In Artenoma, the animation would proceed outside of the
consumer’s control and would be common to all users. Examples of Artenoma iterations can be
seen in Figure 174 and an SLS prototype, in Figure 175.
202
Figure 174
Artenoma iterations
Figure 175
Artenoma SLS prototype
Frames from the Artenoma animation can be seen in Figures 176 and 177.
203
Figure 176
Artenoma Animation (i)
204
Figure 177
Artenoma Animation (ii)
205
Figure 178
Cornuta Iterations
206
7.2
Cornuta
In 2005, the author began a collaboration with software developer Genometri, Singapore (Section
1.8.1). The researcher became a beta-tester of the Genovate software plug-in for Solidworks: a
package that randomly reassigns dimension variables within a Solidworks model. Genovate is
somewhat crude, taking random leaps in solution space, added to this Solidworks, as a solid
rather than a surface modelling package, lacks suitability for the complex 3D forms seen in the
previous, surface modeller derived pieces. Genovate was never-the-less interesting; as working
within a robust parametric CAD package, post generation mapping operations, such as the Tuber
fillets (Section 4.4.3) could take place automatically. Added to this the geometric viability of the
model would be automatically tested.
A case study design, Cornuta, was created for the
Solidworks’ World Conference and Exhibition, Las Vegas, 2006.
The Cornuta form is simple,
compared to earlier FutureFactories pieces: it was significant, however, in that it was the first
meta-design to yield one-off iterations on demand. A series of Cornuta iterations can be seen in
Figure 178.
The design is similar in concept to RGB (Section 6.1), with separate volumes meeting at a
common mounting point. There are three tubular volumes all of which start from the same
hemisphere.
This hemisphere houses a high-intensity LED, which illuminates the internal
surfaces of the volumes, Figure 179.
Figure 179
Cornuta Cross-Section
207
(i) What Morphs?
As with previous work the form is defined by a series of circular control curves, the positions and
scales of which are free to morph within pre-defined bounds. In order to link the morphing
effects, and to create a flowing form, the position of the control curves is not controlled directly.
Instead, CVs on a three dimensional spline curve are moved. The control curves are positioned
on this curve and their positions update accordingly. They remain normal to the smoothed curve
ensuring a flowing form. Rather than use full circles, each volume is modelled in two halves,
Figure 173. This is a device to prevent axial twists in the form. Each of the three ‘horns’ is
modelled in the same orientation and then two rotated to give equal spacing. In fact, the spacing
is not equal, but subject to random morphing 120o +/- X.
Figure 180
Cornuta Geometry
208
7.3
Holy Ghost
In 2006, the author was invited to submit a proposal for one of four commissions for an exhibition,
Perimeters, Boundaries and Borders, to be held in Lancaster, UK, October-November 2006. The
exhibition was organised jointly by Folly and Fast-UK.
Folly is a non-profit digital arts
organisation based in the North West of the UK. Fast-UK (Fine Art Science and Technology in
the UK) is an organisation that supports and encourages practitioners using digital and/or
electronics technologies in their work. The author’s successful proposal for the commission was
to be an adaptation of a well known, pre-existing, design; the Louis Ghost chair.
The Louis Ghost chair was designed by Philippe Stark, for Italian furniture manufacturer Kartell in
2002, Figure 181. The one-piece clear polycarbonate (solid colours were introduced later) chair
is an ironic take on 18th-century Louis XVI salon chairs. It has proved a highly popular piece and
is widely recognised as a symbol of modern design. Stark’s press release rated this piece of
furniture as his, “Most successful, technologically and culturally” and, “A true object of modernity
which represents a dematerialisation of design” (Stark, Network, 2008).
In 2006, design
commentator Stephen Bayley, took a chainsaw to a number of the chairs to promote an article
bemoaning the descent of design, “From ennobling industrial art to the silly designer chair”
(Bayley, 2006). Frivolous or masterful, the chair is ubiquitous and instantly recognisable by even
the most mildly design-aware. It is both an everyday object and an object of desire. It was for
this iconic status, that it was selected for adaptation.
Figure 181
Louis Ghost Chair, Philippe Stark, Kartell, 2002
209
Of the four works commissioned for the exhibition, Holy Ghost caused the most controversy
between the commissioning partners (Marshall 2007). One reason for this was that the author
had a reputation for lighting objects, but was charting new territory with a furniture piece. The
biggest issue, however, and the most significant in terms of this thesis, was in the communication
of the concept. The norm in design industry practice is that projects are initially outlined by
concept sketch-work. A multi-million automotive design project, for instance will typically begin
with a quick loosely defined sketch-work that, whilst short on detail, identifies key lines, proportion
and character. A work that is computer generated via scripts and formula cannot be easily
summed up in this fashion before the rules are created. Computer programming is used to create
complexity that, almost by definition, cannot be readily mentally pictured and committed to paper.
The early sketch of Holy Ghost produced for the exhibition panel, Figure 182, was only intended
to specify the areas of the chair that would be replaced and to hint at a level of intricacy and
ornamentation. This drawing raised issues amongst the panel as inappropriate aesthetic
judgements were made on it. Fortunately Fast-UK were fully behind the project and Holy Ghost
was, “The one work in the exhibition that Fast-UK was completely committed to having” (Marshall,
2007), despite reservations from Folly.
Figure 182
Holy Ghost Concept Sketch
210
In Holy Ghost, the back and arms of the original Louis Ghost chair would be replaced entirely.
These sections would be cut from the one piece polycarbonate injection moulding and a Laser
Sintered component attached in their place, Figure 183. The idea behind adapting an existing,
recognised, design was to highlight the creative possibilities for personalisation and subversion
that digital technologies enable. ‘Sampling’ in which portion, or sample, of one sound recording is
used as an ‘instrument’ in a second piece, has been part of the music industry since the switch
from anoloque to digital production in the eighties. Collages, found footage and video montage
are commonplace in film and graphics. Until recently however, 3D artefacts have been more
difficult to access and manipulate. Digital technologies employed in reverse engineering, along
with additive fabrication RM, make it increasingly possible to access, manipulate and reproduce
3D artefacts.
Figure 183
Portions of Louis Ghost Chair Replaced
A well as being a vehicle for stimulating creative debate, the hybrid structure of conventional and
RP technologies offered some practical advantages. One aim was to build the SLS component in
one-piece, minimising assembly and moving as close as possible to the ‘straight from the
machine’ aspirations of the thesis. A whole chair could not be accommodated in anything other
than SLA equipment (as demonstrated by Patrick Jouin Section 1.6.4).
SLA technology,
however, would not give sufficient strength for the chair to be used and would require significant
hand finishing.
SLS was the only practical option that would produce a functional artefact.
Limiting the RP structure to the back and arms meant that the part could be accommodated in
one piece in the EOS P700 laser sintering machine. An additional advantage was the use of the
standard chair’s injection moulded legs; this removed the areas of highest mechanical stress from
the structure, thus limiting the physical demands on the RM component.
211
The commission budget was not sufficient to cover the commercial production costs of such a
large component.
Sponsorship in kind was provided by RP service providers, 3T RPD of
Newbury. However, 3T requested that the two arms were built separate to the back, Figure 184.
Although the back and arms could technically be built as one, the savings associated with filling
the build volume efficiently were too significant to ignore. In practice the attachment of the arms
caused some difficulty. From a visual point of view the joint was easily hidden below the SLS
nylon surface. Mechanically, however, the SLS with its textured finish, proved very difficult to
bond. The first two chairs had a hole and peg locations built into the part, the bonding of which
proved prone to failure. In #3 the joint was dowelled with aluminium rod and the dowel pinned at
either end.
Figure 184
Holy Ghost Arm Separation
Three iterations of Holy Ghost have been produced to-date, out of a planned edition of 10.
Iterations 1 and 2 were produced for the Perimeters, Boundaries and Borders’ commission in
Autumn 2006. Iteration 3 was commissioned in 2008 by a private Milanese architect. As the
script is quick and easy to run and only a few are being produced there has been an ability to
select preferred outcomes. The biggest factor aesthetically is the number of buttons allocated by
the script from a range of 23-27 (4.4.3). The first two chairs were selected with the aim of
achieving a contrast. Iteration 1 has 26 buttons making it the back visually ‘cramped’ and ‘busy’.
Iteration 2 has 24 buttons and is by contrast, a little sparse with several gaps. For the third
iteration a compromise was sought and an outcome with 25 buttons selected.
212
The Perimeters, Boundaries and Borders Exhibition was held 29th September - 21st October
2006, at the CityLab, Lancaster, UK.
The exhibition comprised of four new commissions
(including Holy Ghost) and 16 selected works.
Alongside the exhibition there was a
complimentary symposium with presentations by the artists.
The aim of exhibition was to,
“Present the very latest examples of work that blur the conventional boundaries of art and design
practice through the use of technology” (Brown, 2008). Iterations #1 and #2 of Holy Ghost were
exhibited on a low plinth alongside a projected animation, Figure 185. The animation was back
projected onto a frosted screen which was suspended alongside the chairs themselves. The
Virtools’ script (4.4.3) was projected running in real-time, constantly rebuilding itself, with a build
cycle lasting approximately 6 minutes. Feedback from the curators suggested that the real-time
animation was too slow and not as exciting as the pieces themselves. For later exhibitions, the
real-time animation was replaced by a show-reel made up of edited Virtools clips. The show reel
is visually more dramatic but less significant from a technical standpoint.
Figure 185
Perimeters, Boundaries and Borders, Lancaster 2006
213
Figure 186
Holy Ghost iterations
214
Three iterations of the design have been physically produced, Figure 186. These have been
included in curated exhibitions, Digitability, Berlin and Trans-Form, Paris.
Digitability held as part of the DesignMai Festival, May 2007, and was curated by Atilano
González, Figure 187. The exhibition aimed to take a, “Close look at the relationship between
digital technology and design, paying special attention to digital crafts (digital technologies such
as laser sintering now being applied to products as well as customization and a revival in
ornamentation)” (DesignMai 2007). Holy Ghost iterations, #1 and #2, were exhibited along with a
new design, Pallavi (Section 7.4).
Figure 187
Digitability DesignMai, Berlin 2007
Trans-Form was curated by Elizabeth Leriche. It was held 25-29th January 2008, to coincide with
the Paris furniture and furnishings fair, Maison et Objet. The exhibition examined transformation
and distortion in design; the application of, “Tools primarily reserved for the digital processing of
the image are applied to the materialised object” (Leriche 2007). The exhibition featured physical
iterations, #1 and #2, along with two widescreen show reels of the build animation, Figures 188 –
184.
215
Figure 188
Trans-Form, Paris 2008 (i)
Figure 189
Trans-Form, Paris 2008 (ii)
216
Figure 190
Trans-Form, Paris 2008 (iii)
Figure 191
Trans-Form, Paris 2008 (iv)
217
218
7.3.1
Metal Plated Holy Ghost
The white texture finish of the SLS marks easily and is difficult to clean. In an attempt to achieve
a more durable product, Holy Ghost #2 was metal plated, Figures 192 and 193. In this process
the SLS is sealed and sprayed with a conductive paint. It is the plated with first copper then
nickel. As well as providing a more durable decorative finish, the metal skin (approximately 300
microns) provides considerable additional strength. The plated chair was exhibited at the Art and
Design Gallery of SIGGRAPH 2009, New Orleans, curated by Makai Smith and the Cheongju
International Craft Biennale, Korea.
Figure 192
Nickel plated Holy Ghost #2, Front
Figure 193
Nickel plated Holy Ghost #2, Rear
219
Figure 194
Pallavi
220
7.4
Pallavi
Pallavi is a 400mm diameter spherical chandelier, Figure 194, created for Designmai Berlin 2007.
It was produced using an adaptation of the Holy Ghost button geometry: sixty of the trumpet-like
forms are clustered into a sphere. Each trumpet is lit by a colour change LED at its base, shining
up the funnel form.
Rather that being positioned at random, the base of each trumpet is
deliberately placed and orientated on a dodecahedral matrix; five to a pentagonal face, Figure
195.
Figure 195
Pallavi geometry
The outer edges of the trumpets are allowed to move and distort within a spherical volume, while
their bases remain fixed.
The aim is to have sufficient curvature in the funnel-like internal
passageway of each trumpet, to prevent direct line-of-sight with the LED at its base, Figure 196.
221
Figure 196
Pallavi LED position out of direct view
The Pallavi trumpets are manipulated independently and each one is different. They are built in
laser sintered nylon in groups of five on a pentagonal mounting, Figure 197. The finished design
can be seen in Figures 198 and 199.
Figure 197
Pallavi Trumpet Cluster
222
Figure 198
Pallavi Lighting Effects
Figure 199
Pallavi Close-Up
223
224
7.5
Jewellery
To this point, commercial interest had been in either serially produced identical retail products, for
example, Creepers and Entropia; or in selling high value one-off gallery pieces, such as Tuber9
and Holy Ghost. The quest remained to achieve a significant run of individualised pieces, beyond
the collections produced for exhibition. An exploration of jewellery seemed a logical step (Dean
2008, Appendix 7). Jewellery is a market in which intricacy and complexity are valued and where
premiums may be paid for exquisite form. It is also a market in which the boundaries between
industrial production and craftsmanship are blurred and where the value of a bespoke piece is
readily appreciated. The jewellery market is, however, generally conservative; particularly in
terms of material. Precious metals associated with jewellery have intrinsic value, are durable,
inert and long lasting. The polymers of RP, beautiful as they can be, would not win over all but
the most avant-guard consumer in this field. A metal, if not a precious metal, seemed essential to
‘compete’ in this market.
Experimentation with jewellery came about through a connection with the Jewellery Industry
Innovation Centre (JIIC), in Birmingham, during the organisations tenth anniversary celebrations
in 2007. The JIIC is part of the School of Jewellery, within Birmingham City University (formally
UCE Birmingham). Based in Birmingham’s jewellery quarter, the centre encourages innovation
within jewellery and high value goods industry SMEs (small and medium enterprise), through
design, research and technology. As a case study for a two-day industry event, the centre
prototyped a series of FutureFactories design proposals. These designs were produced indirectly
in silver. A wax pattern was built by additive manufacture and the piece investment cast from this
master. Three designs were produced, Aorta, Puja and Icon.
7.5.1
Aorta
The starting point for Aorta was the traditional heart necklace and the notion of substituting the
caricature heart with something closer in form to the organ it represents. Aorta uses similar
geometry to the Tuber family of lights: in this case with two tuboid volumes that wrap around each
other, Figure 200. A closed loop of chain passes through both volumes end to end. Each volume
is punctured around its middle revealing the chain and a red enamel inner surface.
225
Figure 200
Digital Rendering of Aorta 2007
7.5.2
Puja
Puja continued the exploration of a more literal heart organ form. In this case four limbs emerge
from a volume. A single closed length of chain loops twice through the volume, passing one
though each limb, Figure 201.
Figure 201
Digital Rendering of Puja, 2007
226
7.5.3
Icon
In Icon the geometric techniques and leaf forms first explored in Entropia (6.3) were developed
further. Whereas in the interwoven leaf-forms of Entropia were used to create a hollow sphere,
the geometry of Icon is allowed to fill an ellipsoid volume to create a pendant form. A continuous
string of leaf-forms passes to and fro across the bounding envelope with the form of the leaves
conforming to its contours, Figure 202.
Figure 202
Digital Rendering of Icon 2007
Production of the three jewellery pieces was far from straight forward. The designs were first
digitally printed in wax, then coated with refractory material to form a mould for the investment
casting process. It was essential that the coating filled the fine internal passageways of the wax
pattern and that this material could be cleared away after casting. Aorta and Puja both presented
problems with keeping the internal passageways clear enough to allow the chain to pass through.
This places design limitations on the length and internal diameter of such internal voids. In Icon it
was the fine strands of the design that gave rise to problems. Molten metal was required to flow
through and fill an irregular assortment of fine passageways and voids, which again placed
limitations on what could be produced.
The results from all three pieces were beautiful: but in production terms, heavily reliant on the
skills of the JIIC to cast and polish the intricate forms (Dean 2008, Appendix 9). Complexity can
be expected in Rapid Manufacture in order to exploit the process’s freedoms and to justify the
expense relative to conventional methodologies. FutureFactories designs intentionally push the
227
boundaries of what the RP industry can produce. Whilst these forms can be ‘printed’ in wax, or
substitute polymer honeycomb, it is clear that coupling this digital fabrication with a secondary
conventional casting process would impose significant limitations.
Even if the forms could
ultimately be cast, setting up the casting with appropriate sprues and risers requires considerable
experience: once worked out, the process is reasonably robust but it is seldom right first time. In
an extended run this is not a problem with set-up costs amortised over the run, with
individualisation, however, it would be.
A small change in form might require a complete
reassessment of how the piece would be cast.
It became clear that the only way to exploit the full potential of additive manufacture was to build
directly in metal.
Metal sintering whist technically feasible early in the project had been,
prohibitively expensive, hard to access and crude in terms of finish. During the course of the
project the technology had improved and become more widely available. Working with EOS and
3T-RPD the possibility of using Direct Metal Laser Sintering in the EOS M270 machine was
explored. Of the three designs produced, Icon received by far the strongest reaction and the
decision was made to start with this piece.
(i) The Direct Metal laser Sintering, DMLS, Production of Icon
In spite of their free form potential, RP technologies each have their own technical limitations to
accommodate.
Powder-based methods of rapid prototyping are generally self-supporting for
features such as overhangs and undercuts. In production these features are supported by loose
un-fused powder thus avoiding the need to fabricate a support structure along with the part. If
support structure is required, it must be removed in potentially difficult and costly secondary
operations. The lack of support structure in the laser sintering of plastics, coupled with the
materials’ mechanical performance, accounts for the dominance of this methodology in Rapid
Manufacture.
In spite of the fact that it is powder based, the Direct Metal laser Sintering process with the
associated forces of contracting metal, does require support, at least to anchor the part to the
machine platform. Each element of the form begins with its underside surfaces linked directly to
the bed. In addition to the support underneath, any surface that develops beyond a given angle
to the vertical, typically around 40 degrees requires supplementary support from below, Figure
203.
228
Figure 203
DMLS Support Requirements
Support material is designed for easy removal. In practice this removal is time consuming and
leaves surface marking. The problem of support structure removal becomes more acute if the
free form potential of RM is to be exploited. In the Icon design, support material within the form,
attached to inaccessible inner surfaces would be virtually impossible to remove completely. In
addition to the issues of access, the support structure, being of the same material in this process,
only breaks away easily when the removable support is finer and weaker that the geometry it
supports. The treads of Icon are as fine as will build reliably: this makes the safe removal of
support structure difficult.
To use DMLS as an effective industrial production tool, support
structure, whilst it could not be avoided entirely, had to be limited to easily accessible areas.
Working with EOS and 3T technical staff, the original Icon concept was adapted for DMLS
production, Figure 204.
229
Figure 204
The Original Icon Design on the Left Alongside its DMLS Adaptation
The first decision was to orientate the parts in the build chamber upright, building from top to
bottom. Building upside down meant that more area for support was available at the base of the
tampering form as built. For cost effectiveness, parts should be orientated with the minimum
height in the build chamber. This would have meant building the part on its side and had to be
discounted, owing to the internal support structure this orientation would have required.
The original Icon is one chaotic intertwined loop. This was replaced by a four, simpler, separate
loops generally running top to bottom in order to restrict the angle to the vertical. Each of the
loops starts at the same point touching, and running parallel to, the supporting platform. The
loops form a filled-in convex surface where they are overlaid at the base regardless of the
morphing. The support structure could be limited to this convex surface at the top of the design
(the base as built) where it can be easily removed and the inevitable witness marks polished off,
Figure 205.
230
Figure 205
Icon DMLS Build Orientation
The only other area of support required was to form the round ‘eye’ feature, a key element in the
character of the design that, it was felt, had to be retained. The supports for the eye feature,
across its centre, could be removed with a punch with relative ease.
Stainless steel, cobalt chrome and titanium are commonly used in DMLS. Gold is technically
possible.
Swede Lena Thorsson has developed a gold alloy for DMLS and together, with
designer, Towe Norlén, created gold chain designs using the process, Figure 206. In practice
however, gold as a build material is currently inaccessible, amongst other reasons the build
chamber must be filled with powdered metal to the build height, requiring substantial investment.
Smaller sub-chambers have been developed for gold research and dentistry but the availability as
a bureau service is not great.
231
Figure 206
18K DMLS Gold Chain, Towe Norlén, Lena Thorsson
Titanium, whilst not a precious metal; is recognised and valued as a contemporary jewellery
material.
For this reason, it was selected as the best option for Icon. Titanium cannot be
soldered, which limits the forms that can be created conventionally. Laser welding is possible but
it is an expensive, skilled and time consuming process. The use of titanium in intricate pieces is
consequently rare. Trial pieces of #1 were successfully built by EOS, Germany, in titanium and
stainless steel in January 2008. As a cost saving measure, the design was scaled to 85% of its
original height, Z, with X and Y dimensions unchanged. Following this success, a series of 25
pieces were produced as an initial batch of what is ultimately intended to be an edition of 100
individualised pieces, Figure 207. Out of the batch of 25, 24 built successfully with one failure.
All 25 were built together alongside each other on the platform, minimising production cost.
232
Figure 207
First batch of titanium Icons
Icon iterations #1 to #24 can be seen in Figures 208 and 209.
233
Figure 208
Icon Iterations 1 -12
234
Figure 209
Icon Iterations 13 - 24
235
Added to the challenges of the casting process, the intricate forms produced require skilled hand
finishing. The need to access the form for polishing can again limit the design possibilities.
Tailoring the form to the DMLS process had solved the production issue, but the problem of
surface finish remained. After building and shot peening, the surface is regular but textured,
Figure 210. This finish is not unattractive and would have a market. It does not, however, show
off the flowing curves of the form, nor does it fit with the highly finished surfaces of the jewellery
market.
Figure 210
Icon Surface Finish After Building and Shot Peening
A level of hand finishing is common in jewellery, but for industrial production this work should not
be too lengthy and should ideally be undertaken with the aid of power tools.
Inner faces,
inaccessible with a polishing mop, and different issues occurring with each piece would not be
tenable. To meet the industrial production aspirations of the project a methodology was needed
that could be automated and would cope with a variety of geometries.
A study was made of conventional polishing procedures, with the help of the JIIC, Birmingham,
Figures 211 - 213. Tumbling with steel shot and a water blast with glass beads had virtually no
effect. A centrifugal barrel polisher, with ceramic cones as the medium, had some effect but only
on the external surface and this after 25 hours. There are two reasons for the ineffectiveness of
these processes: the grade of titanium, which is particularly hard, and the size of the polishing
media relative to the forms being processed. Tests are ongoing at the JIIC using smaller media.
Historically, the surface of titanium would be treated with perchloric acid. This approach was
236
discounted due to the dangers of using this chemical; the use of which is restricted by Heath and
Safety legislation.
Figure 211
Conventional Polishing – Steel Shot
Figure 212
Conventional Polishing – Wet Blast
237
Figure 213
Conventional Polishing – Ceramic Cones
Polishing the external surfaces only using the barrel polisher, followed by the use of a polishing
mop whilst leaving the internal surfaces untouched, was considered as a possible option and the
approach adopted for the first batch. A longer term solution came with the use of the MicroMachining Process, MMP of Best in Class, Switzerland. Working with 3T, Newbury, pieces were
prototyped in stainless steel and cobalt chrome. Figure 214 shows a Cobalt Chrome part with the
support structure still in place. Some damage can be seen at the top left of the part, as built.
Such flaws in the building of delicate parts, such as Icon, necessitated a modification to the
standard DMLS machine. The re-coater arm that spreads the layers of powder was changed
from a rigid blade to a carbon fibre brush, to avoid the wiper fouling and breaking delicate
sections. The revised re-coater is now standard.
Figure 214
A Full Z height Icon in cobalt chrome
238
Two of the stainless steel parts were finished by the Best in Class Company’s process, which
employs a mechanical-chemical reaction to produce an automated equivalent of traditional
polishing. The process requires a relatively substantial mounting boss to be built onto the part by
which it is fastened for the operation. This is not a huge problem in the case of Icon, as the boss
was built in place of the support structure at the build base. This feature needed to be more
substantial than the fine support structure it replaced, but on a convex external surface it could be
ground off and polished with relative ease. The automated polishing proved most successful on
the external surfaces, but there was also a marked improvement on the internal surfaces, Figure
215.
Figure 215
MMP Polished Stainless Steel Icon
The results could be improved further with development, principally by increasing the level of
energy used; the security of the mounting now being proven. The individualisation of Icon would
present no problem for this process and would simply require the addition of a standard boss to
each part.
The efficacy of the Best in Class Company’s process coupled with DMLS, was proved further with
the production of a trophy for the 3rd International Conference on Rapid Manufacturing,
Loughborough.
