Addendum 44 rev 1 Toshiba IK-HR1 mini camera

Colorimetric and Resolution requirements of cameras
Alan Roberts
ADDENDUM 44 rev 1: Tests and Settings on a Toshiba IK-HR1 minicamera
Data for this section is taken from the handbook and a very brief examination of Toshiba IK-HR1S and IKHR1D mini-cameras as part of a group test of HDTV mini-cameras. The cameras are self-contained, there is
no separate control unit, all controls are on the camera.
The cameras are small (44x44x78mm) and weigh only 146 grammes. The specification states that they have
a single CMOS sensor of 2.1Megapixels, which must be Bayer-patterned and is probably 1920x1080 pixel
count. The lens mount is the mini-cam standard C mount. Sensitivity is claimed to be F/4 at 2000lux, which
is typical of single-sensor 1920x1080 cameras with ⅓” sensor. The HR1S has only HSDSDI output, the
HR1D has only DVI output which also carries analogue outputs as VGA. There are menus, allowing some
rudimentary image control, which have great similarity to those in the IK-HD1 mini-camera.
Power consumption is 3.5 watts at 12V DC for the HR1S, 4.2 watts for the HR1D.
Tests were made only on the HR1S, since the DVI output of the HR1D is difficult to use in a broadcast
environment. However, the HR1D does provide output at 1080p/59.94, and so could be useful under special
circumstances. The differences between the cameras are small, and the measurement results should apply
equally to both cameras. Differences between them are noted in the menus.
Unfortunately, the cameras show significant response to infra-red illumination.
This revision corrects minor errors which do not affect the measurements or conclusions.
1
Colorimetric and Resolution requirements of cameras
Alan Roberts
ADDENDUM 44 rev 1: Tests and Settings on a Toshiba IK-HR1 minicamera
Many of the menu items have little or no effect on image quality. Those that have significant effect are
highlighted. The full set of menu items is given for completeness. In boxes with a range of numeric settings,
e.g. -99~+99, the values indicate the range, and zero means no alteration to factory setting, not zero effect,
and no scales are given in the manuals. For each item, the factory setting is underlined where known. “BBC”
recommended settings are in the last column, where appropriate. In some instances, it is possible to alter the
menus such that they produce more meaningful numbers.
Settings have been derived and are shown in the “BBC” column. Although the camera has all the options for
interlaced and progressive shooting, no attempt has been made to derive a ‘film-look’ for it, since the menus
do not allow sufficient control over the gamma curve to make it worthwhile.
Settings are only starting points, recommendations. They should not be used rigidly, they are starting points
for further exploration. However, they do return acceptable image performance.
Measurement results are given in section 2, after the menus.
This listing of the menus and contents is complete, but this should not be used as an excuse for not reading
the manuals.
Auto Black balance can be carried out directly from a button on the camera, and the output format
(1080/720, 50/59.94) is set by slide switches on the HR1S. The output format is set differently on the
HR1D, which runs only at 59.94Hz: hold down the DATA UP button while powering up to select 1080p,
hold down the DATA DOWN button while powering up to select 1080i. This can also be done from menu 6
OPTION. Note that the HR1D really can output 1080p/59.94, and not psf.
There are 5 scene files, selectable directly from the buttons. All other camera control is via the menus, the
contents of which form the scene file.
2
1
Menu items
1 SHUTTER
item
Mode
Manual
SS
Level
Peak/ave
Speed
Area
range
comment
Auto, Manual, SS
Auto links shutter to iris and gain.
HR1S: Off, 1/100, 1/125, 1/250,
1/500, 1/1000, 1/2000, 1/4000
Manual:
HR1D: Off, 1/100, 1/125, 1/250,
1/500, 1/1000, 1/2000, 1/4000,
1/8000, 1/16000, 1/32000
HR1S 1080: Off, 15/1125~1123/1125
Synchro-scan: set shutter in line increments
HR1S 720: Off, 10/750~748/750
HR1D: Off, 2/1125~1123/1125
-100~0~+100
Auto: set average video aim level
00:10~05:05~10:00
Auto:
1~10~20
Auto: response speed
Auto: A=full frame, B=diamond, C=small rectangle,
Preset A, Preset B, Preset
C, Preset D, Preset E
D=mid column, E=bottom row
BBC
2 GAIN
item
Mode
Manual
range
Manual, Off
HR1S: 0~18dB
HR1D: 0~12dB
comment
Off=fixed gain, 0dB
Manual: fix in 1dB steps
3 WHT BAL
item
Mode
R paint
B paint
C. temp
Area
R paint
B paint
R gain
B gain
C. temp
range
AWB, ATW, Manual
-10~0~+10
-10~0~+10
3200, 5600K
Preset A, Preset B, Preset
C, Preset D, Preset E
-10~0~+10
-10~0~+10
-100~0~+100
-100~0~+100
3200, 5600K
4 PROCESS
item
Gamma on/off
Gamma
DTL gain
DTL B.freq
M. ped
range
On, Off
-10~0~+10
-7~0~+7
High, Normal, Low
-128~0~+127
BBC
comment
White balance
BBC
AWB: Red offset
AWB: Blue offset
AWB: Only two settings, equivalent to optical filter
AWB: Active area, as for shutter
ATW: Red offset
ATW: Blue offset
Manual: Red gain
Manual: Blue gain
Manual: colour temperature
General controls
BBC
On
No clue in the manual as to what the numbers mean
0
Ditto
-3
Boost frequency
High
comment
5 MATRIX
item
Matrix
R Hue/Gain
G Hue/Gain
B Hue/Gain
Ye Hue/Gain
Cy Hue/Gain
Mg Hue/Gain
range
On, Off
-15~0~+15 / -15~0~+15
-15~0~+15 / -15~0~+15
-15~0~+15 / -15~0~+15
-15~0~+15 / -15~0~+15
-15~0~+15 / -15~0~+15
-15~0~+15 / -15~0~+15
comment
Simplified version of colour correction or multimatrix. Separate control for hue and gain in each 60
degree sector
BBC
On1
8, 0
0, 0
9, 0
15, 0
0, 0
15, 0
________________________________________________________________________________________________
1
This is the only way to adjust colouring.
