rawspeed: SONY ILCE-7C support

[frantic1048]

Part [2/2] of darktable-org/darktable#6669

Camera info was extracted with dngmeta.sh and Adobe DNG Converter 13.0 (Macintosh)

---

Update at Sun, 08 Nov 2020 02:35:01 +0800:

output from dngmeta.sh

/mnt/waifuh1t/repo/darktable/tools> ./dngmeta.sh  /tmp/raws/_DSC0021.dng
DNG created by : Adobe DNG Converter 13.0 (Macintosh)
DNG Illuminant : 21 (should be 21)
$ nano -w src/external/adobe_coeff.c
{ "Sony ILCE-7C", { 7374,-2389,-551,-5435,13162,2519,-1006,1795,6552 } },
$ git commit -a -m "adobe_coeff: SONY ILCE-7C support" --author " chino@amausaan"
$ nano -w src/external/rawspeed/data/cameras.xml (mind the tabs)
<Camera make="SONY" model="ILCE-7C">
	<ID make="Sony" model="ILCE-7C">Sony ILCE-7C</ID>
	<CFA width="2" height="2">
		<Color x="0" y="0">RED</Color>
		<Color x="1" y="0">GREEN</Color>
		<Color x="0" y="1">GREEN</Color>
		<Color x="1" y="1">BLUE</Color>
	</CFA>
	<Crop x="0" y="0" width="0" height="0"/>
	<Sensor black="512" white="16383"/>
</Camera>

NOTE: The default crop exposes the full sensor including garbage pixels, which need to be visually inspected. (negative width/height values are right/bottom crops, which are preferred)
NOTE: Sensor black and white levels sometimes vary based on ISO, please run this tool on raws for each of the camera's supported ISOs
$ git commit -a -m "rawspeed: SONY ILCE-7C support" --author " chino@amausaan"

[codecov]

Codecov Report

Merging #250 (084255a) into develop (1c58d20) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff            @@
##           develop     #250   +/-   ##
========================================
  Coverage    60.71%   60.71%           
========================================
  Files          234      234           
  Lines        12594    12594           
========================================
  Hits          7646     7646           
  Misses        4948     4948           

Flag	Coverage Δ
integration	47.91% <ø> (ø)
rpu_u	47.91% <ø> (ø)
unittests	23.11% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1c58d20...084255a. Read the comment docs.

[trougnouf]

The cam_xyz info seems to differ from libraw's (whereas it is consistent on another camera). I don't know how either is generated. This results in a different "camera reference" white balance on darktable.

libraw's cam_xyz = [[ 0.5271, -0.0712, -0.0347], [-0.6153, 1.3653, 0.2763],[-0.1601, 0.2366, 0.7242]] (obtained with rawpy 0.17, libraw 0.20.2)

darktable's adobe_xyz_to_cam (from rawspeed) = [[0.737399995, -0.238900006, -0.0551000014], [-0.543500006, 1.31620002, 0.251899987], [-0.100599997, 0.179499999, 0.655200005]]

[kmilos]

The basic matrix comes form the ColorMatrix2 of the DNG produced by the Adobe converter, and is the same for both libraries.

https://github.com/LibRaw/LibRaw/blob/master/src/tables/colordata.cpp
https://github.com/darktable-org/rawspeed/blob/develop/data/cameras.xml

libraw could be additionally multiplying this by the equivalent of the DNG CameraCalibration2 matrix and/or the AnalogBalance matrix (no idea how it gets these from the ARW) which can also differ per camera unit and per shot (this is outlined in the DNG spec as XYZtoCamera = AB * CC * CM), while rawspeed does not (i.e. I guess it assumes AB and CC are identity).

[lclevy]

http://lclevy.free.fr/cr2/#color

[kmilos]

Et voilà ! Again, no idea if ab and cc are really not identity in this case and where to get them in Sony MakerNotes...

[kmilos]

Hang on, being too smart here - this matrix returned by libraw looks suspiciously like the ILCE-7 one, so this could be just a loose matching bug/fall-though? ILCE-7C matrix was added after 0.20.2 it seems (202101 snapshot)...

[trougnouf]

Hang on, being too smart here - this matrix returned by libraw looks suspiciously like the ILCE-7 one, so this could be just a matching bug/fall-though? ILCE-7C matrix was added after 0.20.2 it seems (202101 snapshot)...

Indeed you are right! I get the same value as rawspeed / exiv2 after upgrading libraw to 0.20.2.r81.gadcb898a (current git master)