A design was created, by the author, which was produced by three
methodologies, DMLS, SLS and SLA, to signify 1st, 2nd and 3rd. The DMLS piece made was
approximately 220mm high and was polished using the Best in Class MMP process, Figure 216.
239
Figure 216
DMLS Trophy 2008
Titanium can be anodised and this option was explored, once again with the aid of the JIIC,
Birmingham. Test pieces were produced with blue, gold and purple finishes, Figure 217. As a
spectrum of colours is produced according to the time that the current is applied, this could also
become a randomly generated variable.
Figure 217
Anodised Icons
240
7.6
Puja Table Lamp
The Puja jewellery design was revisited in the Puja table lamp, Figure 218. The design was cast
first in aluminium and then later in stainless steel. The stainless steel lamp is 600mm tall and
weighs 12kg. The pattern was manufactured by printing cross-sections on paper and cutting
these shapes from thin board. The board sections were then assembled on dowel pegs and the
resulting, stepped form, smoothed. This is a laborious method, but an effective way to achieve
strong, inexpensive, large scale pieces. The system could be automated further by CNC laser
cutting, or routing, the sections direct from CAD data.
Figure 218
Puja Table Lamp Design
The lamp is lit by four high intensity LEDs, one per trumpet. This design was initially intended for
individualisation using the Genemetri software (Section 1.6.1). In the event, the character of the
form seemed too attached to single configuration and no further variants were produced. Even
the vagaries of casting the form proved a problem. The design was sand cast in sections and
welded together before polishing. The first attempts were inaccurate and, while the discrepancies
were relatively minor, the character of the design seemed lacking, Figures 219.
241
Figure 219
Aluminium (right) and First Stainless Steel Prototypes
The first finished Puja can be seen in Figure 220.
Figure 220
Puja Table Lamp, Stainless Steel, 2008
242
7.7
Section Summary
The First Collection (Section 5) effectively demonstrated the concept. The retail design projects
resulting from this work (Section 6) then proved the commercial viability and desirability of the
design ideas. The practice documented in this section proves the objectives set out at the start of
the thesis. The case studies presented begin with and are driven by computer programming.
The programs for Cornuta, Holy Ghost, DNA (I and II) and Superkitch all generate unique
solutions each time they are run. Differing approaches are demonstrated, the morphing form
seen in the first collection and a building block approach developed to allow for more fundamental
change. The artefacts produced come far closer to the project ideal of retail products direct from
the machine and there is little of the post-finishing seen in the first collection. In the Icon pendant
design, a collection of 25 iterations, is the most extensive individualised production run to-date.
Each is visibly unique yet, at the same time, recognisably an ‘Icon’.
243
244
8.0
Summary and Conclusions
This thesis develops the ideas set out by Soddu (Soddu 1997). Whereas Soddu’s focuses on the
generative computation and the artefacts remains for the most part a graphic exercise; this thesis
has placed emphasis on the viability of the designs produced. It was a project objective that the
artefacts generated be both desirable and functional.
Functionality has been demonstrated
through lighting that has operated throughout an extensive program of exhibitions. In consumer
product terms pendant lighting has limited functional requirement which was to a large extent why
it was selected, nevertheless the need to house fixed dimension components and maintain
internal clearances for wiring requires technical consideration that is not seen in previous work.
The desirability of the artefacts has been demonstrated through the extensive publication and
exhibition of the work, perhaps most notably the acquisition of Tuber9 by MoMA the Museum of
Modern Art in New York, 2005.
Ron Arad’s bouncing vases (Arad 2000) demonstrate the potential for extensive editions of
unique artefacts by combining parametric CAD with animation and direct digital manufacturing. In
this work iterations are derived from fixed length animations and the animations created from a
relatively simple predictable mechanism. Whilst in principle each animations can yield hundreds
of frames (Arad 2000) the potential variation in the models presented is limited. They are based
on a simple mechanical (in the virtual realm) coil spring action with predictable outcomes. Whilst
there will stark contrast between the ‘compressed’ and ‘extended’ extremes there will be little
discernable difference between individual frames. An aim of this thesis is to achieve obvious
visible difference between iterations while maintaining a coherent design identity.
successfully demonstrated in the Icon pendant jewellery.
This is
The 25 pieces produced are all
identifiable as the Icon design, while each remains distinctly unique.
This thesis has successfully combined parametric CAD, computer scripts and digital
manufacturing. The Virtools scripts created for Holy Ghost, DNA and Superkitsch can be run on
virtually any pc with the freely available Virtools player. In a matter of minutes they will generate
a unique iteration of the respective design each and every time they are run.
245
8.1
Technology
8.1.1
Additive Fabrication
When the project began, Direct or Rapid Manufacturing, the production of end–use artefacts
direct from CAD data, was only a theoretical possibility. There were examples of functional parts
being built by additive fabrication for engineering applications in the automotive and aerospace
sectors. There were also examples of customisation via RM in medical applications, hearing
aids, for example. There were however no examples of commercial retail products. High capital
equipment costs, slow cycle times, inferior high cost materials and poor surface finishes, seemed
to limit Direct Digital Manufacturing to technical niches for the short to medium term.
Less than a year after the project began the first RM retail products appeared. Arguably this was
initially as much a PR opportunity from a major RP service provider as a commercial venture;
never the less the euphoric media and public reaction that the products generated established
RM as a mainstream manufacturing process. Two years later Kundalini, Italy, became the first
established manufacturer, other that an RP bureau, to mass market an RM product with Entropia
from this project.
Counter intuitively, current examples of RM have tended to be high end domestic interior
products, a market in which surface finish is of high importance. It is however also a market that
embraces new technologies. Rather than being seen as a coarse equivalent of moulded plastics,
RP materials have been accepted in their own right. The stair-step striations of the layer build
processes are accepted provided the resolution is fine enough.
The finish may even be
celebrated, especially in lighting objects, as discussed in Section 6.3.1.
Today there seems little need to justify Direct Digital Manufacturing, at least at low volumes. It is
however, and will remain for the foreseeable future, a premium process over conventional mass
market methodologies, injection moulding and die casting for example. The use of RM has to be
justified by exploiting the freedoms of additive manufacture and exploring new modes of both
design and consumption.
This project represents one of these alternative approaches to
production and consumerism. It seems clear that computer-based design and fabrication tools
have the potential to affect the wider cultural landscape in profound ways (Rahim 2005, cited in
Marshall 2007).
Wilson suggests that, “Rapid Prototyping is culturally significant because it
moves into territory that is under explored, namely, the linkage of the virtual and the physical”
(Wilson 2002).
246
The excitement of digital manufacturing is not in its gradual adoption by mainstream
manufacturing, this is already underway; but in the increasingly blurred boundaries between
industry and the creative arts.
Exploiting the freedoms of Digital Manufacturing will require new approaches across the
corporate landscape and not merely in production.
The lack of physical production tooling
eliminates a key stage from the design process; the point at which production drawings would be
passed to a toolmaker. After this point, the design would become fixed with further modifications
costly. In Direct Manufacture design development can continue up to the eve of product launch
and even beyond. In the design of Entropia (Section 6.3), design development continued after
the first saleable batch had been produced. A situation can be envisaged, similar to computer
software releases, in which a product improves by degrees through its lifetime, versions 1.0, 1.1,
1.2 etc. In view of the potentially large design investment required in complex designs, products
could evolve in stages. The early design could be sold in modest quantities to avant-garde
buyers in order to fund ongoing development; a 3D equivalent of computer software beta testing.
Over time, the product could gradually become sophisticated and ‘mainstream’ appealing to a
widening market.
Marketing practices, such as an annual printed catalogue mitigate against flexibility. The 2006
Kundalini catalogue ran to 166 pages and featured Entropia on the cover, Figure 221. Over time,
new marketing practices will emerge; consumers’ behaviour will also need to adapt.
Figure 221
Kundalini Brochure, 2006
247
8.1.2
Bureau Culture
Rapid Manufacture is something of a misnomer in the industry generally. Product development
speed was not a primary consideration in this project.
Speed is more applicable to the
prototyping side of additive fabrication rather than industrial production. The emphasis placed on
rapid turn around by service bureaus, which is such an asset in product development, is often a
barrier to Rapid Manufacture. The prototyping bureau industry is characterised by feast/famine
workloads and large amounts of overtime. This model does not lend itself to manufacturing
where efficiency is paramount and profits are made over a longer term. If Direct Manufacture is
to develop, the approach of service providers will need to change. The marketing approach of
the machine manufacturers has already changed in recognition of the sales opportunities RM
offers. EOS, the largest provider of sintering equipment now refers to themselves as EOS emanufacturing solutions (http://www.eos.info/en/home).
In current industry practice the set-up and optimisation of the additive fabrication build is the
province of the RP bureau. Outside of the automotive and aerospace industries, it is rare for a
manufacturer to possess in-house equipment. The placement and arrangement of models within
the build volume is undertaken by RP bureau staff, who are presented with the model when it is
complete or in the final stages of development. As has been demonstrated in the case studies,
the efficient use of the build volume is the most important factor in determining the viability or
otherwise of Digital Manufacturing projects. It is vital that the design is tailored, from the concept
stage, to use the chamber efficiently and to allow the accommodation of a commercially viable
number of units.
Ideally, software for this function would be brought within high-end CAD
packages so that it can be referred to by designers, throughout the design process, rather than
existing as a stand-alone package aimed at RP service providers.
8.1.3
Materials
Early RP materials and equipment were split between the roles of visualisation and testing. This
divide is familiar from the aesthetic models and test rigs of traditional design development.
Models for visualisation would be non functional solid blocks; test rigs would prove function but
convey little of the design’s finesse. The early technological limitations and the inability of RP
equipment to couple surface finish with engineering performance tended to recreate this divide.
Processes such as Selective Laser Sintering (SLS) and Fused Deposition Modelling, FDM, could
provide structure and demonstrate ‘moulded-in’ features such as live hinges and springs clips, the
surface finish however would be coarse. Stereolythography, SLA, with a little hand finishing,
248
could achieve an excellent surface finish but would be too brittle for practical use and would be
dimensionally unstable over time, as the material absorbs water from the atmosphere.
Recent developments in materials and processes have narrowed the gap between visualisation
and testing roles considerably. Increased resolution has given sintered and FDM components if
not a smooth, then a much finer finish.
SLA epoxies have brought increased stability and
improved functionality. SLA resin supplier, DSM Somas, have registered the trademark protofunctional to describe their products. True prototypes are now possible, that both allows aesthetic
assessment and proves function.
In Rapid Manufacture, for the decorative design market, public perceptions of material value are a
significant issue. Plastics, however exotic from a material’s science perspective, do not have the
same cache as, for example, glass or ceramic. Whilst RP plastics have been well received
amongst avant-guard buyers, this may not last and a wider customer acceptance must be
achieved for the market to grow.
Material strength and build resolution can improve perceptions of quality; fine detailing and a
delicacy of proportion communicate craftsmanship. This is exaggerated in lighting applications,
where translucency accentuates fine section thicknesses. He researcher has used thicknesses
as low as 0.5mm and these have built reliably.
recommended for the process.
This is well below that which would be
Many bureaus give a standard warning if model section
thicknesses drops below 2mm. Many FutureFactories designs do not have any sections above
2mm. Robust sections are encouraged in the RP industry because of the cost of the process and
the risk of scrap. This caution however means that the full potential of the processes may not be
realised.
As seen in the RGB ‘eyelashes’ (6.1), the nature of the material can limit the design as much as
the production technology. Material strength must develop to match what can be built. Designs
also should accommodate potential build shortcomings. Ultra-fine sections can be used in areas
that are purely decorative and where the overall structure would not suffer unduly, should fine
edges not build completely.
Redundancy could mitigate against the risk of failure in minor
decorative elements. Where there are massed numbers of features, following no discernible
pattern, the absence of one or two may pass unnoticed should they fail. Creepers (6.2) features
groups of leaf like reflectors on delicate stems, Figure 222. One or two of these reflectors have
periodically been damaged, during transport or assembly. In these cases the remainder of the
stem has been removed and the product distributed as normal.
249
Figure 222
‘Redundant’ Creepers’ reflectors
8.1.4
Sustainability
In consumer product development, the sustainability or otherwise of prototyping equipment is of
little significance compared to the volume production it will influence. More effective prototyping
should, through more considered designs, lead to more sustainable production. Consequently,
little or no emphasis has been placed by manufacturers on RP equipment and processes. As the
industry turns its attention to manufacturing, this issue needs to be addressed.
Additive manufacture is in principle a more sustainable practice than anything subtractive.
In
theory no more material is used in additive manufacture than is employed in the artefact itself. In
many technologies however considerable additional material is indirectly consumed. In the laser
sintering of plastics, artefacts are built within a block of un-fused powder. This unused material is
degraded by the heat of the process forming longer polymer chains. Only a proportion of recycled
material can be used in the process (up to 20% is recommended) if surface finish is not to suffer.
Other processes such as SLA require a physical support structure to be built along with the artefact
requiring both energy and material. There are however less energy intensive processes such as
those of Z-Corp, USA, whose materials are lower tech; being based on the likes of plaster and corn
starch. In this process the powder is 100% recyclable.
Aside from the equipment and processes, the flexibility inherent in digital manufacturing offers
significantly more sustainable models of a consumer society than the one we take for granted
today. In particular:
250

Products can be produced on demand and not stocked;

Ideas can be transferred electronically rather than freight being shipped;

Products can be manufactured locally for local needs; and

If consumers are engaged with the creative process, there is a greater likelihood that there will
be an emotional attachment to the product, which will consequently become less disposable.
251
252
8.2
The Consumer Interface
A key component in any customisation or individualisation system is the consumer interface.
Rather than there being a static product outcome there are potentially infinite solutions. To
appreciate this capacity for change the consumer needs to see more than a single solution. In
the context of this research a capacity for change over time needs to be demonstrated. Screen
based Internet shopping offers the possibility of digital animation: however, a single video clip
played on demand would be insufficient as there should be no cycle to the mutations. Ideally the
consumer should witness the script driven model itself ‘morphing’ in real time.
Systems are envisaged where the consumer is presented with the virtual meta-design via a
website. The consumer may access the website directly on their own computer equipment or via
a sales outlet, a pop-up shop in a gallery or in a department store for example. The web site
would offer a series of current designs. Having selected a product the user would be presented
with a real-time animation of that design in metamorphosis. At any given point, the consumer
may ‘freeze’ the animation, effectively creating a one-off design on screen. Should they wish,
they might then proceed with an order, in which case the relevant digital production files, for
example Standard Triangulation Language (stl), would be generated automatically and sent to
one of a group of participating bureaus for manufacture using appropriate rapid prototyping
techniques. The animation is changing in real time and is thus outside the users control; this
should be the allure of the process. A unique solution is generated which can be accepted or
replaced by an alternative as the animation continues. The interface therefore is aimed solely at
ensuring effective communication of the form rather than being any form of design tool in itself.
8.2.1
Virtual Reality, VR
Visualisation is a significant issue in internet shopping generally where there can be no physical
sample to examine. Web based marketing can be used to access global niche markets, the
difficulty lies in convincing would-be consumers of the suitability of products they are unable to
physically handle.
Visitors to an individualisation website must be able to appreciate the
potentially complex forms with which they are presented, a difficulty exasperated by designs in a
constant state of flux.
Some form of web-based Virtual Reality, VR, experience might help
overcome this difficulty, enabling the consumer to ‘move around’ the design in their own time and
at will.
‘Virtual Reality’ refers to a computer-based activity that mimics real experience. The primary
characteristics of such an experience are that it is three-dimensional and that it is interactive. At
253
its simplest VR may be a 3D image that the user can rotate to view from various angles. At an
intermediate level, it can be an entire scene or virtual world that the user enters and interacts
with. Advanced systems require the user to wear special equipment, goggles and gloves for
instance, that make the illusion of reality more complete.
High-end VR is expensive and
complex: it is the province of games or specialised applications like military training and is not
usually web-based. The less sophisticated forms of VR however, offer significant potential for
consumers to interact with a virtual design. This can be web based and need not require any
specific equipment.
Conventional marketing usually centres on still studio photography. VR content enables websites
to bring the products 'to life' with interactive animation allowing the product to be seen from all
angles. Potential customers are able to examine the virtual 3D model moving, rotating, and
zooming in and out at will. This ‘hands-on’ interaction allows something of a “try before you buy”
experience. Internet shoppers may spend 50% more time in the part of the site that offers
interactive 3D images (Hurwicz 2000). Historic barriers to uptake of VR, the power of PCs
(Personal Computers), the speed of internet connections and availability of authoring tools and
expertise, are now greatly reduced; although there is likely to always be a trade off between
advances in audio visual technology and the computational power to supply them. A level of
audiovisual interactivity should therefore be considered.
Discounting expensive gloves, VR remains an audiovisual experience and consumers will always
prefer to handle the merchandise. A middle ground solution may be to provide a VR interface
within high street retail outlets, where physical samples and a tactile experience of material are
also available.
8.2.2
The User-Script Interface
In the FutureFactories consumption model the consumer needs to interact with the scripted metadesign in real time. This can be achieved in one of two ways; the script can be run on either the
host server or the customer’s PC, Figure 223. Each approach has both merits and drawbacks.
(i) Host Server Approach
The script can be run on the vendor’s equipment, giving the developer complete control over its
operation. Only video content, and no 3D data, is transmitted to the consumer or client PC. This
is a security benefit and there is no requirement for specific content playing software. Video
streaming is extremely common: there are several suitable players available and the vast majority
of PCs will already have appropriate software installed.
254
Figure 223
Host Server and Client PC Approaches
255
A drawback is the need for the server to handle unknown numbers of users simultaneously. This
capacity may be redundant for the vast majority of time. Then there is the issue of pausing the
animation. The animation could not be truly halted (as it would stop for everyone) and some form
of ‘live-pause’ of the kind seen in digital broadcasting would be required. As a positive, the
consumer’s lack of direct control might encourage the air of the on-line auction with the
imperative to act before it is too late.
The host server approach has advantages from a digital manufacturing perspective as no design
data is required from the customer’s computer. The only information required from the client PC
is the precise moment or frame that the animation is stopped. This frame identifies a particular
configuration of the meta-design scripted on the host server and any amount of CAD data can be
extracted from it regardless of the technology at the user end. This form of interface would suit
procedural animation based designs (Section 4.2) such as Tuber where there is no particular
beginning and end to the metamorphosis.
(ii) Client PC Approach
In this approach the computer script controlling the virtual model is downloaded and run on the
client’s PC using web browser plug-in software. 3D software applications are available that offer
both developer and plug-in player levels. The developer version is typically commercial and
allows the creation of sophisticated 3D content for the web.
Content created may then be
distributed via the internet and ‘played’ on web browsers using free ‘plug-in’ software.
Running the script on the client’s PC means that each customer is seeing their own
metamorphosis, which can be stopped and started again from the same point as they see fit. If
appropriate, the design can build from scratch each time it is loaded allowing the user to see
much more dramatic change than might be evident in the mature mutations.
The most obvious drawback with this methodology is that the script has to be tailored to the likely
power of the client’s PC and to a file size which can be acceptably transferred. Added to this a
specific plug-in is required which will need to be installed on first use. There is also an issue with
information transfer. If and when a design is ordered, sufficient information must be transferred
from the client’s PC to either build the part directly or replicate the virtual design on the host
server.
A further potential problem is that of security:
3D data, albeit not necessarily
sophisticated CAD data, is generated on the client’s PC. These issues mitigate against complex
surface models that required heavy scripts and large data sets. There is however great potential
for more simple designs, specifically designed to run on plug-in applications such as the Virtools
3D Life Player (Section 4.4). An example of this is the SuperKitsch bangle created as a case
study using Virtools’ software as the scripting tool.
256
8.3
The Role of the Designer
The development techniques developed though this thesis have in aspects matched the
researcher’s regular practice and elsewhere been diametrically opposed to it. Defining forms
technically through parameters, relationships and cross sections has come very naturally. The
need to begin by setting up rules and relationships rather than through inspirational sketchwork
has been more of a shift. In the early work there was a tendency for form to come before the
script and to drive its development, in part perhaps due to the timing of the exhibition programme.
In the later work however the script came to lead the design process. Script driven geometry
would be conceived and then consideration given to forms that might be derived from it.
The lack of exciting visuals in the early stage of design development may present problems in an
industry with commissioning processes geared around the artist impression, rendering or concept
sketch. This issue was highlighted in the commissioning of Holy Ghost (Section 7.3).
8.3.1
Design for Manufacture
As product design moves into a post-industrial era creative industries and design education will
need to adapt. Design for industrial manufacture is central to, if not the definition of product
and/or industrial design. This has previously meant design for mass-production and the tailoring
of components to specific manufacturing technologies. Design for manufacture has been the
central tenet of design education. There are of course manufacturing considerations in design for
additive fabrication, as is evident in the case studies and the notion of ‘free-form’ fabrication is
something of a myth. Such restrictions as there are, however, do not carry the same weight as
those in conventional manufacturing. The injection moulding of plastics is conceptually similar to
the pressure die casting of metals.
Not withstanding their similarity, switching production
between these two methodologies would be an onerous task requiring new tooling and a huge
investment. In additive fabrication a part can be changed from metal to plastic simply by sending
the ‘stl’ file to a different machine.
Consumer products are often complex assemblies. There is often a need to work around fixed,
conventionally produced components. This restriction is likely to diminish however, as it becomes
possible to build in a greater variety of materials and material combinations. Absolute dimensions
no longer have the relevance they once had. Materialise-MGX offer to produce their designs to
order in any reasonable size (http://materialise-mgx.com). Manual drafting has been replaced by
2D CAD drafting packages, such as AutoCad, both in industry and in design education. As 3D
CAD becomes the common methodology and with it the provision to output technical drawings
257
automatically, the need for any training in 2D drafting can be questioned. Is there even a need to
dimensions a drawing if dimensions can be taken directly off a 3D virtual model? Arguments
such as these are set to develop as digital technologies shift the emphasis onto abstract formula
underpinning a design rather than absolute solutions. There is a gradual shift in emphasis from
the physical datum, printed drawings,models and the like, to virtual computer-based master data.
8.3.2
Communication
Traditionally, product design development models begin with loose concept ideas that are
gradually refined as the project develops.
Practicalities are fed in gradually, so as not to
‘suffocate’ the style of the raw concept. Early sketches communicate the character of the design
before detailed technical work has begun.
This allows the design to be assessed before
significant investment is made.
Computer generated forms are hard to envisage and complexity, by its very nature, cannot be
easily reduced or predicted.
In computational design the initial priority is in setting up
relationships, parameters, and formula, rather that looking ahead to what they will ultimately
produce. Creating such rule-based systems takes time and requires something of a leap of faith
on the part of those commissioning the process (as was required in the commissioning of Holy
Ghost 7.3). It is not usually possible to generate outcomes before the system is established.
Once the rules are in place, however, any number of design iterations can be generated with
relative ease.
8.3.3
Authorship
Interactive selection by the consumer and computer generated form, raise philosophical
questions around definitions of terms such as 'design' and 'designer', challenging accepted
notions of authorship. There is debate as to whether the end results are ‘art’, ‘design’, ‘craft’, or
'computer generated'. FutureFactories is among a wave of emergent digital practices which are
forming a new position for practitioners, with alternative routes to manufacture and modes of
consumption. The potential impact of FutureFactories has been noted and described since the
first stage of this work (Atkinson and Dean 2003, Appendix 3) and are still considered significant.
As the project has developed, additional elements have been recognised with respect to the
system's impact on issues of authorship and accepted notions or definitions of design practice
(Atkinson and Hales 2004).
Where a consumer is interacting directly with generative software, the original ‘designer’ may not
even be aware of products selected and produced in his or her name. Computer control and
258
autonomous production potentially act to isolate the author of the work from the outcome, and
raise questions of responsibility and ownership. The potential for remote manufacture brings
isolation in itself. Digital sculptor, Keith Brown, commented that he had contributed work to
exhibitions in other continents without having seen the work in the flesh. Digital files had been
sent to remote exhibitions, where the pieces were digitally manufactured and exhibited, without
the creator seeing them. Freedom of Creation founder, Janne Kyttänen, spoke of designing a
trophy for an organisation in the USA. The award was manufactured and presented without the
designer seeing it. In these examples the practitioner had at least seen the virtual design, in an
automated production system this might not be the case. A careful setting of constraints and
considerable faith would be required for designs to be allowed to evolve and mutate unseen by
their creator.
The individualization concept of this project requires inputs from designer, software and
consumer. If the user, the consumer ordering a piece, makes an aesthetic judgment on a form,
the precise configuration of which has been generated by a piece of software; then who has
‘designed’ it? The author has deliberately limited the consumers’ input to the selection of one
variant over another. This is in contrast with the slider bar approaches of some systems (e.g.
