AXIOM Beta TestsTODO

From apertus wiki
Revision as of 16:19, 18 December 2015 by A1ex (talk | contribs)
Jump to: navigation, search
  1. raw12 snapshot of colorchecker passport with 2010 offset (regs 87/88) gain x1 and x2 - exposure so nothing clips
  2. repeat Hutch and IT8 charts - same framing for both Nikon and Apertus, please! (surroundings do matter, and so does the camera orientation)
  3. 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)
  4. 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)
  5. raw12 snapshots of real world scenes (outdoors, people, etc.) with high & low gain, long & short exposure times
  6. 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