3
6 OPTION
item
Monitor
I/P mode
Baud rate
range
PC, TV
1080p, 1080i
9600, 19200bps
comment
HR1D only
HR1D only
Control data rate
BBC
To do a factory reset, select a scene file using the FILE button, press DISP is necessary to disable colour bars, press MENU UP and
MENU DOWN together for at least a second.
4
2
Measurement results
Measurements were made with a Fujinon lens, TF4DA-8, 4mm wide angle. All measurements were made
using the HDSDI output from the HR1S; the HR1D was not specifically tested, but the results and
observations should apply equally to both cameras. Pictures were displayed on a Sony 32” grade 1 CRT
monitor, a waveform monitor, and recorded using proprietary software for analysis.
2.1
Sensitivity
Sensitivity was not measured directly. The specification claims F/4 at 2000lux (59.94Hz), equivalent to
about 100ASA with 0dB gain.
2.2
Colour performance
Using a Colorchecker chart, the colour performance was judged to be unacceptable. The skin tone has a
distinct pink hue, while red, pink and magenta are all desaturated. The yellow is distinctly green and the
oranges desaturated.
After some experiments with the matrix settings, some improvements were forthcoming, but not enough to
produce satisfactory pictures. It would be quite difficult to correct the colouring even in a full grading
operation.
Figure 1 Macbeth chart (a) native performance
(b) with matrix settings
The camera shows significant response to infra-red illumination which can seriously pollute some colours
under some illuminants.
2.3
Resolution and aliasing
All testing was done with a circular zone plate test chart having 6 sinusoidally modulated patterns. The six
patterns explore luminance and chroma channels on the top row, RGB channels on the bottom row, the
samples shown here are each only one quadrant of the luminance (grey scale) pattern. Images were captured
uncompressed from the CCU via HDSDI.
The camera was not tested in 720p, only in 1080i.
Both horizontal and vertical aliasing are clearly present, and coloured. There is also some diagonal aliasing.
These are very good indicators that the camera has a single sensor, with Bayer filter patterning, and that the
sensor pixel-count is 1920x1080. The clean resolution limits area about 1280x720, all frequencies beyond
these limits are spectrally folded and become coloured aliases. The strength of the aliases is normal for
cameras with Bayer single sensors of inadequate resolution.
Clearly, there is no optical filter to prevent high-frequencies from reaching the sensors.
Detail enhancement is a little severe, but at level -3 the aliases have not been enhanced too much. The
camera actually performs better with low levels of detail enhancement, zero is acceptable but negative values
better still. Level -7 is equal to no detail enhancement. Higher, positive levels greatly enhance the aliasing.
5
Figure 2 Zone plate (a) no detail (=-7)
2.4
(b) detail=-3
Video Noise
The specification claims the luma channel noise level to be -54dB, with factory settings (0dB gain).
Measurements were taken on an evenly lit white card, exposed at various levels. Image files were captured
via HDSDI as data files, then transcoded and decoded in software before performing a software noise
analysis. The plot shows the unweighted noise
level in dB versus video signal level.
-48
In order to make the measurements more certain,
0.0
0.2
0.4
the camera gain was set to +18dB, and the results -49
modified by 18dB to compensate. Also, the
measurement files were high-pass filtered to
-50
remove any image shading and tilt, and a further
6dB gain applied to avoid any effects due to
premature data quantising. So, a further 6dB -51
compensation has been applied to the results, and
the graph is representative of the camera -52
performance at normal 0dB gain setting. The blue
curve has no value at high luma level because the -53
source data was accidentally clipped by slight
overexposure. The rise in noise near white level is -54
due to shot noise in the electronics and is Figure 3 Video noise levels
unavoidable in small sensors with small pixels.
0.6
0.8
1.0
R
G
B
Y
The distribution of noise level versus signal level should, ideally, follow the slope of the gamma curve
(presumably ITU709 in this case), and the values at about mid-grey are then representative of the
performance in linear mode (since the slope of the ITU-709 curve is unity at mid-grey). Clearly, the luma
noise value at mid-grey is about -52dB, and is admirably close to the specified level. This figure agrees well
with subjective assessment of the images. It is not unusual for measurements of HD cameras to differ
significantly from the specification claims, and the closeness of this measurement to the specification is quite
refreshing.
The noise levels near black should, ideally, rise as the slope of the gamma curve rises. That does not happen
in this camera, and is a possible indication that gamma-correction is done in the analogue signals before
other processing, and with amplifiers of limited gain-bandwidth product. This is a sensible economy, and
contributes towards the rather odd colour performance.
6
2.5
Conclusions
The camera is tiny, and needs no control unit. However, although the noise level is nice and low, the colour
performance is poor and the resolution significantly limited by the single-sensor. The level of aliasing is
high and could cause problems in motion-dependant compression such as MPEG2 and MPEG4.
7
Open as PDF
Similar pages