Fluidforms, Section 1.5.3). Nevertheless; the consumer is able to express an aesthetic
preference. Is the selection of one variant over another a significant creative decision? Similar
issues were identified by Todd and Latham, with respect to virtual sculptures created via the
'Mutator' code: “Who owns the copyright? What is copyright? The generative system? The
genetic code for a final form? The computer form? The computer image? The artwork on a gallery
wall?” (Todd and Latham 1992). Limiting the ‘customer’ to a simple start, stop, and order
commands has proved untenable. In the trial computer interfaces created for the First Collection
exhibitions, 2003, the user was allowed to temporally save a series of selections rather than
choosing between ordering a single selection or restarting the animation. A on-screen folder
would display up to three selected iterations. The user could then choose which of the selected
iterations to ultimately print. Whilst this does not represent significant creative control, aesthetic
judgement is being exercised which cumulatively, over a population of users, could potentially
steer the design direction.
259
Figure 224
The Experimental User Interface
The potential effects on the practice of design are considerable. Whatever the future holds, there
are certain to be serious changes occurring. Obviously, the future role of the designer and where
he or she fits into the design process is one that will need to be examined closely and perhaps
readdressed.
8.3.4
The Geometry Comes Free ‘Myth’
Additive fabrication can do more than replicate the intricate moulded components of
contemporary consumer products. If Direct Manufacture is to be viable, it must take advantage of
inherent flexibility and freedoms to justify elevated costs. Digital manufacturing can go far beyond
the bounds of moulded plastics producing components that would be far too expensive to tool for
conventional manufacture. Multi-piece assemblies can be replaced by single components not by
merely ‘stitching together’ conventional parts, but by building the function of one component into
another. Components become multi-layered simultaneously addressing aesthetic, structural and
functional requirements.
The integration of features is common in design and engineering
practice; it is the extent to which this becomes possible that presents both the challenge and the
opportunity of digital manufacturing. Traditional product development roles become blurred as
individual components address a range of criteria and require input from a range of disciplines. In
the Rapid Prototyping industry, geometry is often said to ‘come free’ with parts of similar
dimensions costing the same regardless of complexity.
This is true in terms of manufacture;
exploiting this freedom however requires considerable investment in design. Instead of being
centred on tooling; production investment moves to the development of ever more complex virtual
models. Engineers and designers need new skills and understanding to manage complex multilayered projects, in order to fully exploit the potential of digital manufacture.
260
When a complex design is being created, visualisation in the virtual world becomes an issue. It is
difficult to orientate ones self in a complex on-screen model. It becomes hard to identify issues
that would be readily apparent when handling an actual prototype. A greater range of more
flexible visualisation tools are required; ideally built into the CAD software systems themselves.
For the trans-disciplinary potential of digital design and manufacturing technologies to be
realised; interface improvements must, however, go further than providing increased functionality
for the formally trained. Currently, “Digital tools and media change almost as rapidly as anyone
can master them”, and “New features keep appearing, interaction methods evolve, and
underlying contexts such as operating systems change” (McCullough 1996).
Much software
development centres on increased functionality, which only benefits full time operators, while
constant updates confuse the casual user. One software package or suite rarely offers all the
potential required. In this project regular use has been made of the following packages:



Alias StudioTools
SolidWorks

Rhino

AutoCad

3D StudioMax
Virtools
Each software package has a different interface, different terminology and different sequences of
commands for broadly similar operations.
8.3.5
Art, Craft and Design
The use of software processes and real-time networks as generative tools, questions existing,
transient, boundaries of practice, and exposes the irrelevance of conventional definitions of role.
It is clear that the outcomes of such a new model of creative production cannot be thought of as
traditionally conceived pieces. Existing definitions convey little of the reality of their production
and they lie in some new, as yet unspecified, arena of production.
Despite the highly complex issue of defining ‘design’ per se (Micklethwaite 2000), and without
wishing to enter an extended debate about the distinctions between craft and design; a simplistic
approach can be adopted. If the outcome is dependant on a particular individual making skills it
can be considered a craft. If the artefact could be equally well manufactured by a number of
appropriately skilled companies, then it is surely a piece of design.
The FutureFactories
production model is intended to be a design process in that sense. The designer creates a
system capable of generating endless design iterations. A specification is created thereafter
outputs are mechanistically generated. The practitioner is only required to create the initial metadesign rather than for day to day production
261
Pye differentiates the processes of making things into either “the workmanship of risk” or the
workmanship of certainty” (Pye 1968) and describes a spectrum of operations from “free-“ to
“regulated-“ (Pye cited by Gordon, Woodwork 1996).
The economics of conventional mass
manufacture depend upon standardisation and uniformity; the highly regulated workmanship of
certainty with no room for risk.
FutureFactories introduces or reintroduces in the industrial era,
risk in mass produced goods and in doing so blurs the boundaries between traditional notions of
craft and design. The ethos remains clearly within design however and the ambiguity indicative
of a need to separate design from traditional notions of mass manufacture.
The consumers input must also be considered. The fact that customers can make an aesthetic
decision to buy one variant over another does not seem significant when considered in isolation.
One can imagine that volumes of consumers making particular selections would steer the design
in the longer term, but sales pressure is also a factor in conventional design and manufacture.
The FutureFactories’ concept attempts to remain solidly within design, dismissing the creative
input of a user offered given little control.
There is again some blurring of boundaries and
distinctions; the preferences of consumers will be more readily felt than in conventional models of
consumption. Added to this there is the possibility, and therefore pressure, to react to demand
now that product life cycles are no longer fixed.
8.3.6
Design Skills and Education
As far as the impact on design education is concerned, every aspect of the curriculum may need
to be addressed.
Design for manufacture will be less concerned with specific production
processes and the generation of a specification; and more concerned with the creation of
producible entities by a digital manufacturing process and creative systems. The teaching of
materials and processes for manufacture premised on mass-production would, of course, also
have to be reduced or replaced with a higher level of emphasis given to digital manufacturing
techniques and computational design techniques.
The emphasis on achieving mass market production volumes will need to be re-balanced. Target
markets can become more focused and respond to immediate opportunities. Market research will
become more concerned with discovering untapped niches, rather than common denominators.
There will be a need, perhaps, to consider in far more depth the user needs of individuals: more
attention paid to personal preference and the celebration of diversity over convergence. If the
notion of brand ethos is to continue in this scenario, surely it will have to move further into the
realm of individual interpretation; rather than some manufactured and marketed ‘lifestyle’ or
‘brand value’ heterogeneity. Certainly there would at least be a requirement for learning more
about ‘people’ and less about ‘markets’: more about the choices people make about objects and
the emotional relationships people have with them.
262
8.4
Achieving Balance Between Order and Chaos
There has been a constant tension in the project between a desire to deliver unexpected drama
and a need for coherent identifiable designs. The need to balance the degree of random form
generation has been acutely felt. Completely random form could not develop ‘brand recognition’
and would be of dubious artistic merit. At the same time there has to be some relinquishing of
control practically, to achieve the stated aim of variance, and creatively, for the drama of the
unexpected. As Colson states in respect of generative art: “Computers offer access to data that
does not have to be sequential. It too can be random. Digital artists need to negotiate with
delicacy the boundary between logical methodologies that are part of the way they have to work
and the undefined artistic intentions, which sometimes require withdrawal of conscious control in
order to be really compelling. Their work is often tipped more one way than the other but when
evenly balanced, these two elements can result in work that is tense and exciting” (Colson 2007).
In the jewellery design Icon (8.3.3), it has been demonstrated that a coherent design identity can
be achieved over an extended run of 25 pieces; this has been the greatest number of iterations
produced from a FutureFactories design to-date. These 25 pieces remain distinctly unique and it
would seem likely that the full target edition of 100 pieces could be achieved without identity
clashes. The scope for this design is limited however. The form is made up of only a few
elements with restricted movement and there is one key feature that to a large extent defines the
character of the piece. If the production were extended, to 500 units, and beyond for instance, a
greater perception of similarity would be expected.
As the individualisation is randomly
generated, there is the potential for similarities even in a small batch. The concern, however, is
not a single instance of deja-vu, but in repeated similarity which would undermine the concept
and value of the system. Similar pairs from the first batch have been identified in Figure 225.
Iteration #4 has been mirrored to increase the similarity. Whist there is little similarity in the
overall forms the dominance of the ‘eye’ feature makes even slight similarity noticeable.
263
Figure 225
Similar Pairs in Icon Production
It is clear from the works created that some geometries carry more scope for individualisation
than others. The SuperKitch bangle is made up of approximately 30 distinct pieces taken from a
library of 50 pieces, arranged in a random string.
For an identical set of pieces to be selected in the same build order, the probability of selecting
the first piece would be 1 in 50, the second 1 in 49 and so on for the 30 pieces.
In a niche product, where a significant marketing emphasis is placed on individuality, the
production run should be limited according to the geometries’ potential for difference. Contour
changes in surfaces can be used to create difference, but in extended runs the distinction
between iterations may not always be obvious. In Tuber, for example (5.1.3), the individualisation
was driven by key-frame animation. Five iterations were produced from an animation clip of 3600
frames. The virtual model changes with each frame of the animation and this clip could therefore
have yielded 3600 discreet artefacts. Iterations generated from neighbouring frames, however,
would be perceived as identical without careful measurement and iterations even 10 -20 frames
apart would be hard to distinguish, Figure 226.
264
Figure 226
Tuber Animation Frames
The more elements that are subject to change, the more scope there is for generating difference.
This favours more detailed, chaotic forms over minimalist simplicity.
Models that can be
reconfigured rather than merely adjusted offer significantly more potential.
A change in
configuration is immediately apparent, whereas adjustments to a form only become noticeable
over certain amplitude, Figure 227.
Figure 227
Adjustment vs. Reconfiguration
265
Individualisation lends itself to forms in which there are arbitrary design decisions, for instance the
number of ‘buttons’ in the back of the Holy Ghost chair (3.5).
Increased complexity reduces the visual impact of details. In Tuber the interfaces between the
four volumes require careful filleting in a post production process. This involves trimming back
surfaces and adding a fillet of varying radius. In Tuber9 the increased visual complexity means
that less emphasis is placed on these intersections allowing them to be treated with a single
command constant radius fillet.
The Puja is a table lamp design (Section 7.6) intended as a follow-up to Cornuta (Section 7.2)
and was to be the second design created using Solidworks and driven by Genovate. Whilst the
design proved successful and has been widely exhibited, it did not prove possible to create
design variants to the researcher’s satisfaction. The design is sand-cast in sections which are
subsequently welded together and the joints polished out. The first prototypes of the design were
produced in India and even the variance introduced by the production process (as opposed to
deliberate individualisation), proved problematic. The combination of an arbitrary dressing of
edges and the positioning of parts by eye, prior to welding, introduced detrimental changes to the
design’s character. It is clear, therefore, that not all designs are suitable for individualisation and
that in some designs the aesthetic is too heavily bound up with a discreet set of proportions.
The SuperKitch jewellery design demonstrates that computer script can be combined with CAD to
generate a potentially infinite stream of discreet 3D outcomes. Superkitch typically generates a
new form in 4 -5 minutes depending on the processors speed. The script configures pre-existing
CAD models in a Constructive Solid Geometry approach. There is no post-generation CAD
modelling, other than to mirror the virtual model configuration from a library of CAD components,
a process that could easily be automated.
266
8.5
Recommendations for Further Research
From the early residency period of this study, opportunities to expand its’ scope have constantly
arisen. Inevitably the thesis had a specific focus and this leaves future research opportunities
that are worth highlighting.
8.5.1
CAD and Programming
As the project progressed the geometries created became increasing driven by programming.
The researchers’ prior experience however, has had a significant bearing.
The influence of
various approaches, programming techniques and software platforms would be worthy of further
study, for example:



NURBS geometry vs. Polygon mesh;
Writing code versus script writing packages such as Virtools; and
Dedicated generative design packages vs. custom written scripts.
Different approaches to computer programming have been experimented with in this study.
Successful 3D output was only achieved when the researcher was involved directly in the
programming. This was perhaps necessary with a novel concept and where there was little preexisting output to reference. With the example of this thesis and with a growing number of
creatives experimenting with generative techniques, future researchers can perhaps have greater
clarity regarding scripting requirements enabling a separation of roles.
The involvement of
specialist programmers would allow the designer to focus on their creative aims rather that the
limitations of particular software.
8.5.2
The Designs
Throughout the study the researcher has placed emphasis on the need for control and for
keeping the output true to the ‘designer’s intent’. The value of this should be explored and a
computer programmers raw interpretation of a design brief compared and contrasted with
programmer/designer collaboration.
In this thesis interactivity with the consumer has been deliberately restricted and individualisation
favoured over customisation.
This distinction is worthy of in-depth exploration and the
introduction of random input (as in this study) compared with co-design.
267
8.5.3
The Artefacts
The premise of this thesis was based on the availability of additive manufacturing. The high cost
of these technologies has tended towards expensive gallery outputs which inevitably influence
perceptions. To increase the relevance to commercial production, simple, low cost, high volume
outputs should be explored. This could be achieved through the use of emergent lower cost
additive technologies or through the use of lower cost digital manufacturing techniques e.g. laser
cutting.
The thesis has proved the possibility of generating significant numbers of variants from a metadesign template. It would be interesting to explore preference though such a collect of pieces
with the public offered samples to select from;
for example:



Would there be significant preferences for individual pieces?
Would there be significant preferences for recognisable types (8.4)?
Are there disliked ‘failures’?
268
Technical Glossary
3D printing, 3DP
Three Dimensional Printing (3DP): is at the budget end of the Rapid Prototyping spectrum. The
processes use inkjet printing technology and the equipment costs are around a tenth of those for
STL/SLS. Each layer is formed by a jet of material or binder from a multi-nozzle print head. 3DP
tends to be used for concept rather than functional models. The term 3DP has considerable
appeal; aptly summing up the technology and its context, the use of the term however has been
limited by its becoming a trade mark.
Boolean Operations
Boolean Operations in the CAD (Computer Aided Design)
context are a set of logical (Boolean logic) operations that can be
applied to 3D closed volumes. Intersecting entities can:i)
ii)
iii)
Be united into a single part with a Union operation;
Have the intersecting portion of one entity removed
from the other, in a subtraction or difference
operation;
Have everything removed apart from the intersecting
portions in an intersection operation.
C++
C++ is a general purpose computer programming language that is widely used throughout
software development. It is generally regarded as a middle level language combining basic
functionality with some advanced features.
Control Curve
Control Curves are a feature of CAD geometry in which surfaces
are defined by 2D or 3D profiles.
Construction History: Please refer to parametric design.
Constructive Solid Geometry, CSG
Constructive solid geometry is a technique used in Solid Modelling. It allows the creation of
complex models from a simple primative. The primatives are either assembled or used to cut
away ‘material’ from each other using Boolean Operations.
One of the advantages of CSG is that is easy to ensure that a water-tight, closed, volume is
achieved; an important factor in Digital manufacturing.
Computational Design
Computational Design is the combination of Computer Aided Design with computer programming.
Computer scripts can be embedded in CAD software, such that aspects of the geometry are
computer generated.
269
Computer Aided Design, CAD
CAD is the generic term for computer-based geometry authoring tools that are used throughout
design and engineering industries. CAD initially stood for ‘Computer Aided Drafting’ and was
seen as a direct replacement for the manual drawing board. Software packages soon developed
capabilities beyond traditional drafting; hence the switch in terminology to Computer Aided
Design. There are different levels of CAD software: 2D drafting systems (e.g. AutoCAD,
MicroStation); mid-range 3D modellers (e.g. SolidWorks, Rhino); and high-end 3D hybrid
systems (e.g., Alias, Pro/ENGINEER, CATIA). The boundaries between these categories is fluid;
with the 2D packages including some basic 3D capability and the mid-range packages increasing
their functionality. Within the mid and high end packages there are Parametric and nonparametric packages, Solid Modellers and Surface Modellers.
Computer Generated Design, CGD
Computer Generated Design is geometry that has been created or modified by computer
algorithms.
CV, Control Vertex
See NURBS Geometry
Cybersculpture
This represents computer-based 3D geometry, or virtual ‘structures’, that may be impossible to
realise with the physical constraints of the real world.
Delphi Programming Language
Delphi is a programming language, first released in the mid-nineties, that runs under Microsoft
Windows. It uses visual programming tools that make programming for Windows easier.
Direct Metal Laser Sintering (DMLS)
Selective Laser Sintering applied to metals; please refer to SLS.
Edit Point (curve)
Edit Point is a computational geometry curve defined by points on that curve.
Genotype
A biological term used to refer to the coded representation of a possible structure. In biology the
genotype is composed of DNA and contains information that will determine the development of an
organism (the phenotype).
Java3D
This is an application programming interface, API, for drawing 3D graphics using the Java
language.
Key-frame Animation
In key-frame animation an entity is created along with a series of developmental states for that
entity over time. Software then extrapolates between these stages to create a seamless
animation. A key-frame animation is of fixed length and yields only as many solutions as there
are key frames.
Lofting
Lofting is a 3D CAD modelling operation that creates of a 3D transitional surface (or solid volume)
between usually 2D profiles and is a common feature in high end CAD programs. Also known as
a Skin command (Alias Studio Tools).
270
NURBS Geometry
NURBS (Non Uniform Rational Basis Spline) geometry is
controlled by ‘weighting’ Control Vertices (CVs) that lie off the
curve or surface. NURBS geometry was developed to enable
engineers to reproduce accurately curvature for naval and
aerospace application. NURBS geometry allows representation
of controlled geometrical shapes in a mathematically compact
form that can be scaled indefinitely. NURBS’ curves and
surfaces are highly intuitive and predictable. CVs are connected
directly to the curve/surface, or act as if they were connected by
an elastic link. Manipulating them changes the character of the
curve or surface.
A NURBS Curve
Phenotype
A biological term for the physical solution, in biological systems: an organism that is derived from
a particular genotype.
Parametric Design
Convectional CAD defines objects with fixed dimensions. Typically, artefacts’ geometry will be
defined in Cartesian space via a series of vectors. Parametric CAD defines the nature of the
design via relationships between degrees of freedom rather than attributing absolute dimensional
values to the degrees of freedom themselves. For example, if a feature is positioned two thirds
of the way along a face; it is not important how long that face is, this can be modified at will: the
importance is that the two thirds proportion is maintained. If a particular attribute of a design is
modified, the whole model will update to maintain a set of relationships which are pre-defined
within the CAD file.
In parametric CAD, the sequences of geometric operations underlying a design are recorded in
what is often referred to as a Construction History. The user is able to go back through this
history of parent/child features and to adjusted elements on which later (in the history) operations
depend.
Polygon Based or Mesh CAD Geometry
A polygon mesh or unstructured grid is a collection of vertices, edges and faces that defines the
shape of a polyhedral object in 3D computer graphics and CAD modeling applications. Polygon
based models are used in web, animation and visualization applications where ease of
manipulation outweighs mathematical purity. Unlike NURBS geometry, which can be scaled
indefinitely, mesh surface have a fixed resolution.
Procedural Animation
In a procedural animation entities are modified by a procedure or algorithm. A set of
developmental rules and relationships are set out along with an initial condition. Solutions are
then generated automatically. A procedural animation by contrast can yield a potentially infinite
series of solutions.
Generative Design
Generative design can be defined as the approach of developing applications, or systems which
can develop, evolve, or design structures, objects, or spaces more or less autonomously
depending on the circumstance (Krause 2003).
Rapid Manufacturing (RM) (Direct Manufacture, Digital Manufacturing)
“The use of a CAD based automated additive manufacturing process to construct parts that are
used directly as finished products or components” (Hopkinson 2006). Rapid prototyping is now
271
well established; the new frontier is direct manufacture. Direct or Rapid Manufacture is
essentially the adaptation of RP technologies to the manufacture of end-use products. There are
several niche markets emerging: the mass customization of medical products, low volume-high
unit cost applications, such as aerospace and F1.
Rapid Prototyping, RP
Rapid Prototyping (RP) is a catch-all term that applies to the automated manufacture of
prototypes directly from digital CAD data. Simplistically speaking, RP allows the designer to ‘print
out’ on-screen models in 3D. Most of the recently developed processes are layer additive.
Software ‘slices’ the CAD model into thin layers (down to 0.05mm). The model then ‘grows’ one
thin layer at a time, as each data ‘slice’ is replicated in 3D from the bottom up. The layers are
built on a moving platform, each built on its predecessor as the platform steps down in layer
thicknesses. There is no tooling or cutting away of material. This allows unlimited geometry.
Forms may be produced that would be almost impossible to mould or machine. RP processes
can be additive or subtractive (subtractive where material is cut away from a solid blank).
However, the term is more commonly applied to additive processes.
Robosculpture (1988-Christian Lavigne)
Computer aided sculpture, manufactured automatically, via rapid-prototyping technologies.
Selective Laser Sintering, SLS
Laser Sintering is powder based. Here the layers are ‘drawn’ on fusible powder which is sintered
by laser, the powder being ordinarily maintained just below its melting point. The finished part is
formed within a cake of un-fused powder, which supports the features: the final step is to remove
this loose powder. SLS components are much more than appearance models, they have
sufficient durability to be considered functional. A wide range of powder based materials are
available to the process including nylon, stainless steel, synthetic rubber and ceramics.
Skin Surface
A Skin Surface is a 3D modelling operation within Alias CAD software that creates of a
transitional surface between curve profiles. It is a common feature in high-end CAD programs
and is alternatively known as a Lofting operation.
Solid Modelling (CAD)
Solid Modelling is a CAD technique for representing and manipulating solid objects. In general
terms, features are defined rather than surfaces: for example, block with a hole in it is defined by
a block plus a hole. Due to fact that forms are defined by operations performed on them, solid
modellers are best at ‘regular’ forms produced by commands that mirror the physical world, for
example, lathe and extrude.
SLM
SLM is a metal RP process similar to SLS/DMLS developed by MCP. As the name suggests, the
material is melted rather than sintered with claimed structural benefits.
Spline Curve
A tool commonly used in computational geometry for generating smooth curves that can be
scaled indefinitely: please refer to NURBS geometry.
Stereo Lithography, SLA (Stereo Lithography Apparatus)
Stereolithography was the first of the modern layer build systems. It was developed by 3D
Systems in the eighties. It accounts for 45% of new machine installations (Wohler’s Report
2003). In this process each layer is ‘drawn’ by laser in a photosensitive liquid polymer; the laser
light solidifies the liquid. The advantage of this process is the surface finish, which is good; the
disadvantage is that delicate features and overhangs require thin supports that must be removed
post-production.
272
STL , Standard Triangulation Language, file format
An STL file is a triangular representation of a 3D surface geometry native to additive fabrication,
that was created by 3D Systems. An STL file describes a raw unstructured triangulated surface
by defining the vertices of each triangular face. STL files describe only the surface geometry of a
three dimensional object without any representation of colour, texture or material attributes.
Surface Modelling (CAD)
Surface modelling is a CAD technique, in which a form is constructed from a number of discrete
surfaces. Usually the NURB’s surfaces are constructed from curves.
Telesculpture (Christian Lavigne, Alexandre Vitkine1994)
The remote production of digitally created sculpture, using RP technologies, via the electronic
transfer of data (Networks, Internet, ISDN).
273
274
Bibliography
ADAMSON, G., 2007. Thinking Through Craft, New York: Berg Publishers.
AISH, R., 2004 Smart Geometry Conference January 2004, Bentley Systems, London, United
Kingdom
AKS, D. & SPROTT, J., 1996. Quantifying Aesthetic Preference for Chaotic Patterns in Empirical
Studies of the Arts, 14, pp.1-16.
ALDERSEY-WILLIAMS, H., 2003. Zoomorphic. New Animal Architecture. London: Laurence
King.
ALEXANDER, A., 2006. Cyberpunk pioneer has designs on a better world: Bruce Sterling
interview, The Guardian. June 1. Available on line at:
http://technology.guardian.co.uk/weekly/story/0,1786641,00.html [Accessed September 2007].
ALFONDY, S., 2008. NeoCraft: Modernity and the Crafts (Ed),
Halifax, Nova Scotia: The Press of the Nova Scotia College of Art and Design.
ARAD, R., 2000. Not made by hand not made in China, bouncing vases collection,
http://www.ronarad.com/archive2000.swf, [accessed January 2007].
ARANDA, B. & LASCH, C., 2006. Tooling. New York: Princeton Architectural Press.
ARATA, L., 1999. Reflections about Interactivity. Available online at: http://web.mit.edu/commforum/papers/arata.html [Accessed August 2009].
ASCOTT, R., 2002. Turning on Technology. Available online at:
http://www.cooper.edu/art/techno/essays/ascott.html [Accessed June 2006].
ATKINSON, P., 2003 Future Factories: Design work by Lionel Theodore Dean, University of
Huddersfield, United Kingdom
ATKINSON, P. and DEAN, L., 2003. Teaching Techné, 5th European Academy of Design
Conference, EAD, 2003, Barcelona, Spain
ATKINSON P., 2004. Post Industrial Manufacturing Systems: The Impact of Emerging
Technologies on Design, Craft and Engineering Processes', Challenging Craft, Conference, 810th September 2004, Gray's School of Art, Aberdeen, UK.
