Difference between revisions of "AXIOM Beta TestsTODO"
From apertus wiki
| Line 1: | Line 1: | ||
#raw12 snapshot of colorchecker passport with 2010 offset (regs 87/88) gain x1 and x2 - exposure so nothing clips | #raw12 snapshot of colorchecker passport with 2010 offset (regs 87/88) gain x1 and x2 - exposure so nothing clips | ||
#repeat Hutch and IT8 charts - same framing for both Nikon and Apertus, please! (surroundings do matter, and so does the camera orientation) | #repeat Hutch and IT8 charts - same framing for both Nikon and Apertus, please! (surroundings do matter, and so does the camera orientation) | ||
#grey out of focus wall, gain x1,x2 with 100 frames of different exposure | #grey out of focus wall, gain x1,x2 with 100 frames of different exposure settings (listed below), then run a script and upload the results (no need to upload the raw files) | ||
#one small video of the out of focus wall (5 frames or so), with exposure on the dark side (around 7 stops below white), gain x2 (to check FPN correlation between frames) | #one small video of the out of focus wall (5 frames or so), with exposure on the dark side (around 7 stops below white), gain x2 (to check FPN correlation between frames) | ||
#raw12 snapshots of real world scenes (outdoors, people, etc.) with high & low gain, long & short exposure times | #raw12 snapshots of real world scenes (outdoors, people, etc.) with high & low gain, long & short exposure times | ||
| Line 14: | Line 14: | ||
* Please use a static scene (same background in both Nikon and Apertus), and try to take the picture from the same location (they should have the same framing *after* cropping the Nikon image to APS-C). | * Please use a static scene (same background in both Nikon and Apertus), and try to take the picture from the same location (they should have the same framing *after* cropping the Nikon image to APS-C). | ||
Exposure | Exposure settings for the gray frames: | ||
exp_time reg: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17, 18, 19, 21, 22, 24, 26, 28, 30, 33, 35, 38, 41, 44, 48, 52, | exp_time reg: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17, 18, 19, 21, 22, 24, 26, 28, 30, 33, 35, 38, 41, 44, 48, 52, | ||
56, 60, 65, 70, 75, 81, 88, 95, 102, 110, 119, 128, 138, 149, 161, 174, 188, 202, 218, 235, 254, 274, 296, 319, | 56, 60, 65, 70, 75, 81, 88, 95, 102, 110, 119, 128, 138, 149, 161, 174, 188, 202, 218, 235, 254, 274, 296, 319, | ||
| Line 27: | Line 21: | ||
1692, 1826, 1970, 2125, 2292, 2473, 2668, 2878, 3105, 3349, 3613, 3898, 4205, 4537, 4894, 5280, 5696, 6145, 6629, | 1692, 1826, 1970, 2125, 2292, 2473, 2668, 2878, 3105, 3349, 3613, 3898, 4205, 4537, 4894, 5280, 5696, 6145, 6629, | ||
7152, 7715, 8323, 8979, 9687. | 7152, 7715, 8323, 8979, 9687. | ||
To compute the exposure times, please write down the LVDS clock value used during the test. | |||
Script to be run on those 100 frames: | Script to be run on those 100 frames: | ||
raw2dng *.raw12 | raw2dng *.raw12 | ||
python2 [http://dl.dropboxusercontent.com/u/4124919/bleeding-edge/apertus/raw12_stats.py raw12_stats.py] | python2 [http://dl.dropboxusercontent.com/u/4124919/bleeding-edge/apertus/raw12_stats.py raw12_stats.py] | ||
Revision as of 16:19, 18 December 2015
- raw12 snapshot of colorchecker passport with 2010 offset (regs 87/88) gain x1 and x2 - exposure so nothing clips
- repeat Hutch and IT8 charts - same framing for both Nikon and Apertus, please! (surroundings do matter, and so does the camera orientation)
- grey out of focus wall, gain x1,x2 with 100 frames of different exposure settings (listed below), then run a script and upload the results (no need to upload the raw files)
- one small video of the out of focus wall (5 frames or so), with exposure on the dark side (around 7 stops below white), gain x2 (to check FPN correlation between frames)
- raw12 snapshots of real world scenes (outdoors, people, etc.) with high & low gain, long & short exposure times
- colorchecker passport zoomed in on the bottom half and defocused a bit (just the boxes area) - for FPN
For all samples:
- Please set the black offset registers (87,88) to 2010. 1910 is too low (clipped shadows = data lost). To get a normal picture, use raw2dng --black=100.
- Please save the register metadata block (on past samples, many files were mislabeled).
For the color charts:
- Please use a static scene (same background in both Nikon and Apertus), and try to take the picture from the same location (they should have the same framing *after* cropping the Nikon image to APS-C).
Exposure settings for the gray frames:
exp_time reg: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17, 18, 19, 21, 22, 24, 26, 28, 30, 33, 35, 38, 41, 44, 48, 52,
56, 60, 65, 70, 75, 81, 88, 95, 102, 110, 119, 128, 138, 149, 161, 174, 188, 202, 218, 235, 254, 274, 296, 319,
344, 371, 400, 432, 466, 503, 542, 585, 631, 681, 735, 793, 855, 922, 995, 1074, 1158, 1249, 1348, 1454, 1569,
1692, 1826, 1970, 2125, 2292, 2473, 2668, 2878, 3105, 3349, 3613, 3898, 4205, 4537, 4894, 5280, 5696, 6145, 6629,
7152, 7715, 8323, 8979, 9687.
To compute the exposure times, please write down the LVDS clock value used during the test.
Script to be run on those 100 frames:
raw2dng *.raw12 python2 raw12_stats.py