ATKINSON, P. & HALES, D., 2004. A Chance Would be a Fine Thing: Digitally Driven PracticeBased Research at Huddersfield. Pixelraiders2 Conference Sheffield.
BAILEY, S., 2006. Chair Wars. In the Observer, 10th September 2006,. Available online at:
http://www.guardian.co.uk/artanddesign/2006/sep/10/architecture.shopping [Accessed June
2008]
BAR, M. & NETA, M., 2006. Humans Prefer Curved Visual Objects in Psychological Science, 17,
pp. 645-648.
BENJAMIN, W., 1936. The Work of Art in the Age of Mechanical Reproduction. Available online
at: http://www.marxists.org/reference/subject/philosophy/works/ge/benjamin.htm [Accessed
August 2009].
275
BENTLEY, P., 1999. Evolutionary design by computers. San Francisco: Morgan Kaufmann
Publishers.
BENTLEY, P. & CORNE, D, 2001. Creative evolutionary systems. San Francisco: Morgan
Kaufmann Publishers.
BLEMINK, B., 2004. Design = Art: functional objects from Donald Judd to Rachel Whiteread.
London: Merrell Publishers.
BLUNDELL JONES, P., 1999. Hugo Häring. The Organic versus the Geometric.
Stuttgart/London: Edition Axel Menges.
BOURRIAUD, N., 2002. Relational Aesthetics. Dijon: Les Presses du Réel.
BOVILL, C., 1996. Fractal Geometry in Architecture and Design. Basel: Birkhaüser.
BRITTAIN, M., 2002, Meeting Mr Laser, Jewellery in Britain, British Jeweller's Association, Issue
12, Nov. p. 2.
BROWN, K., 2008. Perimeters, Boundaries and Borders. Manchester, United Kingdom: Fast-UK
/Manchester Metropolitan University.
BROWN, P., 2001. Breaking the Art and Science Standoff in Leonardo. Vol. 34, No. 4, pp. 335336.
BUNNELL, K., 2004. Craft and Digital Technology. Available online at:
http://www.autonomatic.org.uk/team/kb/craft%20and%20digital%20technology.pdf [Accessed
April 2010].
CALLICOTT, N., 2001. Computer-Aided Manufacture in Architecture the Pursuit of Novelty.
Oxford: Architectural Press.
CAMPBELL, E., 2006. Personal Touch in Crafts May/June 2006.
CANDY, L., & EDMONDS, E., 2002. Interaction in Art and Technology.
CARDOSO, R., 2004. Putting the magic back into design: from object fetishism to product
semantics and beyond. Available online at:
http://www.bristol.ac.uk/artontheline/journal_20041/articles/pdf/20041_02.pdf [Accessed April
2010]
CENTURY, M., 1999. Pathways to Innovation in Digital Culture. Montreal: McGill University.
CHAPMAN, J., 2005. Emotionally Durable Design: Objects, Experiences and Empathy. London:
Earthscan Publications Ltd.
CHUA, C., LEONG, K., & LIM, C., 2003. Rapid Prototyping: Principles and Applications.
Singapore: World Scientific.
COOK, D., 1996. Digital Darwin. Creative Technology. February pp. 14-19.
COLES, A., 2005a. DesignArt. London: Tate Publishing.
COLES, A., 2005b. On Art’s Romance with Design. Design Issue 21[3] p17-24.
COLES, A, 2007. Design and Art. Cambridge, Massachusetts: The MIT Press.
276
COMPUTER ARTWORKS, 2003. Computer Artworks. Available online at:
http://www.artworks.co.uk/index2.htm [Accessed 27 January 2003]
COOPER, G., 2001. Rapid Prototyping Technology: Selection and Application. New York: Marcel
Dekker, inc.
COOPER-HEWITT, 2004. National Design Museum. Design ≠ Art. Available online at:
http://www.cooperhewitt.org/EXHIBITIONS/dina/site/webpages/dina.html [Accessed June 2008].
COSS, R., 2003. The Role of Evolved Perceptual Biases in Art and Design. In VOLAND, E. &
GRAMMER, K. (Eds.), Evolutionary Aesthetics. Berlin/ Heidelberg: Springer-Verlag, pp. 69-130.
CRAYTON, T., 2001. The Design Implications of Mass Customisation in Architectural Design. In
Architectural Design, April 2001 pp. 74-81
CRUZ, M. & PIKE, S. (Eds.), 2008. Neoplasmatic Design. London: John Wiley & Sons.
DAUERER, V., 2007. Interview with Digitalability curator Atilano González-Pérez. Available online
at: http://pingmag.jp/2007/05/11/digitalability/ [Accessed July, 2008].
DAVIES, S., 1987. Future Perfect. Boston: Addison-Wesley Longman Publishing Co.
DAWKINS, R., 1986. The Blind Watchmaker. London: Longman.
DAY, M., 2005. Generative Components. In AEC Magazine 18th January 2005
DEAN, L., ATKINSON, P. & UNVER, E., 2005 (a). FutureFactories: Inverting the MassProduction Paradigm. Available online at: http://www.falmouth.ac.uk/img/pdf/ooooo452.pdf
[Accessed March 2006].
DEAN, L; ATKINSON, P. and UNVER, E., 2005 (b). Evolving individualised consumer products,
European Academy of Design EAD, 2005, Bremen, Germany
DESIGNMAI, 2007. Digitability press release, Designmai, Berlin, May 2007.
DEVERAUZ, J., 2002. Mass Customisation: Let Consumers Collaborate on Product Designs.
Metropolis Magazine August/September 2002. Available online at:
http://www.Metropolismag.com/html/content_0802/cus/ [Accessed July 2008].
DINKLA, S., 1994. The History of the Interface in Interactive Art. Available online
at:http://www.kenfeingold.com/dinkla_history.html [Accessed August 2009].
DORMER, P., 1997. The Salon de Refuse? In Dormer, P. (Ed.) The Culture of Craft, 7.
Manchester: Manchester University Press.
EOS ,2009, Available online at http://www.eos.info/en/home, [Accessed September 2009].
[email protected], 2007. [email protected]: The Open-Source Personal Fabricator Project. Available online
at: http://www.fabathome.org/ [Accessed October 2007].
FABJECTORY, 2007. Fabjectory: Virtual Objects in Real Life. Available online at:
http://www.fabjectory.com/ [Accessed October 2007].
277
FAIRS, M., 2004. What is Design? In Icon magazine Issue 018 2004. Available online at:
http://www.iconeye.com/index.php?view=article&catid=328%3Aicon+018&layout=default&id=267
4%3Awhat-is-design--icon-018--december-2004&option=com_content [Accessed October 2009]
FEUERSTEIN, G., 2002. Biomorphic Architecture. Human and Animal Forms in Architecture.
Stuttgart/London: Edition Axel Menges.
FLAKE, G.W., 2000, The Computational Beauty of Nature, Cambridge, Massachusetts: The MIT
Press.
FLUIDFORMS, 2005. Fluid Forms, Your Unique Designs. Available online at http://fluidforms.com [Accessed Dec 2007].
FLUSSER, V., 1999. The Shape of Things – A Philosophy of Design. London: Reaction Books
Ltd.
FRANCIS, E., TAY et al., 2001. Distributed Rapid Prototyping – A Framework for Internet
Prototyping and Manufacturing. In Integrated Manufacturing Systems 12(6) pp. 409-415.
FRAZER, J., 1995. An Evolutionary Architecture. London: Architectural Association Publications.
FRAZER, J & JANSSEN, P., 2003. Generative and Evolutionary Models for Design. In
Communication & Cognition, 36, pp. 187-215.
FU, P., 2002. Custom Manufacturing: From Engineering Evolution to Manufacturing a Revolution.
In Time Compression Technologies. April 2002 10/2 pp. 44-48.
FULLER, M., 2003. Behind the Blip: Essays on the Culture of Software. New York: Autonomedia.
FUNES, P; and POLLACK, J., 1997. Computer Evolution of Buildable Objects, Fourth European
Conference on Artificial Life, Cambridge, MA:MIT Press. pp.358-367
GANIS, W., 2004. Digital Sculpture: Ars ex Machina. In Sculpture Magazine. September, Volume
23, No. 7.
GERE, C., 2006. Art, Time and Technology. Oxford: Berg Publishers.
GERSHENFELD, N., 2005. Fab: The Coming Revolution on Your Desktop – From Personal
Computers to Personal Fabrication. New York: Basic Books.
GIBART, T., 2002. Objecthood. Available online at:
http://csmt.uchicago.edu/glossary2004/objecthood.htm [Accessed June 2008]
GOHLKE, G., (Ed.) 2003. Software Art - A Reportage about Source Code. Berlin: The Media Arts
Lab.
GREENE, R., 2004. Internet Art. New York: Thames and Hudson.
GRIMM, T., 2003. Rapid Prototyping Benchmark: 3D Printers. Irvine, California: T.A. Grimm &
Associates.
GRIMM, T., 2004. Users’ Guide to Rapid Prototyping. Michigan: Society of Manufacturing
Engineers.
GROSSMAN, B., 2005. Available online at: http://www.bathsheba.com/ [Accessed October 2007]
278
GULLICHSEN, E. & CHANG, E., 1985. Generative Design in Architecture Using an Expert
System. In the Visual Computer. Berlin/ Heidelberg: Springer-Verlag, pp.161-168.
KAMRAI, A., & ABOUEL NASR, E., 2006. Rapid Prototyping: Theory and Practice. New York:
Springer Science+Business Media Inc.
HARROD, T., 2002. Otherwise Unobtainable: the Applied Arts and the Politics and Poetics of
Digital Technology, Pixel Raiders Conference, Victoria & Albert Museum, 19th March 2002.
HEATH, T., SMITH, S. & LIM, B., 2000, Tall Buildings and the Urban Skyline. The Effect of Visual
Complexity on Preferences. In Environment and Behaviour, 32, pp. 541-556.
HEERWAGEN, J., 2003. Bio Inspired Design: What Can We Learn From Nature? Available
online at: http://biomimicry.typepad.com/bioinspire/files/BioInspire.1-01.15.03.pdf [Accessed Sept
2009].
HENSEL, M. MENGES, A. & WEINSTOCK, M. (Eds.), 2004. Emergence: Morphogenetic Design
Strategies. London: Wiley-Academy.
HENSEL, M., MENGES, A. & WEINSTOCK, M. (Eds.), 2006. Techniques and Technologies in
Morphogenetic Design. London: Wiley-Academy.
HENSEL, M. & MENGES, A. (Eds.), 2008. Versatility and Vicissitude: Performance in Morpho
Ecological Design. London: Wiley-Academy.
HESSELGREN, L. ,2004. Smart Geometry Conference, January 2004, Bentley Systems, London,
United Kingdom
HIGHT, C. & PERRY, C. (Eds.), 2006. Collective Intelligence in Design: New Forms of Distributed
Practice and Design. London: John Wiley & Sons.
HODGES, N. (Ed.), 1994. Art and Technology. London: Art & Design.
HODGSON, E., 1998. The CALM (Creating Art with Layer Manufacture) Project. Available online
at http://www.uclan.ac.uk/clt/calm/overview.htm [Accessed June 2006]
HOLTZMAN, S., 1994. Digital Mantras: The Languages of Abstract and Virtual Worlds.
Cambridege: MIT Press.
HOPKINSON, N., HAGUE, R. & DICKENS, P. (Eds.) 2005. Rapid Manufacturing: An Industrial
Revolution for a Digital Age. Chichester: John Wiley and Sons Ltd.
HUGHES, R., 1991. The Shock of the New: Art and the Century of Change. London: Thames and
Hudson.
HURWICZ, M., 2000. It's Ready When You Are - Web Virtual Reality and 3D - in VRML or XML?
In Web Developers Journal, June 21, 2000
HUYSSEN, A., 1986. After the Great Divide: Modernism, Mass Culture, Postmodernism.
Bloomington: Indiana University Press.
IMAMOGLU, A., 2000. Complexity, Liking and Familiarity: Architecture and Non Architecture
Turkish Students’ assessments of Traditional and Modern House Facades. In Journal of
Environmental Psychology, 20, pp. 5-16.
279
INTERSCULPT, 2007. Available online at:- http://www.arsmathematica.org/sculptbio/sculptbioexpo.html [Accessed August 2009].
JACKSON, L., 2004. Craft Wars in ICON issue 016 October 2004. Available online at:
http://www.iconeye.com/index.php?option=com_content&view=article&id=2691:craft-wars--icon016--october-2004 [Accessed October 2009].
JACKSON, T., 2001. Towards a New Media Aesthetic, Reading Digital Culture. (Ed). David
Trend. Oxford, U.K.: Blackwell Press, pp. 347-353.
JOHANSSON, T., 2006. Art in the Context of Design, Design in the Context of Art. Working
Papers in Art and Design 4. Available online at:
http://www.herts.ac.uk/artdes1/research/papers/wpades/vol4/tdjabs.html [Accessed September
2009].
JOYE, Y., 2007. A Tentative Argument for the Inclusion of Nature-Based Forms in Architecture.
University of Gent.
KAPLAN, A. & HAELEIN, M., 2006. Toward a parsimonious definition of traditional and electronic
mass customization, Journal of product innovation management.
KOLAREVIC, B. & MALKAWI, A. (Eds.), 2004. Performative Architecture: Beyond instrumentality
New York: Routledge.
KOLAREVIC, B., 2005. Architecture in the Digital Age: Design and Manufacturing. London: Taylor
& Francis Ltd.
KOLAREVIC, B. (Ed.), 2008. Manufacturing Material Effects: Rethinking Design and Making in
Architecture. New York: Routledge.
KRAUSE, J., 2003. The creative process of generative design. In architecture pg. 1
KRISH, S., 2006. Genometri. Available online http:// www.genometri.com [Accessed November
2006]
LEACH, N. (Ed.), 2002. Designing for a Digital World. London: John Wiley & Sons.
LEACH, N., TURNBULL, D. & WILLIAMS, C. (Eds.), 2004. Digital Tectonics. London: WileyAcademy.
LERICHE, E., 2007. Trans-Form press release, Maison et Objet, Paris, France, January 2007
LÜ, L., FUH, J. & WONG, Y-S., 2001. Laser-Induced Materials and Processes for Rapid
Prototyping. Norwell, Massachusetts: Kluwer Academic Publishers.
LYNN, G., 1998. Folds, Bodies and Blobs: Collected Essays. Brussels: La Lettre Volée.
LYNN, G. & Rashid, H., 2003. Architectural Laboratories. Rotterdam: Netherlands Architecture
Institute.
LYNN, G., 2008. Greg Lynn Form. New York: Rizzoli International Publications.
MACCGREGOR, B., 2002. Cybernetic Serendipity Revisited. Proceedings of the Fourth
Conference on Creativity & Cognition. New York: ACM Press.
280
MAEDA, J., 2006. The Laws of Simplicity. Cambridge, Massachusetts: The MIT Press.
MAGNUSSON, T., 2002. Processor Art: Currents in the Process Oriented works of Generative
and Software Art. Available online at: http://www.ixi-software.net/thor/pa_lowres.pdf
MALINS, J., PENGELLY, J. & MARSHALL, J., 2007. The Post Disciplinary Digital Practitioner. In
BEHEMIA, E., HILTON, K., MCMAHON, C & CLARKE, A., (Eds.) Shaping the Future? Basildon:
Hadleys Ltd. pp. 437-441.
MALONE, E., 2008. Factories at Home, Factory of the Future, Euromold 2008, Frankfurt,
Germany
MARSH, P., 1997. Making to Measure. In Design, Spring 1997 pp 32-37
MARSHALL, J., 2007. An Exploration of Hybrid Art and Design Practice Using Computer-Based
Design and Fabrication Tools. Aberdeen: The Robert Gordon University
MARSHALL, J. & PENGELLY, J., 2006. Computer Technologies and Transdisciplinary
Discourse: Critical Drivers for Hybrid Design Practice? CoDesign, Vol.2 No.2. Taylor and Frances
Ltd. pp. 109-122.
MASTERTON, D., 2004. The Hunt for Complexity. Available online at:
http://www.autonomatic.org.uk/archive/The%20Hunt%20for%20Complexity.pdf [Accessed April
2010].
MATERIALISE.MGX, 2009. Available online at: http://www.materialise-mgx.com, [Accessed
September 2009]
MCCULLOUGH, M., 1996. Abstracting Craft: The Practiced Digital Hand. Cambridge,
Massachusetts: The MIT Press.
MCCULLOUGH, M., 2004. Digital Ground. Cambridge, Massachusetts: The MIT Press.
MCDONALD, J., RYALL, C. & WIMPENNY, D., (Eds.), 2001. Rapid Prototyping Casebook.
London: Professional Engineering.
MICHL, J., 2004. Form Follows What? The Modernist Notion of Function as a Carte Blanche.
Available online at: http://janmichl.com/eng.fff-hai.html [Accessed April 2010].
MICKLETHWAITE, P., 2000. Conceptions of Design in the community of Design. PhD Thesis,
University of Huddersfield.
MILLER, D. (Ed.), 1998. Material Cultures: Why Some Things Matter. London: UCL Press.
MILLER, J. & VAN LOON, B., 1992. Darwin for Beginners. Cambridge, UK: Icon Books.
MITCHELL, I. & McGRAVIE, D., 2004. Crafting the 3D Object, Challenging Craft, Conference, 810th September 2004, Gray's School of Art, Aberdeen, UK. Available on-line at:http://www.rgu.ac.uk/challengingcraft/ChallengingCraft/pdfs/mitchellmcgravie.pdf [accessed
October 2006].
MITCHELL, M., 1996. An Introduction to Genetic Algorithms. Cambridge: MIT Press.
MITCHELL, M., TSENG, J. & CHUAN-JUN, S., 1998. Virtual Prototyping for Customized Product
Development. In the Integrated Manufacturing Systems Journal, Vol. 9, Issue 6. Bradford, UK:
MCB UP Ltd.
281
MOUSSAVI, F. & KUBO, M. (Eds.), 2006. The Function of Ornament. Barcelona: Actar.
NIEDDERER, K., 2004. Why is there the need for explanation? – Objects and their realities, In
Working Papers in Art and Design, Vol.3, Available on-line at:http://sitem.herts.ac.uk/artdes_research/papers/wpades/vol3/knfull.html [Accessed January
2007].
NORMAN, D., 1988. The Design of Everyday Things. New York: Doubleday.
NORMAN, D., 2004. Emotional Design: Why We Love (or Hate) Everyday Things. New York:
Basic Books.
OBJECTS OF VIRTUAL DESIRE, 2005. Objects of Virtual Desire. Available online at:
http://www.objectsofvirtualdesire.com/ [Accessed July 2008].
PAPANEK, V., 1999. Design for the Real World: Human Ecology and Social Change. Chicago:
Academy Publishers Ltd.
PASQUARELLI, S., (Ed.) 2002. Versioning: Evolutionary Techniques in Architecture. London:
Wiley-Academy.
PAUL, C., 1999. Fluid Borders: The Aesthetic Evolution of Digital Sculpture. Available online at:
http://www.sculpture.org/documents/webspec/digscul/digscul.shtml [Accessed April 2008].
PAUL, C., 2003. Digital Art. London: Thames and Hudson Ltd.
PETROSKI, H., 1994. The Evolution of Useful Things. Vintage Books, Random House: New
York.
PINE, B., 1993. Mass Customisation: The New Frontier in Business Competition. Boston:
Harvard Business School Press.
PONOKO, 2007. Ponoko – make it real. Available online at: http://ponoko.com/ [Accessed July
2008].
POPPER, F., 1993. Art of the Electronic Age. London: Thames and Hudson.
POYNOR, R., 2005. Art’s Little Brother, in Icon issue 023, May 2005. Available online at:
http://www.iconeye.com/index.php?view=article&catid=323%3Aicon+023&layout=default&id=262
8%3Aarts-little-brother--icon-023--may-2005&option=com_content [Accessed September 2009].
PYE, D., 1978. The Nature and Aesthetics of Design. London: Herbert Press.
PYE, D., 1995. The Nature and Art of Workmanship. London: Herbert Press.
RAHIM, A., 2005 Catalytic Formations. Abingdon: Taylor and Francis.
RAHIM, A. & JAMILLE, H. (Eds.), 2007. Elegance. London: John Wiley & Sons.
REICHARDT, J., 2005. Cybernetic Serendipity. Available online at:
http://www.medienkunstnetz.de/exhibitions/serendipity/ [Accessed June 2008].
RENDELL, J., 2006. Art and Architecture: A Place Between. London: I.B. Taurus & Co. Ltd.
REPRAP, 2007. RepRap. Available online at: http://reprap.org [Accessed June 2008].
282
RIDDELL, A., 2001. Data Culture Generation: After Content, Process as Aesthetic in Leonardo,
Vol. 34, No. 4, pp. 337-343.
RISATTI, H., 2007. A Theory of Craft: Function and Aesthetic, Chapel Hill, North Carolina: The
University of North Carolina Press.
RISCOE, C., HOWARD, G., SEKERS, A. & VINER, D., 1988. Art & Computers. Middlesborough:
Cleveland Gallery.
RHOADES, L., 2005. The Transformation of Manufacturing in the 21st Century. The Bridge. Vol
35, No.1, Spring, national Academy of Engineering Publications. Available online at:
http://www.nae.edu/Publications/TheBridge/Archives/V351CelebratingManufacturingTechnology/TheTransformationofManufacturinginthe21stCentury.aspx
[Accessed September 2009].
RUTSKY, R., 1999. High Techné: Art and Technology, from the Machine Aesthetic to the
Posthuman. Minneapolis: University of Minnesota Press.
SAKAMOTO, T., FERRE, A. & KUBO, M., (Eds.), 2008. From Control to Design:
Parametric/Algorithmic Architecture. Barcelona: Actar.
SAN FRANCISCO MUSEUM OF MODERN ART, 2001. 010101: Art in Technological Times.
Available online at: http://www.sfmoma.org/exhibitions/2 [Accessed September 2009].
SKOV HOLT, S. & HOLT SKOV, M., 2005. Blobjects and Beyond: The New Fluidity in Design.
San Francisco: Chronicle Books.
SEVALDSON, B., 2005. Developing Digital Design Techniques: Investigations on Creative
Design Computing. PhD Thesis. Oslo School of Architecture and Design, Oslo, Norway.
SHACKLETON, J., & SUGIYAMA, K., 1999. Prototype Theory and the Modeling of New Product
Perception, in Proceedings of the Conference of the Design Research Society, University of
Central England, 8-10 September 1998, pp. 155-166. UK: Taylor and Francis.
SHEIL, B. (Ed.), 2005. Design through Making. London: John Wiley & Sons.
SHEIL, B. (Ed.), 2008. Proto Architecture: Analogue and Digital Hybrids. London: John Wiley &
Sons.
SILVER, M. (Ed.), 2006. Programming Cultures: Architecture, Art and Science in the Age of
Software Development. London: Wiley-Academy.
SIMS, K., 1999. Evolving Three-Dimensional Morphology and Behaviour. In BENTLEY (Ed)
Evolutionary Design by Computers. Morgan Kaufmann Publishers, Inc.
SODDU, C; & COLABELLA, E., 1997. Argenic Design. The European Academy of Design
Contextual Design / Design in Context Conference, Stockholm 23-25 April 1997, pg.1
SPILLER, N., 2008. Digital Architecture Now: A Global Study of Emerging Talent. London/ New
York: Thames and Hudson.
SPUYBROEK, L., 2004. NOX: Machining Architecture. London: Thames & Hudson Ltd.
STARK NETWORK, 2007. Available online at: www.starck.com/ [Accessed October 2007].
283
STEADMAN, P., 1979. The Evolution of Designs: Biological Analogy in Architecture and the
Applied Arts. Cambridge University Press.
STERLING, B., 2000. Blobjects and Biodesign. Artbyte Magazine. March–April Issue. pp. 42-53
and pp. 90-91.
STERLING, B., 2004a. The Dream Factory: Any Product, Any Shape, Any Size – Manufacture on
Your Desktop! Wired Issue 12.12 December 2004. Available online at:
http://www.wired.com/wired/archive/12.12/view.html?pg=4 [Accessed Sept 2009].
STERLING, B., 2004b. When Blobjects Rule the Earth. Available online at:
http://www.boingboing.net/images/blobjects.htm [Accessed July 2008].
STERLING, B., 2005. Shaping Things. Cambridge, Massachusetts: The MIT Press.
STORKERSON, P., 1997. Defining Design: A New Perspective to Help Specify the Field.
Available online at:
http://www.communicationcognition.com/Publications/ConstructivistDesign.pdf [Accessed June
2008].
TELESCULPTURE GALLERIES, 2005. available online at:
http://telesculpture.prism.asu.edu/ [Accessed June 2008].
TERZIDIS, K., 2006. Algorithmic Architecture. Oxford, UK: Architectural Press.
THACKARA, J., 1998. Design after Modernism: Beyond the object. London: Thames and
Hudson.
THACKARA, J., 2005. In the Bubble: Designing in a Complex World. Cambridge, Massachusetts:
The MIT Press.
THE ADVANCED COMPUTER CENTRE FOR THE ARTS AND DESIGN, Undated: history
http://accad.osu.edu/history.htm [Accessed September 2009]
THOMPSON, A., 1995. On Growth and Form. Canto, Cambridge University Press.
TODDS. and LATHAM. W., 1992. Evolutionary Art and Computers. Academic Press Ltd.,
London.
TOFFLER, A., 1970. Future Shock. New York: Bantam Books.
TOMES, A. & ARMSTRONG, P., 2003. Dialectics of Design: How Ideas of ‘Good Design’
Change. Available online at: http://www.ub.es/5ead/PDF/6/TomesArmstrong.pdf [Accessed June
2008].
TRIBE, M., 2005. New Media Art. Los Angeles: Taschen America.
TSENG, M. & JIAO, J., 2001. Mass Customization. In Handbook of Industrial Engineering,
Technology and Operation Management. 3rd. Ed. P 685.
UMEMOTO, N. & REISER, J., 2006. Atlas of Novel Tectonics. New York: Princeton Architectural
Press.
UNTO THIS LAST, 2007. Unto This Last. Available online at: http://www.untothislast.co.uk/
[Accessed June 2008].
284
UNVER, E; DEAN, L and ATKINSON, P., 2003. ‘FUTURE FACTORIES’: DEVELOPING
INDIVIDUALISED PRODUCTION METHODS, Advanced Engineering Design, AED, 1-4 June
2003, Prague, Czech Republic.
VEROTSKO, R., 2002 (a). Epigenetic Painting: Software as Genotype, A New Dimension of Art.
Available online at: http://www.verostko.com/epigenet.html [Accessed June 2008].
VEROTSKO, R., 2002 (b). Algorithms and the Artist. Available online at:
http://www.verostko.com/alg-isea94.html [Accessed June 2008].
WANDS, B., 2006. Art of the Digital Age. New York: Thames and Hudson, 2006.
WATERS, J., 2003. Blobitecture: Waveform and Organic Design. Gloucester, MA: Rockport
Publishers Inc.
WILSON, S., 2002. Information Arts: Intersections of Art, Science and Technology. Cambridge,
The MIT Press.
WOHLERS, T., 2003. Wohlers Report 2003. Fort Collins, Colorado: Wohlers Associates.
WHITE, H., 2004. Hybrid Practice – Challenging Traditional Craft Boundaries: Authenticity:
Anxiety: Autonomy. Available online at: http://www2rgu.ac.uk/challengingcraft/Challenging
Craft/pdfs/hazelwhite.pdf [Accessed June 2006].
Z, P., 2002. A Tool is a Tool in Women in New Media. MIT Press. Available online at:
http://www.pamelaz.com/tool.htm [Accessed June 2008].
ZELLNER, P., 2000. Hybrid Space: New Forms in Digital Architecture. London: Thames &
Hudson, Ltd.
285
286
Appendix 1
‘FUTURE FACTORIES’: DEVELOPING INDIVIDUALISED PRODUCTION METHODS
©Dr Ertu Unver, Lionel T. Dean, and Paul Atkinson, 2003
School of Design Technology
Huddersfield University
Queensgate, Huddersfield HD 3DH
United Kingdom
KEYWORDS
Mass Customisation, Mass Individualization, Rapid Prototyping, Virtual Reality, Virtual Merchandising,
Animation, Organic design, Interactive, Computer Generated, 3D modelling, 3D Printing.
ABSTRACT
'Future Factories' is an exploration of the possibilities for flexibility in the manufacture of artefacts inherent in
digitally driven production techniques. The concept considers individualised production – in which a random
element of variance over parameters such as the relative positioning of features, scale, proportion, surface
texture, and the like is introduced by the computer within a parameter envelope defined by the designer. This
paper is the feasibility study of, and design of, a production system for the 'Future Factories' concept.
In 'Future Factories', a production system is envisaged in which the consumer is presented with a 3D digital
model of the design. The design is presented as an animation showing the design morphing within a parameter
envelope specified by the designer. At any given point the consumer may freeze the design, place an order, and
generate the relevant digital production files (.stl etc.). A unique, individual artefact will then be manufactured
using Rapid Prototyping techniques. This may be achieved directly, via Stereo Laser Sintering in a suitable
material for example, or indirectly via the production of a single use tool or pattern.
This paper presents results from research conducted as part of the Designer in Residence project at the School of
Design Technology, University of Huddersfield. Firstly a selection of design concepts with associated parameter
envelopes are created using relevant 3D design software. Animations are then created showing the design
moving within its parameter envelope. A new computer program is being developed to enable the generation of
digital production files direct from a selected animation frame. There will be a study of existing rapid
prototyping techniques with regard to their suitability for direct manufacture of this type and speculation on
future potential.
INTRODUCTION
The School of Design Technology at the University of Huddersfield recently decided to allocate an amount of
research funding to provide an ‘Artist-in-Residence’ to work alongside Fine Art students, and a ‘Designer-inResidence’ to work alongside Product and Transport design students for a period of one year. The work
currently being undertaken by the Designer in Residence along with contributions made by other academic staff
are the subject of this paper. The title of the project ‘Future Factories’ describes an exploration of the creative
potential inherent in digital design and manufacture to offer more than a single discrete 3D outcome. The
outputs from this practice-based research project are expected to consist of a number of inspirational products
which will be exhibited in a traditional gallery environment and later digitally – either on-line or by CD-ROM
dissemination. Alongside the practice-based research outputs there will be a number of different academic
papers (such as this one) addressing the different technical, theoretical and contextual issues raised by the
content of the ‘Future Factories’ project.
THE 'FUTURE FACTORIES' CONCEPT
Rapid Prototyping technologies, developed to compress product development cycles, offer the potential for much
more. Layer-build production processes allow for the direct transfer of virtual CAD models to real objects. As
in reprographics, model files can be emailed to an agency for production or desktop printed on machines using
inkjet technology. Through direct digital production a revolution is underway in 3D Product Design that is likely
to be as radical as that already seen in Graphic Design.
In essence the project proposes an inversion of the mass production paradigm to one of individualised production
– in which a computer generated random element of variance is introduced. Each artefact physically produced is
a one-off variant of an organic design. The design is defined by parametric relationships and is maintained in a
constant state of metamorphosis by the computer software.
1
287
The variance introduced may be over factors such as the relative positioning of features, scale, proportion,
surface texture, pattern, and the like. These variable factors may be multiple and interrelated. The intention is to
achieve subtly different aesthetics around a central theme rather than mere differentiation that might be achieved
by, say, scale or colour change alone. We do not claim here that the notion of computer generated random form
is in itself an original one. Perhaps some of the best-known computer generated forms are those resulting from
the collaboration between the artist William Latham and the mathematician and computer graphics expert
Stephen Todd. Although the resulting ‘sculptures’ were only ever intended to be seen as 2D representations of
complex 3D models presented as art in a gallery context, the principle behind it can just as easily be used to
create variations on ‘usable’ forms to produce designs for ‘anything from buildings to shampoo bottles’
(Computer Artworks 2003). The potential of this proposition has not yet been fully realised. The ‘Future
Factories’ concept explores an aspect of this potential.
MASS CUSTOMIZATION
It is perhaps pertinent here to specify what 'Future Factories' is not - and it is not 'Mass Customization'. The term
'Mass Customization’ was coined by Stan Davies in his book Future Perfect (Davies 1987). The term is
deliberately paradoxical. There are many different models for mass customization suiting different products and
market sectors. They are all however, consumer driven. Products are "decomposed" into modular components or
subsystems that can be recombined to more nearly satisfy consumer needs.' (Crayton 2001: 78). In contrast to
mass customization, the 'Future Factories' model derives no input from the consumer. Where mass customization
consists of consumer selection and specification, 'Future Factories' allows the consumer only to select the
moment at which the process of form generation is arrested.
DESIGN FORMULA
In the 'Future Factories' model, rather than specify a discrete design solution, the designer sets up a series of
rules and relationships that achieve a desired aesthetic over a potentially infinite range of outcomes. This is
achieved using parametric CAD software. The aim is a degree of random mutation. This does not mean a series
of staccato jumps as one random value is replaced by another. The model should appear to ‘grow’ with one
mutation flowing seamlessly into the next. Each solution is intended to be unique and not repeated in a cycle,
however long. The desired result can be achieved by setting variables to cycle through specified ranges.
Different variables are set to cycle at different rates, with the differential providing the random element. In
addition, the rates of change can themselves vary, increasing or decreasing at random (though with smooth
implementation) over time.
Fig 1: A solid formed by lofted square profiles
EXAMPLE OF THE RANDOM ELEMENT GENERATION PRINCIPLE
To understand the principles proposed we can start with a simple box, as illustrated in figure 1, this will become
a product structural leg. A simple solid model is created by ‘lofting’ three square sections. ‘Lofting’ is the
2
288
creation of a 3D transitional form between usually 2D profiles and is a common feature in high end CAD. The
dimension value of each the squares (D1, D2, D3) is allowed to cycle 100% - 70%. Figure 2 shows the effect of
this applied to D3.
Fig 2: The effect of scale variance on one lofted profile
The cycling of the variable D3 is set at a given rate. The other two variables, D1 and D2 are set to cycle though
the same value range but at different rates. The effect of this is illustrated in figure 3.
Fig 3: The effect of differing scale variance on all three lofted profiles
Another element of variance that could be considered is the addition of a twist about a vertical axis formed by
rotating the horizontal profiles relative to each other as illustrated in fig 4.
Fig 4: The effect of rotating lofted profiles to produce twist
3
289
So far the mid-profile of the loft construction has always been located mid-way up the form, but this profile can
be allowed to rise or fall (figure 5).
Fig 5: The effect of altering the height of the central lofted profile
The rotation about the vertical axis and the asymmetric placement of the mid profile are assigned ranges and
independent rates of change. These transformations are overlaid on the earlier figure 3 model, and the resulting
forms illustrated in figure 6.
Fig 6: Combined effects of transformations
The Twist II candlestick (figure 7) was developed using these principles. The design has three legs which meet
at the top. Each of the legs has elements of variance similar to those used in the box example (figures 1-6). The
three legs morph independently but with a constraint to ensure the tops match. The legs are equally spaced at a
separation specified for stability. The footprint of the legs is allowed to both twist about a vertical axis and move
in a horizontal plane relative to the top to create further distortions. It can be seen that the scope for variance is
vast. It is important to highlight, however, that the changes in form are not arbitrary. Each of the variables has
been applied so that through their combination a desired aesthetic is achieved in an organic form.
4
290
Fig 7: The 'Twist II' Candlestick
Other models developed for the project include the Twist I candlestick (figures 8 and 9) and Lampadina Mutanta
(figures 10 and 11). Lampadina Mutanta is a luminaire with a light source of high intensity white Light Emitting
Diodes (LED's). The LED's are mounted in the ends of ‘tentacles’ which appear to grow at random from the
bulb form. The end of each ‘tentacle’ is dimensionally constrained to accept an LED and the direction in which
the LED points restricted to certain angles from the vertical (to avoid glare).
Fig 8: The 'Twist II' Candlestick
5
291
Fig 9: The 'Twist II' Candlestick
Fig 10: ‘Lampadina Mutanta’ lumiere
6
292
Fig 11: ‘Lampadina Mutanta’ lumiere
PRESENTATION
Each organic design is defined by a production formula which can yield an infinite range of equally valid
outcomes – How should such a design be presented to both clients and consumers? To appreciate their organic
nature the designs must be seen continuously ‘morphing’ in real time - this requires animation. A video clip
played on demand would not be sufficient as there is no cycle to the mutations. The consumer must be offered a
‘webcam’ window onto a design which is changing whether they are watching or not. Clearly the project lends
itself to some form of ‘virtual’ web-based merchandising. A system is envisaged in which the consumer is
presented with a 3D animated model of the artefact via a website. The consumer may access the website directly
or via a sales outlet, at a gallery or in a department store for example. The web site, the ‘Future Factory’ itself,
would have a series of ‘production lines’ corresponding to different products. When a particular production line
is selected the user is presented with a computer animation showing that particular product in metamorphosis.
At any given point the consumer may freeze the animation, effectively creating a one-off design on screen.
Should the consumer wish, they might then proceed with an order, in which case the relevant digital production
files (stl etc.) would be generated automatically and sent to the appropriate RP production facility. The unique,
one-off would then be manufactured using layer additive manufacturing (Rapid Prototyping) techniques. This
may be achieved directly, via laser sintering in a suitable material for example, or indirectly via the production of
a single use tool or pattern. It should be pointed out that the intention is not for the consumer to use the
animation to adjust design features to their liking. The animation is changing in real time and is outside their
control (this is part of the allure of the process). A variant can be ‘designed’ for them and they can choose to
order it or not. Visitors to the website must therefore be able to ‘read’ the designs, to appreciate the form of the
artefact with which they are being presented. This is made difficult by the fact that the object is changing. A
Virtual Reality experience would help overcome this difficulty, enabling the consumer to ‘move around’ the
design in their own time and at will.
WEB-BASED VIRTUAL REALITY (VR)
"Virtual Reality" refers to technology used to provide computer-based experience that mimics real experience.
The two primary characteristics of such an experience are that it is three-dimensional and that it is interactive. At
its simplest, it is a 3D image of a single object that the user can rotate to see from various angles. At a middle
level, it can be an entire scene or virtual world that the user enters and interacts with. High-end VR requires the
7
293
user to wear special goggles and gloves that make the illusion of reality more complete, this is the province of
games or specialised applications like military training, which are not typically Web-based. VR applications on
the Web include entertainment, education, medicine, marketing, training and, as in this application,
merchandising.
VIRTUAL MERCHANDISING
Conventional marketing usually centres around a glossy photograph of the product shown from its best angle.
VR content enables websites to bring the products 'to life' with 3D models, user interactivity, animation, sound,
and detailed views. An interactive image allows the product to be seen from all angles. Potential customers are
able to examine the design in the form of a 3D model moving, rotating, and zooming in and out at will. This
‘hands-on’ interaction allows something of a “try before you buy” experience. Internet shoppers have been
reported to spend 50% more time in the part of the site that offers interactive 3D images (Mike Hurwicz, 2000),
yet VR on the Web is not yet mainstream or widespread. Why has this exciting technology made such slow
progress? This may be due to the fact that VR content is costly to develop, mostly because the expertise to create
it is still rare, and a direct link to sales revenue is as yet unproven. In addition, content creators typically make a
substantial investment in computers, software, and digital equipment. Technical difficulties have discouraged the
take up of website based VR marketing. The technology is most convincing and pleasing when it uses realistic
textures, lighting and sounds. The use of these elements requires large files which leads to slow performance,
and web designers fight a constant battle between high graphical appeal and slow download times. Highresolution graphics and elaborate animations are notoriously slow to download, especially through a dial-up
connection. Add to that bandwidth limitations and the possibly unreliable connections of the Web, and you have
the potential for a deeply dissatisfying experience.
INHERENT LIMITATIONS OF VR
We can assume that consumers will always prefer hands-on experience with a product before purchase. Barring
cumbersome and expensive gloves and goggles, VR is still strictly an audiovisual experience. Even gloves have
serious limitations in that they cannot provide a tactile experience of texture. This might be solved if the website
access was via a retail outlet and samples were available. It would also avoid ‘user end’ technical issues. VR
content cannot be viewed with a standard browser - a special-purpose browser or a plug-in is required. In
addition there is no single viewer that can handle all VR content. The requirement to download additional
software merely to, in effect, browse a shopping catalogue is a severe disincentive. The user may not even have
administrator rights to the computer or the desire to involve themselves in IT issues. The lack of an industry
standard also affects content creators (and their cost), as each viewer typically has its own authoring tool. There
is no single tool that a programmer can learn with any expectation of addressing more than a fraction of Web
users. The limitations of computers and the technological demands of VR should not be exaggerated however.
VR files are not necessarily huge. Files consisting mostly of vector graphics are relatively small. Problems of
speed and resolution will be solved over time as standard-issue desktop computers gain speed and are optimized
for 3D graphics. Higher-bandwidth and more reliable connections to the Internet will also become more
common. Proprietary viewer and content creation software offer increasingly high-quality images within
compact, efficient files.
HTML, SGML, VRML, XML and X3D
Conceived when the Internet was still in its infancy, SGML (Standard Generic Mark-up Language), defined by
ISO 8879, was created to describe the ‘look’ of a document. SGML is very complicated and is best suited to
solving large, complex problems that justify its use. Viewing structured documents sent over the web rarely carries
such justification. HTML (Hypertext Mark-up Language) was developed in the early 1990’s, essentially from a
stripped-down version of SGML. HTML has become widely used in spite of never becoming standardized, with
vendors such as Microsoft and Netscape adapting the language to their needs. Neither SGML nor HTML are
particularly well suited to VR applications SGML being overly complex and HTML too rigid and inflexible for
development. In the mid-1990’s VRML Virtual Reality Modelling Language was developed with the aim of its
becoming a standard (supported by the ISO). VRML is a 3D interchange format and is essentially a 3D analogue
to HTML. It defines most of the commonly used semantics found in today’s 3D applications such as hierarchical
transformations, light sources, viewpoints, geometry, animation, fog, material properties and texture mapping. It
integrates three dimensions, two dimensions, text and multimedia into a coherent model (Carey and Bell 1997).
The uptake of VRML, however, has not been without problems. The language has an extensive set of required
features which demand large browsers and plug-ins. The need to integrate with a complex existing feature set also
hinders innovation. XML (Extensible Mark-up Language) has emerged as an alternative to, or fix for, VRML. The
Core Profile of XML is much lighter than VRML, and additional profiles are implemented as software components
to be downloaded as necessary. The user only uses the profiles needed to view the current content, whereas a
VRML browser may have many features they don't need. When an innovator comes up with a new profile, they
8
294
can achieve minimal compatibility by testing only with the Core Profile. The VR community has recognized the
growing success of XML, compared to the very limited success of VRML. In response, the Web3D Consortium
(http://www.web3d.org), in concert with the W3C (World Wide Web Consortium), has defined an XML-compliant
3D standard for the web: "Extensible 3D" (X3D). X3D extends the capabilities of VRML and provides a means of
expressing the geometry and behaviour capabilities of VRML using XML.
3D MODELLING IN THIS STUDY
In this study a series of organic product designs have been created using Alias Wavefront, Solidworks and 3D
Studio Max. Domestic interior products, principally lighting and tableware, have been considered for the project
thus far. Domestic interior products is a market well used to paying a premium for design and materials
technology. Lighting and tableware have been selected to keep the artefacts relatively small. This consideration
is based on cost rather than capacity - the largest laser sintering machine commercially available in the UK is
700 x 500 x 350mm for manufacture in one piece, and building in sections could also be considered. The designs
selected thus far are for production in cast metal. They make use of layer additive production methods to
achieve complex forms almost impossible to achieve with multiple use tooling. This necessitates the use of
investment casting, with the wax patterns for use in the process being produced by a layer additive process.
ANIMATION AND PUBLISHING TO THE WEB
There were two options for publishing the designs to the Web. The first was to employ a high end programming
language Java3D (an application programming interface, API, for drawing 3D graphics using the Java language)
or C++. We evaluated this option in our study using Parasolid Kernel to access the 3D information directly from
CAD and then animating/manipulating coordinates using C++ codes. This proved difficult and time consuming.
Large file size was also problem as no optimisation procedure was easily available. The adoption of this
methodology would require the designer to work via a specialist computer programmer.
The second option was to employ a proprietary ‘plug-in’. We have already generated 3D CAD data (using Alias
Wavefront format in this instance), which can be exported and animated (we used 3DStudio Max). Several
‘plug-ins’ are readily available for the publication to the web of 3D content including animations. One such
package, Viewpoint, was used in this study. Animation and interactive 3D files were created and exported to
Viewpoint for the creation of interactive 3D mtx files using Media Export Utility. These files were then edited
using Viewpoint Scene Builder to set the required controls for scene animation in Javascript. The model was
then transferred to the web using HTML with embedded Viewpoint XML.
CONCLUSIONS
It is clear that the implications of the widescale adoption of such techniques by industry are potentially serious, and
are such that moves to protect the process via patents have been made. The system has the potential to change the
perception of design by consumers and manufacturers alike, and to influence considerably the education and
training of designers. Despite the philosophical questions the process raises for the definitions of terms such as
'design' and 'designer', (which are potentially misleading in this context), and the scope for confusion as to whether
the end results of the process are 'art', 'craft', or 'computer generated' (which are, perhaps, topics for debate in a
different arena to this one), there are a number of more pragmatic considerations. The potential for the process to
impact on manufacturing and retail industries should not be overlooked. 'Future Factories' allows for the economic
large scale production of artefacts while providing important reductions in wastage arising from the overproduction of unwanted items, while promoting the move from reductive to additive manufacturing processes. As
such, it may point the way to a more sustainable model of a consumer society than the one we take for granted
today.
REFERENCES
Carey, R and Bell, G (1997) The annotated VRML 2.0 Reference Manual
Computer Artworks (2003) http://www.artworks.co.uk/index2.htm
Computer Artworks (1995) Organic Art software, GT Interactive http://www.artworks.co.uk
Crayton, T (2001) 'The Design Implications of Mass Customisation' in Architectural Design, April 2001
Davies, S (1987) Future Perfect, New York, Addison-Wesley
Hurwicz, M VRML or XML?, Web virtual reality and 3D, June 2000
Todd S & Latham W (1992) Evolutionary Art and Computers, Academic Press Ltd, London
http://www.web3d.org/
http://www.oasis-open.org
http://www.xml.org/
http://www.cai.com/cosmo/
http://www.viewpoint.com/
9
295
296
Dean, L. Atkinson, P and Unver, E.
Evolving individualised consumer
products
Abstract
The origins of this project began in 2002 with experimentation into the application of
computer generated random form to 3D product design. Advances in the Rapid
Prototyping industry were offering the possibility of mss-produced one-off consumer
products. Computer based 3D solid models were created that would randomly mutate
within parameter envelopes set by the designer. At any given point the mutation
could be halted and a real-world product generated via digital manufacture (Rapid
Prototyping). This first stage of the work has already been reported on (Atkinson and
Dean, 2003).
The next phase of the program has been to introduce evolutionary development so
that, via the computer generated random mutation, the model develops generation by
generation in a desired direction (though not necessarily to a predictable outcome).
This requires an element of selection. There are several examples of computer based
evolutionary design experiments that use human by-eye selection methods, notably
Richard Dawkins’ ‘Biomorph’ system (Dawkins 1993). The aim of this project is an
automated system that selects on some measure of desirability and rejects outright any
functional failures.
Each FutureFactories product form is define by a parametric CAD (Computer-AidedDesign) model. When evolution is initiated, a series of mutant designs are generated
each with a single parameter, selected at random, adjusted by a small pre-determined
step. The step may by be positive or negative, this again is determined at random.
The resulting set of mutant progeny is then assessed for their visual ‘success’ using a
quotient. The quotient aims to access the level of visual interest in a form. As the
application is 3D products, there are physical parameters to consider, for instance
‘hard points’ generated by the envelopes of internal components which may not be
intruded upon. If any of the offspring do not meet the necessary physical criteria they
are rejected. Animation is employed to extrapolate between iteration present the
evolution as a smooth metamorphosis. Product forms and associated development
criteria have been created capable of evolutionary development over many
generations. The resulting designs are both surprising and unpredictable.
297
Introduction
Future Factories is a digital design and manufacturing concept for the massindividualisation of products. The project began as a one year Design Residency in
School of Art and Design at the University of Huddersfield. The project has now been
expanded into a practice-based PhD study. Instead of creating a single discreet design
solution (or indeed a finite range of options), the designer creates a template. This
template defines not only the functional requirements of the form but also embodies
the character of the design. Through the design template, the designer establishes a
series of rules and relationships which maintain a desired aesthetic over a potentially
infinite range of outcomes. The design becomes a ‘living’ entity, continuously
morphing within its template envelope (Atkinson and Dean, 2003). In a development
of the project we have looked at coupling random mutation with selection and the
introduction of evolutionary pressure.
This application of computer based
evolutionary design is the subject of this paper.
Technological context
Computer generated artwork has become commonplace, the creation of three
dimensional artifacts from this artwork imposes considerable limitations and is
consequently rare. Advances in digital technologies have made the creation of oneoff products from computer generated models, a realistic, affordable possibility.
There are three principle technologies exploited in the FutureFactories model (fig. 1).
figure 1
Three core digital technologies exploited by FutureFactories
Parametric computer aided design
Parametric computer aided design (CAD) enables the designer to define relationships
that form the character of a design rather than a single, discrete, design solution.
Parametric design considers the relationships between degrees of freedom rather than
the degrees of freedom themselves. When a variable is changed the whole model will
298
up-date to maintain specified proportional relationships. Individual variables in the
computer based 3D model can be modified and the whole form will up-date to
maintain specified relationships.
Digital Manufacture (Rapid Manufacture, Direct Manufacture)
Now that Rapid Prototyping (RP) is well established – the new frontier for the digital
manufacturing industry is Direct Manufacture. Direct or Rapid Manufacture is
essentially the adaptation of RP technologies to the manufacture of end-use products.
“A number of compelling examples of RM suggest that it will span across many
industries in the future. Among these are hearing instruments, dentistry, medicine,
aerospace, military, oil exploration, motor-sports, and consumer products” (Wohlers
2003). Rapid Prototyping (RP) is a catch-all term that applies to the digital
manufacture of prototypes directly from CAD data. Essentially RP allows on-screen
models to ‘printed out’ in 3D. Most of the recently developed processes are layer
additive. Software ‘slices’ the CAD model into thin layers (down to 0.05mm). The
model then ‘grows’ one thin layer at a time as each data ‘slice’ is replicated in 3D
from the bottom up. The layers are built on a moving platform, each built on its
predecessor as the platform steps down in layer thicknesses. There is no tooling or
cutting away of material. This allows unlimited geometry. Forms may be produced
that would be almost impossible to mould or machine. The relatively slow layer-bylayer building means that digital manufacturing is unlikely to ever match production
capacity of die casting and injection moulding. Manufacturing parts without the need
for moulds or dies does however makes the volume production of individualised
forms an economic possibility.
Graphics
In FutureFactories the product forms are not fixed. The designs exist in a constant
state of metamorphosis. To appreciate this, customers should be able to see the
designs continuously ‘morphing’ in real time. The concept lends itself to some form
of ‘virtual’ web-based merchandising (Unver, Dean and Atkinson, 2003). A system
is envisaged in which the consumer is presented with a 3D animated model via a
website. The consumer may access the website directly or via a sales outlet within,
for example, a gallery or a department store. Advances in the graphics capabilities of
home PC’s and the speed of internet connections allow the display of rendered forms
mutating in real time on the customer’s home computer. Memory hungry threedimensional rendering now exploits graphics processors on the video cards instead of
consuming valuable CPU resources when drawing 3D images. These advances in
video cards and the software that manage them, driven hard by the video game
industry, enable the smooth real time display of animated forms complete with
realistic scene lighting and material finishes.
The introduction of selection into the FutureFactories model
In the original FutureFactories concept it was necessary to define a complete envelope
of parameters. The envelope defined ‘solution space’ covering every possible
mutation of the form. Each individual parameter required specified ranges that
considered both the effects of that parameter alone and its effects in combination with
others. It is clear to see that if there are more than a handful of parameters and their
effects interrelate to any significant degree then the task of specifying such an
envelope becomes extremely long and complex. An aim of FutureFactories is to
299
develop generic systems of commercial potential. For commercial viability it should
be possible to introduce new designs with reasonable ease. Ideally one would be able
to apply mutation rules to a conventional 3D model via an intuitive on-screen process
(the system is seen as a plug-in addition to high-end parametric CAD systems). The
complexity of specifying a parameter envelope could be reduced by severe restriction
of the permissible parameter ranges and by isolating their effects where possible. But
this would lead to uninspiring, predictable, movement in the form, repeated
oscillations for example. A way of simplifying the rules for mutation had to be
found. Evolutionary design principles offered a potential solution. Genetic
algorithms permit virtual entities to be created without requiring an understanding of
the procedures or parameters used to generate them. Instead of incorporating expert
systems of technical knowledge into the programming, evolutionary design systems
rely on utilitarian assessments of feasibility and functionality (Sims 1999, Funes and
Pollack 1997). Our parameter envelope was designed to perform two functions; to
ensure that manufacturability be maintained and that the mutated form retains the
‘designers intent’. If these factors can be assessed and scored then mutation, coupled
with selection, can be used to drive and control changes in form.
A model for mutation and selection
In the FutureFactories model, mutation takes place in a series of generations. The
original model had a single parent producing a single, randomly mutated, child per
generation. In this development of the system each generation has a single parent (the
starting point) and ten offspring. In each generation, ten randomly mutated iterations
are generated from the parent model. Each iteration has a single parameter, chosen at
random, modified by a set small amount. The mutant offspring are ranked for
‘fitness’ and the most successful selected as the parent of the next generation. The
scoring for fitness is based on the ‘desirability’ of the last transformation with
reference to the designer’s intent. Before becoming the parent of the next generation
the selected iteration is tested against functional failure criteria. This ensures that
after mutation the design is no less manufacturable. If the parent fails this assessment
the next best offspring is selected. Only selected offspring are tested against the
failure criteria to reduce computation. Animation is employed to provide a flowing
transition between one generation and the next. The generations are a fixed number
of animation frames apart with the software extrapolating to fill in the missing frames
(fig 2). This is known as key frame animation.
300
figure 2
The mutation, selection and animation process
The introduction of an evolutionary pressure
Given the use of random mutation and selection, the introduction of an evolutionary
pressure was a logical step. Indeed it would be hard to avoid creating such pressure
by virtue of the ‘fitness’ scoring. Public reaction to the early stages of the project also
pointed to the inclusion of an evolutionary element.
Public reaction
As part of the Residency program at the University of Huddersfield, a touring
exhibition was arranged to communicate the project to a wider public. The exhibition
toured three regional venues, before going on to London and Milan. At each venue
interactive displays were set up. Visitors were invited to ‘try out’ the system by
selecting their own one-off designs from a computer rendered image of the design as
it mutated randomly in real-time. Users received a 2D printed image of their
individual design, which mimicked the production proposal, in which a 3D model
would be digitally built. This gave the opportunity to assess levels of consumer
interest and expectation.
301
The mutating image proved initially extremely seductive, with visitors drawn to the
image and captivated by it. The selection process proved less of an attraction; users
were often just as happy for the choice to be made for them. To some extent this is
understandable, as no actual purchases were being made and no 3D objects would be
generated. But it was nevertheless apparent that as the mutation was completely
random there was little intrigue in ‘what happens next’. Creating this type of intrigue
is obviously important from a marketing point of view. A level of evolutionary
development is seen as a way of stimulating this type of interest. The idea is that
designs would be available and evolve for a limited period. Different periods of the
evolution process may achieve different levels of desirability. The value of an artifact
would vary according to its position in the evolution. There may be ‘good’ and ‘bad’
generations as there are good and bad vine harvests.
Aesthetic Evolutionary Design
The aim is not the functional optimisation of the designs through evolutionary
computation. The suitability and functionality of the design are present in the initial
seeded product form. Functionality is then maintained by the selection process, rather
than improved upon. The aim is the evolution of aesthetic designs in what is
described by Bentley as Aesthetic Evolutionary Design (Bentley 1999), an area that
borrows from both Evolutionary Design Optimisation and Evolutionary Art (fig. 4).
302
figure 4
‘Tuber’ pendant lamp produced by 3D printing
Failure criteria - feasibility, functionality and manufacturability
Assessing the mutant designs for feasibility, functionality, and manufacturability is
relatively straight forward. The validity of the surfaces created can be assessed
through the ability to export a suitable digital file for manufacture. Problems, such as
overlapping surfaces, either prevent successful export, or are flagged up by error
messages. The manufacturing limitations of the intended digital manufacture process
can be imposed, minimum material section thickness and the machine build envelope
for example. FutureFactories has experienced problems with clearing fine internal
passageways of unused build material, this can be mitigated against with a limiting
bore diameter/length ratio.
Functionality may consider issues such as stability,
checked via the position of the center of gravity, and the appropriate housing of
internal components. These practical assessments are used to impose absolute limits
rather than for relative scoring. The aim of FutureFactories is not technical
refinement. It is not the intention to select the quickest to manufacture or the most
stable; merely to assure that each generation conforms to a minimum functional
standard.
303
Scoring the aesthetic
Maintaining the designer’s intent requires a more relativist approach. Selecting
designs based on a scoring of ‘fitness’ allows the designer to express general ideas for
the design rather than absolute limits. For example the notion, “some rotation is fine
but not too much,” might translate to an exponential decrease in the probability of
further rotation being selected as the angle increases. This less rigid form of
definition simplifies the set up of the model and also allows the possibility of new
unexpected forms (although the possibility of surprising turns has to be balanced
against maintaining a coherent, identifiable design). The effects of the rules are
‘softened’ by the use of probability: a high fitness score can be allocated a higher
probability of selection rather than assured selection. This again broadens the
possibilities allowing from time to time the success of a less fit parent.
Step size – micro-mutation, macro-mutation and the balance between
different transformations
Evolution is the result of accumulated small change. If the geometry of the model
were to be re-arranged at random there would be infinitely more ways of creating a
failure than a success. As Dawkins points out of the natural world, “Even a small
random jump in genetic space is likely to end in death. But the smaller the jump the
less likely death is, and the more likely is it that the jump will be in
improvement…………….The chance of improvement resulting from a transformation
tends to zero with increasing step size and to 50% as it decreases” (Dawkins 1986).
Also the more transformations that are occurring simultaneously, the lower the
probability that they will all be successful. For this reason each offspring ‘bred’ from
the parent form has only one parameter adjusted at random +/- one ‘small’ step. The
step size is an absolute value arrived at through experimentation. A step size is set for
rotation, transformation and scaling. The values of these different steps must be
balanced so that they each achieve a comparable degree of change to the form. If
particular transformations have disproportionate effects, they will inevitably exclude
milder transformations from the evolutionary process. A diagnostics screen has been
incorporated into the system to guide the setting up process. Amongst other
information this screen is shows the percentage breakdown between the three
transformation types that have acted on the model up to the current point in the
evolution. It is possible to see the balance between the operation types as the
evolution progresses.
Evolution – what are the aims?
The designer creates both the initial form of the design and the evolutionary pressure
that will govern changes in that form over its evolutionary lifespan. The aim is to
evolve increasingly visually interesting designs along the path set by the designer. The
use of digital manufacturing favours more complex forms. If the forms are simple or
regular, then the options increase for manufacture via faster, cheaper, conventional
methods. So whilst simplicity may have elegance, FutureFactories evolutions will
necessarily tend toward the more complex.
We have considered surface area divided by volume as a measure of complexity.
Dividing by volume prevents simple expansion. When applied repeatedly to a simple
304
model, the resulting forms, after 200 generations, are clearly related, more intricate,
and yet still manufacturable (machine build area is used as a failure criteria).
figure 5, the initial form and the form after 200 generations
Detail, grouped, and structural changes
The product forms of the FutureFactories models are made up of surfaces defined by
control curves. It is these control curves that are manipulated during the evolutionary
process. Each 3D iteration (in evolutionary terminology phenotype), is defined by a
list of parameter values (the genotype). Parametric CAD generates the 3D form from
this list or genotype. The evolutionary algorithm modifies the list generating mutated
genotypes which in turn, via CAD, create mutated 3D objects (fig6).
305
figure 6
The CAD model phenotype and its Parameter genotype
Manipulating individual parameters results in what could be considered as, local,
detail changes. As well as detail changes it is often desirable to manipulate larger
areas of the form with the same transformation. In a legged structure for example, it
may be desirable to apply transformations to legs as a whole rather than specific areas
of individual legs. For this reason FutureFactories allows the grouping of parameters
at the set up stage, for example, a leg group. The grouped parameters are treated in the
same way as the individual ones with a certain probability of random mutation. A
transformation may be applied to the grouped parameters, or to individual parameters
within the group: the percentage probability of each being dictated by the set-up rules.
The particular transformation may be therefore applied to a small area as a detail
change or may be spread over a particular feature (fig 7).
306
figure 7
Mutation sphere of influence
Structural change involves an alteration of the geometrical make up of the model,
rather than adjustment of it. This could be the addition or removal of features, for
example, an additional leg on a legged structure. This type of change is very difficult
to accommodate in the FutureFactories due to the surface based geometry of the
current models and the requirement that each iteration produces a potentially viable
product. The system does not allow for the evolution of new features from
functionally compromised beginnings. Complex natural systems, such as the human
eye, have evolved from much cruder beginnings, like perhaps the light sensitive spots
processed by some single celled animals (Dawkins 1986). One can imagine the
parallel in the Tuber lamp (fig. 8). A new limb might evolve beginning as a small
protuberance on the surface. This would elongate and develop a slight glow to the tip.
The glow then intensifies, until it becomes the intense focused beam of the LED. This
unfortunately belongs to the virtual world. FutureFactories is able to individualise
product forms, but standard, interchangeable functional components are still required.
An LED has a fixed size and specification. It is either there, or it is not.
307
Figure 8
Tuber9 produced in Laser sintered nylon
A degree of structural evolution is desirable, if not essential. FutureFactories achieves
this by breaking the design down into an assembly of separate models. Tuber consists
of limbs that intersect. These are separate solid models joined by a Boolean
operation. The requirement is that all four limbs remain linked by enough material to
achieve a structural joint. The format of the assembly can change during the
evolution as long as all four limbs remain linked. A link can pass from one limb to
another in the manner of a baton being passed in a relay race (fig. 9).
308
figure 9
changes in model structure
FutureFactories vs. Evolutionary art
FutureFactories employs many of the principles seen in evolutionary art. There are
also major differences. Evolutionary art and FutureFactories are similar in that they
have no fixed target solution. The design is not homing in on an ideal generation by
generation (Although in evolutionary art, with manual selection a user might focus on
his/her preference). What is important, is the level of development: this usually means
complexity. The number of generations “progress” from the start point. The
importance is the distance from the starting point in solution space rather than a
particular region of it.
Organic art often starts with simple geometric primitives; effectively a blank canvas.
FutureFactories starts with well developed, non random, seeded solutions, a viable
design that must be maintained throughout the evolutionary process.
Evolutionary art exists in a virtual world. The constraints of the physical world,
gravity for instance, need not exist. In evolutionary art anything is possible and the
images are usually scaled as required to fit a convenient screen area. “The scale of
forms generated from the same structure can vary by huge amounts as the parameters
change: a single family can easily include both whales and insects” (Todd and
Latham, 1999). Functional products must adhere to physical rules. In commercial
manufacture, certain products would be destined for production in certain machines.
The machines have build envelopes, into which parts must fit (although it is possible
to subdivide a form into smaller components that are subsequently assembled into a
larger structures using built in fastenings). There must be an element of repeatability
in the production process if volume production is to be economically viable.
Iterations of the same design, in spite of differences in form, should be produced in
the same machine and use the same packaging (elements of protective packaging can
be incorporated into the build process). Dimensions in FutureFactories are absolute,
with limits imposed to ensure manufacturability.
309
Evolutionary art often allows a user, or ‘artist’, to guide the evolution. The
FutureFactories selection process is automated: there is no human input during the
evolutionary process.
One of the ‘drawbacks’ for Evolutionary art is that the images generated often have
very distinct styles. “Often the style of the form generated using a particular
representation is more identifiable than the style of the artist used to guide the
evolution” (Bentley, 1999). The representations used are often limited to particular
types of structure and generate forms with common, readily identifiable elements. In
FutureFactories this is an aim - to produce designs that remain identifiable in spite of
mutation.
The nature of the FutureFactories designs
From the project’s inception, communication of the FutureFactories concept was an
important factor. The example designs created had a strong flavour of organic growth
in the aesthetic. The name Tuber and Tuber’s colour – vivid green, were seen as
factors in selling the concept. A frequent question raised is, given that the designs to
date have such a strong organic flavor – could the system be applied to other
aesthetics, to something more geometric?
Beyond the ‘marketing’ of the concept, there are other reasons for the preference of
organic forms. Firstly, the example products produced to date are the work of one
designer; inevitably the work reflects his tastes and ideas. Secondly, the virtual
models are literally growing: natural organic forms are the result of growth and so the
connection is hardly surprising. Thirdly, one of the transformation types employed is
a twisting motion. Twisting a form is almost certain to result in the generation of
curves. Where surfaces are formed between control curves, they are geometrically
constrained to flow smoothly one curve to the next.
Geometric aesthetics are not being overlooked however. One of the areas for future
work is an evolution that favors the creation of flat surfaces, straight edges, and
angular relationships between faces. The evolution would start from a simple organic
base and evolve into something geometric and faceted.
Conclusions
The use of simple fitness scoring and failure criteria has been used to replace the
‘parameter envelope’ of the original work. Using this evolutionary algorithm,
selection based approach, represents a huge reduction in complexity. Consider the
simple limb used in the earlier examples (fig 6). Instead of setting ranges for its 36
parameters, some of which interrelate and cannot be considered in isolation, a scoring
system is used. Surface area/volume and machine build envelope limits control the
model’s evolutionary mutation. Running the evolutionary algorithm for 200
generations results in closely related, but at the same time, distinct solutions. The
design progresses along slightly different pathways to the same region in ‘solution
space’. If we are confident that after a given number of generations the forms, whilst
different, will conform to a broad design concept, we can allow the evolution to run
repeatedly. On top of the initial design, the designer needs to specify selection criteria
that will focus the evolutionary development on a solution space region, as broad or as
310
narrow, as required. This gives the possibility of running the evolution on the
customer’s home computer, rather than on the host server. Computationally, this is a
much more attractive solution than the customer accessing an evolution on a host
server: however, conceptually the main benefit is in the flexibility. Running the
evolution on the customer’s home computer means that the evolution can be started
on-demand. The customer can run the evolution at will, stopping, starting and
resetting as desired.
FutureFactories focuses on a single mutating solution. Evolutionary algorithms are
usually much more sophisticated. They often feature populations of solutions, and
two parents, both of whom contribute to the offspring’s ‘genetic’ make up. This
‘crossover’ contributes to the evolution as well as mutation. So far FutureFactories
has been very broad in its aims for evolution. As the complexity of the models, and
the degree of evolutionary control required increase, it is likely that the models will
become susceptible to ‘noise’ and ‘local optima’ (solutions that score high on ‘fitness’
but are not ultimately the ‘target’).
The scope of the evolution possible within FutureFactories is restricted. It is limited
by the use of standard components, by the geometry of the model, and by the
requirement that the iterations remain recognisable designs, true to the designer’s
intent. Trials have shown us that customer demand is for significant change in the
forms. It is also seen as desirable that, whilst conforming to a design idea, the
evolved form contains some unexpected twists.
The potential for evolution is restricted by the internal components in the sense that,
whist in principle the skin of the design might be allowed to mutate, significant areas
of the form will be dictated by standard functional components. Ideally the entire
product should be allowed to evolve including any functional components. It is
possible that such components could be built digitally along with the body. This
already happens with some simple mechanical devices for example, springs, bearings,
clips and hinges. There are also machines capable of building in more than one
material simultaneously. There are research machines capable of ‘growing’ circuitry
on electronic substrates (de Garis 1999). It is safe to assume that technology will
make components ever smaller and easier to package. New materials and possibilities
for digital manufacturing are emerging all the time, with ever increasing performance.
It will become possible to achieve more and more functionality from digital builds.
Simplicity in the model has been sought to facilitate the creation of generic systems
rather than a discrete examples. We have sought to maintain simplicity whilst
allowing a degree of structural change through a model made up of multiple bodies.
This could be taken further towards the building block approach common amongst
evolutionary systems in which geometric primitives are added, subtracted, and
modified to achieve a desired form. These methodologies however do not in
themselves produce viable products. Further operations are required to translate the
primitive blocks into the functional components. The evolution takes place on a
simplified model. Each time a real product is required a set of mapping operations are
performed, for example, primitive blocks are united into a single volume; this would
then be smoothed and hollowed out. The FutureFactories customer sees the evolution
occurring in real time. Either the customer is presented with the simplified model or
311
the mapping operations must be computed for each generation. The latter approach is
impractical, requiring too much computation. Presenting the customer with a
simplified version is open to misinterpretation. If the mapping operation makes
significant changes to the model, then too much is left to the imagination. In the
FutureFactories multiple body model, the animated evolution shows the separate
bodies simply intersecting. Outputting a 3D model gives the intersection between the
forms a fillet radius. An integrated form is made from overlaid separate entities.
Visually the product becomes more realistic and ‘believable’ after this ‘mapping’ (fig
10): however, a reasonable impression can be gained from the simplified animation
model. From a computational point of view leaving complex modelling operations to
a mapping stage, completed only if a 3D outcome is required is highly desirable. But
visually the animation needs to be close to the final outcome. A square block
representation of a soft sculptural form, for example, would not be acceptable.
figure 10
Animation model vs. 3D output
Scoring surface area/volume represents a crude beginning. Other assessment
methodologies are under consideration. Consideration is also being given to the
number of polygons required to idealise a surface, average surface curvature, and the
spread of surface-normals as fitness criteria: the results to-date are promising.
The potential impact of 'Future Factories' have been noted and described
since the first stage of this work as mentioned (Atkinson & Dean 2003)
and are still considered to be significant. As the project has
developed, additional elements have been recognised with respect to the
system's impact on issues of authorship and accepted notions or
312
definitions
of
design
practice
(Atkinson
&
Hales
2004).
Clearly 'Future Factories' is an example of emerging and converging
technologies and new practices which are forming a new position for the
maker and author as the creative source of finished pieces. In fact, the
designer may not even be aware of products selected and produced in his
or her name. The combination of mathematical algorithmic processes and
autonomous production potentially act to isolate the author of the work
from the outcome, and raises questions of responsibility and ownership.
Finally, the use of software processes and real-time networks as
generative tools questions existing, transient boundaries of practice,
and
also
exposes
the
relevance
or
irrelevance
of
conventional
definitions and accepted nature of the roles, practices, techniques and
processes involved. It is clear that the outcomes of such a new model of
creative production cannot be thought of as traditionally conceived
pieces. They are, without question, art. Outside of that, existing
definitions convey little of the reality of their production, as they
lie in some new, as yet unspecified arena of production.
References
Atkinson, P (2003) Future Factories: Design work by Lionel Theodore Dean,
University of Huddersfield. ISBN 1862180474
Atkinson, P. and Dean, L. (2003). Teaching Techné. 5th European Academy of
Design Conference, Barcelona. http://www.ub.es/5ead/PDF/10/Atkinson.pdf
Atkinson, P & Hales, D (2004) Chance would be a fine thing: Digitally Driven
Practice-based Research at Huddersfield, published in the proceedings of the
pixelraiders2 conference, Sheffield Hallam University, ISBN 1843870606
Bentley, P. (1999). Evolutionary Design by Computers. Morgan Kaufmann
Publishers, Inc.
Dawkins, R. (1986) The Blind Watchmaker, Penguin Science pp.70-74; pp.77-85;
pp.231-235
de Garis, H. (1999) Artificial Embryology and Cellular Differentiation. Evolutionary
Design by Computers. Morgan Kaufmann Publishers, Inc. P282
Funes, P. and Pollack, J. (1997). Computer Evolution of Buildable Objects. Fourth
European Conference on Artificial Life, Cambridge, MA:MIT Press. pp.358-367
313
Sims, K (1999) Evolving Three-Dimensional Morphology and Behavior. Evolutionary
Design by Computers. Morgan Kaufmann Publishers, Inc.
Todd, S. and Latham, W. (1992). Evolutionary Art and Computers. Academic Press.
Unver, E. Dean, L. and Atkinson, P (2003). ‘Future Factories’: developing
individualised production methods. International conference on Advanced
Engineering Design, Prague
Wohlers, T. (2003) The Wohlers Report
314
Appendix 3
Design pedagogy: basic design and academic experiences / 1
'Future factories':
teaching techné
Paul Atkinson and Lionel Theodore Dean
INTRODUCTION: BACKGROUND TO THE PROJECT
,The phenomenon of the ‘Artist-in-Residence’ has a long-standing precedent in many areas of social and business
activity where the imperative to present a different perspective on a number of aspects of everyday activity and
to bring art into otherwise aesthetically impoverished environments has been seen to be of great benefit.
Consequently, their appearance in corporations and state institutions is well known. Their place in an art
education setting is perhaps less frequent, but by no means unusual, as the educational value of regular exposure
to a ‘qualified’ or ‘experienced’ practitioner carrying out their own work has long been recognised. However, the
use of a ‘Designer-in-Residence’ in a design for production education setting (as opposed to a designer-maker or
craft environment) is perhaps even less well documented.
The School of Design Technology at the University of Huddersfield recently decided to allocate an amount of
research funding to provide an ‘Artist-in-Residence’ to work alongside Fine Art students, and a ‘Designer-inResidence’ to work alongside Product and Transport design students for a period of one year.
The detailed description of the role of the Designer-in-Residence in educational terms; the benefits to students in
improving project management and time planning; and seeing the pace of professional design work in real time
are substantial, but perhaps the subject of a slightly different paper to this one. Here, we wish instead to
concentrate not so much on the process of using a Designer-in-Residence, but on the content of the particular
project being undertaken, the far-reaching implications the work has for the practice of design and design
education both on a theoretical and philosophical as well as a more pragmatic level.
The title of the project ‘Future Factories’ describes the exploration of the potential for direct digital manufacturing,
using the latest CAD 3D modelling and rapid prototyping techniques, in which a random element of variance is
introduced by the computer software. The outputs from this practice-based research project are expected to
consist of a number of inspirational products produced as a result of the residency itself, which will be exhibited
in a traditional gallery environment and later digitally – either on-line or by CD-ROM dissemination.
Alongside the practice-based research outputs, it is hoped there will be a publication describing the parallel
Designer-in-Residence and Artist-in-Residence projects at Huddersfield in a pedagogic context, as well as a
number of different academic papers (of which this is one) addressing the different theoretical and contextual
issues raised by the content of the ‘Future Factories’ project.
THE 'FUTURE FACTORIES' PROJECT
'Future Factories' is an exploration of the possibilities for flexibility in the manufacture of artefacts inherent in
digitally driven production techniques. All such production techniques are considered, the focus however is on
the layer additive manufacturing techniques associated with Rapid Prototyping (RP). In essence the project
proposes an inversion of the mass production paradigm to one of individualised production – in which a random
element of variance is introduced by the computer within a parameter envelope defined by the designer. Each
315
2 / Design
pedagogy: basic design and academic experiences
artefact physically produced will be a one-off variant of an organic design that has been defined by the designer
and maintained in a constant state of metamorphosis by the computer software. This variance may be over
parameters such as the relative positioning of features, scale, proportion, surface texture, pattern, and the like.
These variable factors may be multiple and interrelated. The intention is to achieve subtly different aesthetics
based on a central theme rather than mere differentiation that might be achieved by say scale or colour change
alone. This random variance would simulate the lack of uniformity in one-off craft production where the
craftsperson may be guided by a design intent rather than a toleranced production drawing. In this way, ‘Future
Factories’ aims to overcome the split between the technological and the aesthetic, between artistic creativity and
machine production – addressing in essence, ‘Techné’ – the integration of beauty, technical knowledge, and
industry.
WHY INDIVIDUAL PRODUCTION?
Mass production itself is a relatively recent concept. Prior to mass production artefacts would be produced by
craftsmen whose individual skills would be reflected in the product. The artefacts produced would be individual
interpretations of a design formula. Each artefact produced could be more or less faithful to the original
‘specification’. The design might be adapted to suit changes in stock material or to work around a fault or
blemish. As well as variance introduced by the manufacturer the process itself might have an effect. The process
used might not be fully controllable. Many craft processes are a balance between demands made of the process
and the control of it, hand-blown glass for example. A craftsperson's mistake, rather than resulting in scrap,
might produce an interesting twist on an old theme. The design formula itself might be organic, developing and
mutating over time. This lack of uniformity, far from being seen as a negative by the consumer is often valued.
Mass production depends on uniformity. Since the worldwide adoption of the mass production model the goal
of manufacturing has been accurate repeatability. This has had a number of beneficial effects. Mass production
has made desirable objects affordable. The size of the market allows levels of design development and the use
of sophisticated processes not possible at lower volumes.
There is however a perception that something has been lost. In today’s consumer world we are surrounded by
every conceivable product for every possible application, all at affordable prices. This availability and the
omnipresence of mass-merchandise fosters within us a desire for something personal and unique, something we
can imbue with a soul or character of its own. 'Future Factories' considers the automated production of one-off
pieces from organic, ever changing designs, which promotes the notion of the unique and fosters the processes
of personalisation.
MASS CUSTOMIZATION
It is perhaps pertinent here to specify what 'Future Factories' is not - and it is not 'Mass Customization'. 'Mass
Customization' can be defined as 'a delivery process through which mass-market goods and services are
individualised to satisfy a very specific customer need at an affordable price. Based on the public's growing desire
for product personalisation, it serves as the ultimate combination of "custom made" and "mass produced"' (Fu
2002: 44). The term 'Mass Customization was coined by Stan Davies in his book Future Perfect (Davies 1987).
The term is deliberately paradoxical. There are many different models for mass customization suiting different
products and market sectors. They are all however, consumer driven, and the key to mass customization remains
'modularisation and configuration. Products are "decomposed" into modular components or subsytems that can
be recombined to more nearly satisfy consumer needs.' (Crayton 2001: 78). This may be through a combination
of options as in cosmetic customization, where the consumer selects from an extensive but finite range of colours
and finishes. Alternatively the consumers may provide data on personal preferences or accurate measurements
316
Design pedagogy: basic design and academic experiences / 3
of body parts to enable the production of a ‘tailor made’ product. Consequently, examples of mass customized
products range from genuine medical 'needs' such as perfectly fitting hearing aids (Fu 2002) to desired product
differentiation in a kitchen stove or better-fitting bespoke jeans (Marsh 1997).
In contrast to mass customization, the 'Future Factories' model derives no input from the consumer. Where mass
customization consists of consumer selection and specification, 'Future Factories' allows the consumer only to
select the moment at which the process of form generation is arrested. Each artefact produced is therefore a oneoff realization of the designer's formula. It is the automated one-off production of an ever changing organic
design in a constant state of metamorphosis.
COMPUTER GENERATION OF FORM
We do not claim here that the notion of computer generation of random form is an original one in itself. The
capability of computers to add an element of random selection to any mathematical function has been long
appreciated. As computers have increased in power and speed, the capacity to randomly generate complex threedimensional forms can be seen as a logical development of that capability. Perhaps some of the best-known
computer generated forms are those resulting from the collaboration between the artist William Latham and the
mathematician and computer graphics expert Stephen Todd. Latham had developed a hand-drawn system for
generating abstract form called ‘form synth’ in which geometric forms could be added together, undergo a series
of pre-determined deformations and then join with other forms to ‘marry’ and create ‘offspring’ consisting of
complex forms bearing characteristics of both ‘parent’ forms. Using the extensive resources of IBM’s UK Scientific
Centre at Winchester, in the late 1980s Todd developed this method and joined it with elements of Richard
Dawkins' ‘Biomorph’ system (Dawkins 1993) that demonstrated the power of natural selection, in order to create
a powerful piece of software called ‘Mutator’. The detailed explanation of the workings of this system is best left
to those who created it (Todd & Latham 1992), but the end results are staggering. The system has developed a
great deal since, most notably in its widely disseminated form as the ‘organic art’ software package [figures 1-3]
(Computer Artworks 1995); yet its potential has not yet been fully realised. Hopefully the ‘Future Factories’
concept will explore a small part of this potential.
Figures 1 - 3: Organic 3D forms generated (or
created?) by Paul Atkinson using the 'Organic Art'
software package.
Reproduced with permission of William Latham
317
Basically ‘Mutator’ took Latham’s
‘form synth’ principle and expanded
it exponentially. Incredibly complex
forms would mutate and create
eight different offspring. One ‘child’
could then be selected and a new
series of mutations created from
that selection. Mutations could
then be judged as to how ‘good’ or
‘bad’ their forms were considered,
and those judgements used to
‘steer’ the next generation of
evolutionary mutation.
The similarities of the process to
natural evolution have led to
Latham being referred to as a
‘Digital Darwin’ (Cook 1996: 14-16).
The driving force behind ‘Mutator’
was the creation of art. As the
4 / Design
pedagogy: basic design and academic experiences
authors stated, ‘some artists feel that it provides a genuinely new way of working, and it has certainly led to the
creation of forms that would not have been created by other methods’ (Todd & Latham 1992: 105). Although
the resulting ‘sculptures’ were only ever intended to be seen as 2D representations of complex 3D models
presented as art in a gallery context, the principle behind it can just as easily be used to create variations on
‘usable’ forms to produce designs for ‘anything from buildings to shampoo bottles’ (Computer Artworks 2003).
DESIGN FORMULA
The creation of computer generated art has little in the way of physical constraints. The adaptation of these forms
into functional products though, obviously requires stricter control. Advances in computer added design have
brought a shift to parametric solutions as a methodology for the definition of computer models (3d designs). In
parametric design, relationships between the degrees of freedom of a model, instead of the degrees of freedom
themselves, are specified. Using parametric design software designs can be quickly manipulated, and alternate
solutions considered, simply by changing the variables, or parameters that define the product.
The 'Future Factories' designs are defined by 3d parametric models. In these models, ranges are set for certain
parameters within which values are assigned at random by the computer. The range limits, along with further
interdependent parametric relationships are imposed by the designer to maintain function and the desired
aesthetic. This leaves an organic model free to mutate within a series of interrelated parameter envelopes. Each
organic design is defined by a production formula, which can yield an infinite range of equally valid outcomes.
We are able to categorise objects in nature by the recognition of certain common patterns and proportional
relationships in spite of significant variance. 'Future Factories' aims to achieve this same balance between order
and chaos, between manufactured uniformity and individual sensibilities. It aims to develop a system for the
automated production of one-off outcomes that are at once distinctly individual and at the same time of a
recognizable design.
Two fundamental approaches to the concept of product variance in the 'Future Factories' model have been
identified in the work to date; manipulation of the core 3d form and the application to the core 3d form of a
variable feature.
Figures 4, 5: ‘Twist’ candlestick,
Lionel Theodore Dean, 2002
The ‘Twist’ candlestick is an example of
the manipulation of the core 3d form.
The candlestick has a simple structure
consisting of three legs, each of which
carries a candle. The legs are curved in
three dimensions and taper from top
to bottom. They touch, and are joined,
towards the middle of each leg to form
a stable three-legged structure.
The 'Twist' candlestick’s footprint [Figures 4, 5] is fixed, the legs being evenly spaced and at a fixed separation,
for stability. The tops of the legs are also constrained but not fully. Each top is required to remain in the same
radial plane as a foot, this is again is for stability. The height of each leg may vary, separately, between a maximum
and a minimum value. A relationship is applied to ensure an even spread of heights between the legs. This
318
Design pedagogy: basic design and academic experiences / 5
Figure 6, 7: ‘Mutant bulb’,
Lionel Theodore Dean, 2002
This lumiere design is an
example of the application
of a variable feature to a
core 3d form. The core form
is that of a bespoke
incandescent lightbulb.
relationship prevents an outcome with two legs close to maximum height and one close to the minimum, or the
reverse scenario. The only constraints on the form of the legs between top and bottom are the degree of
interference required for a joint to be made, and that the legs spiral in the same sense and in a smooth curve.
In the 'Mutant bulb' [Figure 56, 7], the light source is a series of high intensity white Light Emitting Diodes (LED's).
The LED's are mounted in the ends of ‘tentacles’ which appear to grow at random from the bulb form. The end
of each ‘tentacle’ is dimensionally constrained to accept an LED and the direction in which the LED points is
restricted to certain angles from the vertical (to avoid glare). Three distinct characters of ‘tentacle’ have been
designed;
• ‘Drops’ form like stalactites on the lower half of the bulb tapering as they ‘grow’ downwards as if under
gravity.
• ‘Tentacles’ form like drops from the lower half of the bulb, these however are able to resist gravity to an
extent, they have a tendency to curl and coil.
• ‘Risers’ form like stalagmites rising up from the upper half of the bulb. As they rise they lean out from
the bulb body and begin to curl under gravity.
These ‘Tentacle’ types appear in varying proportion and random positions over the bulb form. Each can then vary
in form based on its type.
THE 'FUTURE FACTORIES' SYSTEM
In 'Future Factories', a production system is envisaged in which the consumer is presented with a 3d digital model
of the artefact via a website. The consumer may access the website directly or through a sales outlet, at a gallery
or in a department store for example. The web site, the ‘Future Factory’ itself, would have a series of ‘production
lines’ corresponding to different products. When a particular production line is selected the user is presented with
a computer animation showing that particular product design metamorphosis within a parameter envelope
specified by the designer. At any given point the consumer may freeze the animation effectively creating a oneoff design on screen. Should the consumer wish they might then proceed with an order, in which case the
relevant digital production files (stl etc.) would be generated automatically and sent to the relevant RP production
facility. An artefact, effectively a one-off, will then be manufactured using layer additive manufacturing (rapid
prototyping) techniques. This may be achieved directly, via laser sintering in a suitable material for example, or
319
6 / Design
pedagogy: basic design and academic experiences
indirectly via the production of a single use tool or pattern. It should be pointed out that the intention is not for
the consumer to use the animation to adjust design features to their liking. The animation is changing in real time
and is outside their control (this would hopefully be part of the allure). A variant can be ‘designed’ for them and
they can choose to order it or not.
WEPROGRESS SO FAR
A small series of organic product designs have been produced and defined parametrically. Domestic interior
products, principally lighting and tableware have been considered for the project thus far. Domestic interior
products is a market well used to paying a premium for design and materials technology. Lighting and tableware
have been selected to keep the artefacts relatively small (though it should be noted that currently, the largest laser
sintering machine commercially available in the UK is 700 Xx 500 x 350mm). The designs selected are for
production in cast aluminium. They make use of the layer additive production methods to achieve complex forms
almost impossible to achieve with multiple use tooling. This necessitates the use of investment casting with wax
patterns for use in the process being produced by a layer additive process.
We are currently developing computer animations that illustrate the designs and the potential variance within
them. A new computer program is being developed to enable the generation of digital production files direct
from a selected animation frame. These STL files would then be used to directly produce wax patterns for
investment casting in aluminium
TECHNÉ
It is clear to see from the detailed description of the ‘Future Factories’ project that it represents a true convergence
of art and science; the aesthetic and the technological; between creativity and production. This integration of
the human perception of beauty; the randomly mutated, computer generation of form; and purely neutral
cutting-edge industrial production is a far more complex issue than it might at first appear. This is no simple
human/machine interface or human/computer binary opposite we are dealing with, but the very nature of 'techné
- art and science as one.
The adoption of such a design/production paradigm as the one being put forward by ‘Future Factories’ raises a
whole series of complex issues about the role of the designer. If the user makes an aesthetic judgment on a form,
the precise configuration of which has been generated by a piece of software, then who has ‘designed’ it? Is the
designer’s role in setting up the algorithms to be employed, the variables and constraints within which the
computer generates the form, and the parameter envelopes which limit the amount of variation a major
contribution to the finished artefact, or a fairly arbitrary minor consideration? The same problems were
encountered by Todd and Latham with respect to the sculptures created via the 'Mutator' code: 'Who owns the
copyright? What is copyright? The generative system? The genetic code for a final form? The computer form?
The computer image? The artwork on a gallery wall?' (Todd & Latham 1992: 210). The potential effects on the
practice of design are considerable. Whatever the future holds for design practice, there is certain to be serious
changes occuring. Obviously, the future role of the designer and where he or she fits into the design process is
one that will need to be examined closely and perhaps readdressed.
Despite the highly complex issue of defining ‘design’ per se (Micklethwaite 2000), and without wishing to enter
a huge debate about the distinctions between craft and design, consider the following definitions of the two
areas to see how the ‘Future Factories’ concept acts to blur those boundaries and distinctions. If ‘craft’ is taken
to be concerned with the conception of form leading to one-off production; and ‘design’ is taken to be
320
Design pedagogy: basic design and academic experiences / 7
concerned with the conception of form leading to the production of a specification for later large-scale
manufacture by a third party, then the distinction between a craft-person and a designer is clear. Following this
distinction, ‘Future Factories’ aims to allow the use of a designed system to select a form randomly generated by
computer software for immediate one-off production by machine. Although such a system has the capacity to
make an infinite variety of related forms, it also has the capacity to reproduce exactly the same form, to a
massively high level of accuracy, for an unspecified number of repetitions.
The ‘Future Factories’ system, then, would be seen to fit both the definition of craft, in that it allows one-off
variations in form; of design, in allowing repetitive production of the same form; or neither as the form is not
generated or conceived by the person who selects it, or even by the designer who specified the parameters within
which it was designed. In this context, the previously understood definitions of ‘craft’ and ‘design’ as discrete
processes become hopelessly blurred, intertwined, inextricable, and as a result, meaningless. Perhaps a
completely new terminology will be required to describe such a phenomenon to define its unique nature. The
impact on the understanding of design and its practice is potentially huge.
TEACHING TECHNÉ
As far as the impact on design education is concerned, every aspect of the curriculum may need to be addressed.
Working backwards from the proposed manufacturing concept, the element of taught CAD would be concerned
less with its integration with tool production and the generation of a specification in the form of a solid model
geometry or a set of tolerance drawings; and more concerned with its output as a producible entity by a direct
digital manufacturing process such as (for example) layer additive rapid prototyping.
The teaching of materials and processes for manufacture premised on mass-production would, of course, also
have to be reduced or replaced with a higher level of emphasis given to complex organic forms and
correspondingly suitable content to support such new digital technologies.
It is possible that even the visualisation skills taught will be affected if, as is entirely plausible, the final production
techniques employed influence the conception of forms in the design process. To what extent are our current
designers’ forms for products influenced by the ease of manufacture in injection moulded components?
And perhaps, going right back to the starting point of project briefs – how many are based on the premise of a
particular product to be produced in quantity for a certain age/gender/lifestyle or so on? There will almost
certainly need to be a move from the ‘accepted wisdom’ of market research in its constant quest for a series of
common denominators aiming to produce a product profile to fit the largest possible group of people. There will
be a need, perhaps, to consider in far more depth the user needs of individuals – more attention paid to personal
preference, the celebration of diversity over convergence. If the notion of the brand ethos were to continue in
this scenario, surely it would have to move further into the realm of communicating individual meaning rather
than ‘lifestyle’ or ‘brand values’ of a group of people, sharing some manufactured and marketed heterogeneity.
Certainly there would at least be a requirement for learning more about ‘people’ and less about ‘markets’ – more
about the choices people make about objects and the emotional relationships people have with them. In short,
less materials, more material culture.
321
8 / Design
pedagogy: basic design and academic experiences
CONCLUSIONS
At the time of writing, the ‘Future Factories’ project can be seen to have been a resounding success both in terms
of a practice-based research project, and as an exercise in design pedagogy. The next stage in the project will be
to expand the range of different items produced, and create more examples of products arising from each design
formula. These will be exhibited in a number of gallery settings as individual or touring exhibitions. Further
dissemination of the work produced will initially be done through the creation of a virtual gallery within the
research section of the School website, and the production of a CD-Rom of the work is a distinct possibility.
The project has demonstrated that the potential of computer-generated organic forms to produce viable artefacts
for one-off production, hinted at by the creators of the original ‘mutator’ code is at last a realistic proposition.
The outputs from ‘Future Factories’ will be used as evidence to support bids for external funding to develop the
required software further, and to purchase the hardware required to realise and trial the production of finished
artefacts in-house.
The ‘one design fits all’ paradigm of modernist mass production may well have been expanded through the use
of interchangeable components to allow for more variation on finishes and textures of standard products. The
sheer range of models produced of most consumer products points towards confirmation of the ‘myth’ of mass
production where the number of identical products is minimal. However, in most cases, the differences are
superficial, and limited to the surface features of products. The economy of scale in producing similar products
still holds sway – for all the variety of any model of mobile phones for example, the internal components remain
standardised, and manufacturing technology is such that at least some elements of the political ideologies of
modernism remain. The phone may look different, but its capabilities as a functional product are dictated by the
manufacturer.
Obviously, ‘Future Factories’ is not a suitable model for the production of complex technological objects (at least
not yet). But the design thinking behind it, and the manufacturing system proposed fits far more comfortably
within the tenets of post modernism, and the drive for individuality associated with that philosophy.
As a piece of pedagogic research, the experience of having a designer-in-residence in an educational setting
needs to be analysed and reflected upon, and the benefits disseminated through publication in journal articles
and conference papers, and possibly in book form. It is clear, however, that the impact on design education of
teaching techné is potentially huge.
Even at this stage of the project, graduate designers will leave the University of Huddersfield with a far wider
perspective on the nature of their design discipline, and an insight into the possibilities technology holds for the
application of craft ideologies to the design process. The project is certainly indicative of future changes that may
well be required in design education.
322
Design pedagogy: basic design and academic experiences / 9
REFERENCES
Computer Artworks (1995) Organic Art software, GT Interactive http://www.artworks.co.uk
Computer Artworks (2003) http://www.artworks.co.uk/index2.htm accessed 27 January 2003
Cook D (1996) ‘Digital Darwin’ in Creative Technology, Feb 1996 pp 14-19
Crayton, T (2001) 'The Design Implications of Mass Customisation' in Architectural Design, April 2001 pp 74-81
Davies, S (1987) Future Perfect, New York, Addison-Wesley
Dawkins, R (1993) Blind Watchmaker software, W.W.Norton And Co.
Fu, P (2002) 'Custom Manufacturing: from engineering evolution to manufacturing a revolution' in Time
Compression Technologies, April 2002 10/2 pp 44-48
Marsh, P (1997) 'Making to Measure' in Design, Spring 1997 pp 32-37
Micklethwaite, P (2000) ‘Conceptions of Design in the community of Design Stakeholders’ in Scrivener, S, Ball, L
and Woodcock, A (Ed.) CoDesigning 2000: Adjunct Proceedings Coventry, Coventry University pp 93-98
Todd S & Latham W (1992) Evolutionary Art and Computers, Academic Press Ltd, London
323
324
Appendix 4
325
Appendix 4
326
Appendix 4
327
Appendix 4
328
Appendix 4
329
330
Appendix 5
INTERNATIONAL DESIGN CONFERENCE - DESIGN 2004
Dubrovnik, May 18 - 21, 2004.
Title: ‘Future Factories’: Supportive technologies as
creative processes
© Atkinson, P., Unver, E. & Dean, L.T. 2004
Keywords: 3D Generative processes, Rapid Prototyping, Direct
Digital Manufacture
1. Introduction
Envisage a future where you could visit a dedicated 'Future Factories' website, accessed in a gallery, a
department store or directly from your own home. This website would display a range of products, and
you could select any one of them. Once you made a choice, you would be presented with an
animation, which would show that particular product in a constant state of metamorphosis, as it grows,
changes and mutates on the screen. At any given moment you could pause the animation and view a
three-dimensional computer model of the product, rotating it to see it from any angle. The animation
would continue, and any number of ‘snapshots’ of the product at various stages of its growth could be
taken, and if required, printed onto paper for closer examination. Each one of these ‘snapshots’ would
be a unique form – never to be repeated. If you then decided to purchase one of the designs, you could
order it directly from the website, and the product you had selected would be manufactured
automatically, exactly as you had seen it on screen, and delivered directly to your door. An original. A
one-off. A work of art?
The School of Design Technology at the University of Huddersfield recently decided to allocate an
amount of research funding to provide an ‘Artist-in-Residence’ to work alongside Fine Art students,
and a ‘Designer-in-Residence’ to work alongside Product and Transport design students for a period
of one year. The work undertaken by the Designer in Residence along with contributions made by
other academic staff are the subject of this paper. The title of the project ‘Future Factories’ describes
an exploration of the creative potential inherent in digital design and manufacture to offer more than a
single discrete 3D outcome. The outputs from this practice-based research project consist of a number
of inspirational products which have been exhibited in a number of traditional gallery environments
and will be later disseminated digitally – either on-line or by CD-ROM. Alongside the practice-based
research outputs there have been a number of different academic papers addressing the different
technical, theoretical and contextual issues raised by the content of the ‘Future Factories’ project.
2. Methodology
'Future Factories' is an exploration of the possibilities for flexibility in the manufacture of artefacts
inherent in digitally driven production techniques. All such production techniques are considered, the
focus however is on the layer additive manufacturing techniques associated with Rapid Prototyping
(RP). In essence the project proposes an inversion of the mass production paradigm to one of masscustomisation. However, unlike previous work on mass-customisation, where many design decisions
331
are taken by the consumer or with reference to the particular consumers needs, ‘Future Factories’
considers individualised production – in which a random element of variance is introduced by the
computer within a parameter envelope defined by the designer. Each artefact physically produced will
be a one-off variant of an organic design that has been defined by the designer and maintained in a
constant state of metamorphosis by the computer software. This variance may be over parameters such
as the relative positioning of features, scale, proportion, surface texture, pattern, and the like. These
variable factors may be multiple and interrelated. The intention is to achieve subtly different aesthetics
based on a central theme rather than mere differentiation that might be achieved by say scale or colour
change alone. This random variance would simulate the lack of uniformity in one-off craft production
where the craftsperson may be guided by a design intent rather than a toleranced production drawing.
In this way, ‘Future Factories’ aims to overcome the split between the technological and the aesthetic,
between artistic creativity and machine production
The creation of computer generated art has little in the way of physical constraints. The adaptation of
these forms into functional products though, obviously requires stricter control. Advances in computer
added design have brought a shift to parametric solutions as a methodology for the definition of
computer models (3d designs). In parametric design, relationships between the degrees of freedom of a
model, instead of the degrees of freedom themselves, are specified. Using parametric design software
designs can be quickly manipulated, and alternate solutions considered, simply by changing the
variables, or parameters that define the product.
The 'Future Factories' designs are defined by 3d parametric models. In these models, ranges are set for
certain parameters within which values are assigned at random by the computer. The range limits,
along with further interdependent parametric relationships are imposed by the designer to maintain
function and the desired aesthetic. This leaves an organic model free to mutate within a series of
interrelated parameter envelopes. Each organic design is defined by a production formula, which can
yield an infinite range of equally valid outcomes. We are able to categorise objects in nature by the
recognition of certain common patterns and proportional relationships in spite of significant variance.
'Future Factories' aims to achieve this same balance between order and chaos, between manufactured
uniformity and individual sensibilities. It aims to develop a system for the automated production of
one-off outcomes that are at once distinctly individual and at the same time of a recognizable design.
Two fundamental approaches to the concept of product variance in the FF model have been identified
in the work to date; manipulation of the core 3D form and the application to the core 3D form of a
variable feature.
As an example of the first approach, a three-legged candlestick was designed, having a series of
functional requirements – to stand upright and support three candlesticks of a fixed size. The
candlestick’s footprint is fixed, the legs being evenly spaced and at a fixed separation, for stability.
The tops of the legs are also constrained but not fully. Each top is required to remain in the same radial
plane as a foot, again for stability. The height of each leg may vary, separately, between a maximum
and a minimum value. A relationship is applied to ensure an even spread of heights between the legs.
This relationship prevents an outcome with two legs close to maximum height and one close to the
minimum, or the reverse scenario. The only constraints on the form of the legs between top and
bottom are the degree of interference required for a joint to be made, and that the legs spiral in the
same sense and in a smooth curve.
As an example of the second approach, a light fitting was designed which took the existing form of a
light bulb, but with a solid metal body. Instead, the light source is a series of high intensity white Light
Emitting Diodes (LED's) mounted in the ends of ‘tentacles’ which appear to grow at random from the
bulb form. The end of each ‘tentacle’ is dimensionally constrained to accept an LED and the direction
in which the LED points is restricted to certain angles from the vertical (to avoid glare). Three distinct
characters of ‘tentacle’ have been designed;
332
Appendix 5
‘Drops’ form like stalactites on the lower half of the bulb tapering as they ‘grow’ downwards as if
under gravity.
‘Tentacles’ form like drops from the lower half of the bulb, these however are able to resist gravity to
an extent, they have a tendency to curl and coil.
‘Risers’ form like stalagmites rising from the upper half of the bulb. As they rise they lean out from
the bulb body and begin to curl under gravity.
These ‘Tentacle’ types appear in varying proportion and random positions over the bulb form. Each
can then vary in form based on its type.
3. Virtual Merchandising
Conventional marketing usually centres around a glossy photograph of the product shown from its
best angle. VR content enables websites to bring the products 'to life' with 3D models, user
interactivity, animation, sound, and detailed views. An interactive image allows the product to be seen
from all angles. Potential customers are able to examine the design in the form of a 3D model moving,
rotating, and zooming in and out at will. This ‘hands-on’ interaction allows something of a “try before
you buy” experience. Internet shoppers have been reported to spend 50% more time in the part of the
site that offers interactive 3D images, yet VR on the Web is not yet mainstream or widespread. Why
has this exciting technology made such slow progress? This may be due to the fact that VR content is
costly to develop, mostly because the expertise to create it is still rare, and a direct link to sales
revenue is as yet unproven. In addition, content creators typically make a substantial investment in
computers, software, and digital equipment. Technical difficulties have discouraged the take up of
website based VR marketing. The technology is most convincing and pleasing when it uses realistic
textures, lighting and sounds. The use of these elements requires large files which leads to slow
performance, and web designers fight a constant battle between high graphical appeal and slow
download times. High resolution graphics and elaborate animations are notoriously slow to download,
especially through a dial-up connection. Add to that bandwidth limitations and the possibly unreliable
connections of the Web, and you have the potential for a deeply dissatisfying experience.
We can assume that consumers will always prefer hands-on experience with a product before
purchase. Barring cumbersome and expensive gloves and goggles, VR is still strictly an audiovisual
experience. Even gloves have serious limitations in that they cannot provide a tactile experience of
texture. This might be solved if the website access was via a retail outlet and samples were available. It
would also avoid ‘user end’ technical issues. VR content cannot be viewed with a standard browser - a
special-purpose browser or a plug-in is required. In addition there is no single viewer that can handle
all VR content. The requirement to download additional software merely to, in effect, browse a
shopping catalogue is a severe disincentive. The user may not even have administrator rights to the
computer or the desire to involve themselves in IT issues. The lack of an industry standard also affects
content creators (and their cost), as each viewer typically has its own authoring tool. There is no single
tool that a programmer can learn with any expectation of addressing more than a fraction of Web
users. The limitations of computers and the technological demands of VR should not be exaggerated
however. VR files are not necessarily huge. Files consisting mostly of vector graphics are relatively
small. Problems of speed and resolution will be solved over time as standard-issue desktop computers
gain speed and are optimized for 3D graphics. Higher-bandwidth and more reliable connections to the
Internet will also become more common. Proprietary viewer and content creation software offer
increasingly high-quality images within compact, efficient files.
4. 3D Modelling and Software Development
333
Figure 1. Organic designs created in Future Factories Software
Initially in order to test user interaction and finding out what each individual person choose during the
exhibition, a new software FF (Future Factories) is developed. The software written in Delphi
Language enabled user to choose, stop, replay, print a 2D picture as well as capturing the selected
design parameters and creating a database.
In this study a series of organic product designs shown in Figure 1 were created using this software.
Domestic interior products, principally lighting and tableware, have been considered for the project
thus far. Domestic interior products is a market well used to paying a premium for design and
materials technology. Lighting and tableware have been selected to keep the artefacts relatively small.
This consideration is based on cost rather than capacity - the largest laser sintering machine
commercially available in the UK is 700 x 500 x 350mm for manufacture in one piece, and building in
sections could also be considered. The designs selected thus far are for production in cast metal. They
make use of layer additive production methods to achieve complex forms almost impossible to achieve
with multiple use tooling. This necessitates the use of investment casting, with the wax patterns for use
in the process being produced by a layer additive process.
The software’s database enables us to see when each design parameter is selected and if there are any
correlations between similar chosen parameters. In future, this facility could provide data on the
preferences of different genders and age groups or other criteria.
In 'Future Factories', a production system is envisaged in which the consumer is presented with a 3D
digital model of the artefact via a website. The web site, the ‘Future Factory’ itself, would have a
series of ‘production lines’ corresponding to different products. When a particular production line is
selected the user is presented with a computer animation showing that particular product design in
metamorphosis within a parameter envelope specified by the designer. At any given point the
consumer may freeze the animation effectively creating a one-off design on screen. Should the
consumer wish they might then proceed with an order, in which case the relevant digital production
files (stl etc.) would be generated automatically and sent to the relevant RP production facility. An
artefact, effectively a one-off, will then be manufactured using layer additive manufacturing (rapid
prototyping) techniques. This may be achieved directly, via laser sintering in a suitable material for
example, or indirectly via the production of a single use tool or pattern. It should be pointed out that
the intention is not for the consumer to use the animation to adjust design features to their liking. The
animation is changing in real time and is outside their control (this would hopefully be part of the
allure). A variant can be ‘designed’ for them and they can choose to order it or not.
To add to this allure of one-off products, there are a number of ways in which the ‘value’ of the
artefacts produced might be increased. An element of exclusivity can be introduced for customers such
as corporate buyers, for whom specific commissions could be undertaken and unique design formulas
334
Appendix 5
produced. They could then order as many of the objects (such as light fittings for a particular chain of
restaurants) as they required, secure in the knowledge that each product would be unique in itself as
well as the design formula being unique to them.
Alternatively, the production of designs can automatically be ‘capped’ to a specified quantity as is the
case, for example, with limited edition screen prints, with a numbered system being used to show how
many have been produced, and how many opportunities to own a one-off variant of a particular design
are left. Another option is not to cap the quantity, but to limit the amount of time for which any
particular product will be produced.
Perhaps the most interesting possibility for increasing value is to employ the model of a single line of
‘evolutionary’ development in which a design is created, adapted and finished over a specified time
span. Imagine a simple design being created for production for a period of, say, six months. Over that
period, the design might become more and more complex, more organic, or more convoluted in form
until it reached the end of its ‘growth’ pattern when it would no longer be able to be turned into a real
object. At any point during that period, customers could view how the object started out and how it
has developed since its inception. They could have the option of purchasing the object at that point
(but not be able to purchase any of the forms from a previous time), or anticipate, like gamblers
playing a game of chance, how the design might look in a month, when they might return and
purchase it. They might plan to purchase a range of objects from a number of different points in its
existence, or vectors along the animated production line. It is possible that ‘early’ incarnations of the
design could become more valuable than later ones (as with limited edition screenprints having lower
imprint numbers). The possible combinations of ways in which the process could be employed are
potentially huge and are currently forming the basis of a funded attempt to commercialise the
technology developed.
5. Conclusions
The ’Future Factories’ project demonstrates that the potential of computer-generated organic forms to
produce viable artefacts for one-off production is at last a realistic proposition. Obviously, ‘Future
Factories’ is not a suitable model for the production of complex technological objects (at least not yet).
But the design thinking behind it, and the manufacturing system proposed fits comfortably with
today’s drive for individuality.
In ’Future Factories’ a direct connection is made between playful desires and the will to take risks
through predictive forecasting, as well as connecting with dominant modes of capitalist production via
the technologies if not the processes of mass production. ’Future Factories’ is not mass customisation,
the mode of production is craft placed momentarily in the hands of the consumer, temporarily
liberating them by engaging them in a culture of chance, variability, selection and playfulness. This
enables the consumer to engage with a plethora of possibilities through chance decisions that
ultimately capture a particular moment, through which a unique object is cast out from a virtual
environment into the real world.
It is clear that the implications of the wide scale adoption of such techniques by industry are
potentially serious, and are such that moves to protect the process via patents have been made. The
system has the potential to change the perception of design by consumers and manufacturers alike, and
to influence considerably the education and training of designers. Despite the philosophical questions
the process raises for the definitions of terms such as ’design' and 'designer', (which are potentially
misleading in this context), and the scope for confusion as to whether the end results of the process are
'art', 'craft', or 'computer generated', there are a number of more pragmatic considerations. The
potential for the process to impact on manufacturing and retail industries should not be overlooked.
'Future Factories' allows for the economic large-scale production of artefacts while providing
335
important reductions in wastage arising from the over-production of unwanted items. At the same time
the system promotes the move from reductive to additive manufacturing processes, cutting down on
waste material from the production of goods. As such, it may point the way to a more sustainable
model of a consumer society than the one we take for granted today.
6. References:
Cook D (1996) ‘Digital Darwin’ in Creative Technology, Feb 1996 pp 14-19
Crayton, T (2001) 'The Design Implications of Mass Customisation' in Architectural Design, April
2001 pp 74-81
Davies, S (1987) Future Perfect, New York, Addison-Wesley
Fu, P (2002) 'Custom Manufacturing: from engineering evolution to manufacturing a revolution' in
Time Compression Technologies, April 2002 10/2 pp 44-48
Todd S & Latham W (1992) Evolutionary Art and Computers, Academic Press Ltd, London
336
Time Compressions Technologies Conference, United Kingdom – 2006
Appendix 6
Abstract
FutureFactories is a design project with the ultimate aim of massindividualisation: the industrial scale production of one-off artefacts. This is to be
achieved by the combination of genetic algorithms, parametric CAD, and Direct
Manufacture. Mass individualisation itself remains a blue-skies goal for the
project (although it need not be that far away). Rapid Manufacture however, is
already with us, and products developed by FutureFactories are currently on
retail sale around the world. In April 2006 there was a significant development:
Italian lighting and furniture manufacturer Kundalini launched, what they are
billing as, the first Rapid Manufactured retail product by a recognised
manufacturer (in other words not by an RP bureau). Prophesies such as this are
dangerous, given that this is a fast developing field: it is certain however, that
Kundalini will be amongst the first.
The significance is that Kundalini’s primary function is retail manufacturing.
Rapid Manufacture could be considered as a logical step for the larger service
bureau, unused machine capacity can be turned to production, and the artefacts
produced serve as marketing tools promoting the capacity, technology, and
expertise, of the company. The products effectively become larger scale
versions of the samples produced by machine vendors, only in this instance
saleable.
For Rapid Manufacture to become ‘mainstream’ it must be adopted
by those who have no vested interest in promoting the production process itself.
The author has long argued that FutureFactories is not about the technologies
themselves, but about their application, and the creative opportunities that they
facilitate.
From the onset of the Kundalini project, the client was dispassionate about
process and entirely focused on form. The brief specified a product that would
baffle, whose conventional production would be inconceivable. This departure
from the predictable would apply not only in terms of physical manufacture, but in
the very nature of the geometry produced. There should be no pattern or order,
no hint of logic that would suggest how the form had been created. For Kundalini
this could be the only justification for the relative expense of direct manufacture.
This paper will discuss, through the Kundalini case study, the implications of
direct manufacture from a design perspective. It will look at the shift in new
product investment from physical tooling to the design of increasing complex
objects. From a Direct Manufacturing point of view, geometry may ‘come free’,
but this in turn will set up new market demands and expectations. The designer
will need new skills, tools, and understanding of the capabilities of Rapid
Prototyping, in order to realise the potential of Direct Manufacture in the
consumer products market.
Introduction
Entropia was launched at Light and Building, Frankfurt, April 2006 by Italian
lighting and furniture manufacturer Kundalini. Kundalini celebrated its tenth year
337
Time Compressions Technologies Conference, United Kingdom – 2006
Appendix 6
in 2006 and has a reputation for blending tradition and technology, mixing hand
blown glass with 5-axis waterjet cutting for example.
Entropia’s principle component is a 120mm diameter spherical diffuser produced
in laser sintered nylon. The design is available in table, suspension, and wall
variants. It retails between 400 and 500 euro depending on the model, a price
comparable with traditionally manufactured artefacts from design-led
manufacturers in materials such as hand blown glass and ceramic. Kundalini
had no previous experience of Rapid Manufacture or even Rapid Prototyping.
They were however familiar with the work of FutureFactories. The company had
previously expressed interest in FutureFactories’ aesthetics, but until recently
had considered that the time was not yet right for the technology to be used in a
retail context. The plethora of R.P. based concept work seen around the world
and the promotion of Direct Manufacture by the R.P. industry began to modify
that view point. In autumn 2005 it was decided that the idea needed serious
consideration and the author was commissioned to create a series of concepts.
At the same time talks began with RP service bureaus and equipment vendors to
explore the economic viability.
From the onset the client was extremely passionate about the aesthetics that
might be achieved but never at the expense of commercial viability. This was to
be a retail product the price of which would be determined by the market. It was
clear from early investigations that the best chance of viability lay in a compact
form and that the more efficiently the build chamber was filled, the more cost
effective the process would become. The cost of a build would be split between
the number of components contained within it. Squeezing in one or two extra
units would have a significant impact on price and may prove to be the difference
between viability or otherwise. Early in the concept design stage a spherical form
emerged as the most likely solution as the design needed to be as compact as
possible, and leave a certain clearance around the light source for reasons of
temperature. A diameter of 120mm was arrived at as offering the best balance of
perceived value and manufacturing cost. Despite this compact nature the design
would use G9 halogen fittings, these run at line voltage, eliminating the cost of a
transformer.
Form
It was important to the client that the design was taken as far from conventional
industrial manufacture as possible. Complexity was a given, any regular form
could be produced more economically by conventional means. The idea was
however, to go beyond awkward geometry. Although certain forms may be
impossible to produce conventionally, undercuts, re-entrant shapes, and the like,
this fact will not necessarily be appreciated by the lay customer. To the
consumer it makes little difference if a product is produced in one piece via some
exotic means or is made as a well disguised assembly of cheaper components.
The freedom of Rapid Manufacture brings the risk of engaging in party tricks that
are only appreciated within the industry itself. The author recently showed a lay
338
Time Compressions Technologies Conference, United Kingdom – 2006
Appendix 6
audience a sample of SLS manufactured chain link. The audience were
unimpressed, plastic chain link could be bought from the local DIY superstore.
The aim with this project was to create a form that would intrigue, baffle, and
captivate everyone. It would be necessary to convince the buying public that a
particular piece of plastic was every bit as valuable, if not more so, than for
example, a piece of hand blown Murano glass that would sit beside it in the
Kundalini collection. The idea was to remove all traces of pattern and logic from
the form, to eliminate any accessibility or key to understanding. The principal of
this was easy to grasp, achieving it in practice less so. The language of
traditional design-for-manufacture had to be abandoned. There could be no
repeats or symmetry. At the same time it needed to be evident that there was
process behind the form, a totally random assembly would not be considered
particularly valuable. The solution was to adopt the rule based approach used in
previous FutureFactories morphing designs. In these designs virtual models
were created that allow a design to morph within a parameter envelope set by
the designer. Rather than yield a discrete 3D solution these designs were
templates from which multiple one-off solutions could be generated, each
functional and true to the designers’ intent. In the case of Entropia, design
templates were created for a series of features that would appear in varying
numbers throughout the form. These templates dictate the underlying style of a
specific feature but allow considerable flexibility in it’s particular embodiment.
Each time the feature is repeated within the form there is a slightly different
outcome. The result is the impression of a natural phenomenon, such as coral.
There are clear patterns to the ‘growth’ but the form appears to have evolved
rather than to have been constructed.
Rapidity
Product development speed was not a primary consideration in this project.
Speed is more applicable to the prototyping side of digital manufacture rather
than series production. The emphasis placed on rapidity by service bureaus
which is such an asset to prototyping, is often a hurdle to Rapid Manufacture.
The bureau industry is often characterised by feast/famine workloads and large
amounts of overtime. This does not lend itself to rapid manufacture where
efficiency is key and profits are made over a longer term.
Whilst not a primary consideration Rapid Manufacture enabled an extremely
short product development cycle. The concept was agreed in early January 2006
and the product launched in April 2006. In spite of the designs compact nature,
the level of complexity in the form required a lot of 3D modelling. Digital
manufacture allowed prototypes to be built based on sections of the design that
were complete, incomplete sections were either replaced with simple
approximations or simply omitted. This enabled technical development to take
place in parallel with the design of the form itself and allowed the compressed
timescale.
Conclusions
339
Time Compressions Technologies Conference, United Kingdom – 2006
Appendix 6
Complexity
Aesthetics are subjective and the merits of this particular design are not an issue
for this paper. A significant factor however is the demand for complexity, a
demand coming from the client but driven by the consumer. This demand is sure
to increase as the almost free-form potential of Rapid Manufacture becomes
more widely appreciated. Greater use of computation in the design process is
called for to manage this complexity.
Visualisation
Visualisation is a major issue. It is difficult to orientate ones self in a complex onscreen model. Locating a specific area in a form as complex as Entropia is far
from easy. It becomes hard to identify issues that would be readily apparent
when handling an actual prototype. A far greater range of more flexible
visualisation tools built into CAD software systems would help here.
Visualisation is also an issue for the consumer where there is on sample product
to assess. FutureFactories is frequently asked where designs, published on the
website, can be seen in the flesh. If only a few artefacts are made to satisfy a
market niche it would not be economically viable for stock to be held at retail
outlets worldwide, manufacturing to order would be more appropriate.
Visualisation beyond the brochure studio photograph is required to convince all
but the most avante-guard of buyers, better interfaces are needed to enable
consumers to assess products remotely and order them with confidence. Some
form of web-based virtual reality, computer-based experience that mimics real
experience, may provide a solution. It is perhaps the high-end VR system that
come first to mind with the user wearing special goggles and gloves. At its
simplest however, VR could be a 3D view of the product that the user can rotate
to see from various angles. In this way potential customers would be able to
examine the 3D model moving, rotating, and zooming in and out, at will. Such
‘hands-on’ interaction would allow something of a “try before you buy”
experience.
Investment
It is worth noting that although Rapid Manufacture eliminates the need for tooling
investment, current marketing practices require investment in printed catalogues
and the like. As a designer, it is tempting to believe that manufacturing will
become far more flexible and versatile, able to experiment and react quickly to
niche markets. In practice it is an attitude that will take time to develop within
industry and consumers’ behaviour will also need to adapt.
The lack of tooling investment removes a key point from the design process; the
point at which production drawings would be passed to a toolmaker. With Rapid
Manufacture design development can continue up to the eve of product launch.
Need development cease at product launch? A situation can be envisaged
340
Time Compressions Technologies Conference, United Kingdom – 2006
Appendix 6
similar to software releases with a product improving by degrees during a
production run, versions 1.0, 1.1, 1.2 etc.
Process
The traditional product design development model begins with loose concept
sketches that gradually become more and more refined as the project develops.
Practicalities are fed in gradually so as not to kill the style of a concept. When
computational design is employed the initial priority is in setting up relationships,
parameters, and formula, rather that looking ahead to what they will ultimately
produce. Complexity by its very nature cannot be easily reduced or predicted.
With Entropia the importance was in defining a language for the elements used.
The initial design issue was the definition of developmental rules that would allow
diversity yet keep the form true to the designer’s intent. Creating such rule-based
systems takes time and requires something of a leap of faith on the part of those
commissioning the process. It is not possible to generate outcomes before the
system is established. Once the rules are in place however, any number of
design iterations can be generated with relative ease. It is easy to imagine a step
on from the current Entropia in which only the size and developmental rules are
pre-defined. A unique form would be ‘designed’ every time a 3D iteration was
generated, this would be the FutureFactories concept of mass-individualisation.
The only issues hold back such an idea are the demands of setting up the model
and customer acceptance. A virtual model capable of generating endless one-off
outcomes must be created, a process that must be automated as, in industrial
scale manufacture, a designer cannot modify each and every item produced.
Consumers would need to be aware of the concept and celebrate difference,
manufacturers considering individualisation fear endless product returns as the
product received does not match up to another.
Optimisation
Currently, the placement and arrangement of models within the build volume is
undertaken by RP bureau staff who are presented with the model when it is
complete or in the final stages of development. The efficient use of the build
volume is the most important factor in determining the viability or otherwise of
Rapid Manufacturing projects. It is vital that the concept is tailored from the
concept stage to use the chamber efficiently and to allow the accommodation of
a commercially viable number of units. Ideally, software for this function, would
be brought within high end CAD packages so that it can be referred to by
designers throughout the design process rather than existing as a stand-alone
package aimed at machine operators.
Material
In Rapid Manufacture for the decorative design market, public perceptions of
material value becomes an issue. Plastics however exotic do not have the same
cache as, for example, glass or ceramic. Hand blown glass artefacts sit beside
Entropia in the Kundalini collection and are comparable in price. This perception
of inferiority can be combated to some extent via design. Fine detailing and
341
Time Compressions Technologies Conference, United Kingdom – 2006
Appendix 6
delicacy of proportion communicate quality and craftsmanship. This becomes
exaggerated in lighting applications where translucency accentuates fine section
thicknesses.
In the laser sintered designs of FutureFactories, thicknesses can be as low as
0.5mm. This is well below that which would be recommended for the process.
Many bureaus give a standard warning if model section thicknesses drop below
2mm, Entropia does not have any sections above 2mm. In fairness, robust
sections are encouraged because of the cost of the process and the risk of
scrapping parts. FutureFactories plays with thickness ensuring that there will
always be an appropriate structure. Ultra-fine sections are used in areas that are
purely decorative and forms are preferred that will not suffer unduly should fine
edges not build completely. Whilst it has not been used as a policy in the output
of FutureFactories, redundancy could be considered to mitigate the risk of failure
in minor decorative elements. Where there are massed numbers of features
following no discernible pattern the absence of one or two may pass unnoticed
should they fail.
342
Time Compressions Technologies Conference, United Kingdom – 2006
Appendix 6
Early FutureFactories work featured hand finished (paint finish) artefacts built
using budget 3D printing processes. It soon became clear however that the
budget RP processes were still expensive in relative terms and that hand
finishing on top of a costly process was unlikely to prove viable. Hand finishing
also went against the principle of automated manufacture. FutureFactories has
fastened upon SLS nylon as currently the best option in terms of the balance
between cost (process), mechanical performance, and appearance. New
materials are appearing all the time many of these have caught FutureFactories
eye in terms of their beauty, in particular ceramic-like epoxies with vivid colour.
RP materials development however has been focused on providing high
mechanical performance rather than cost. This tends to make them unattractive
for Rapid Manufacture. We will have to see if embryonic Rapid Manufacturing
stimulates a drive for cost effective decorative materials.
343
Time Compressions Technologies Conference, United Kingdom – 2006
344
Appendix 6
Appendix 7
345
Appendix 7
346
Appendix 7
347
Appendix 7
348
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement