sony colors mystery

I am not sure that Sony was lazy, it is more a limitation how much processing power you can fit into a small piece of silicone.

If I was Sony or Fuji I would allow customers to pic more than 1 creative style so they can compare effects later. In miss that feature big time, because in most cases those renderings are quite good.

I still would like to have some kind of dynamic preset for DT which simulates typical creative styles from given cameras.

Thank you. While I am somewhat color blind and see colors differently, I noticed that reception of those greens varies indeed.

So, I connected my beloved ASUS Ubuntu laptop running the same version of DT as my minimacm1gen1 to the same quite good but nothing special DELL U2720Q 4k monitor, applied same sidecars and it almost seems I am getting different results. Will double check it though.

Thereā€™s a lot of secret sauce that goes into the SooC jpegs, so I wouldnā€™t expect to see that.

Unless theyā€™re both calibrated, youā€™re likely seeing things differently.

1 Like

I worked the image in my hack software, to include inspecting the metadata. What I found was that the SOOC JPEG was rendered with the ā€œGamma Still - Vividā€ PictureProfile. Accordingly, I was able to process the image close to the SOOC JPEG by using the camera ā€œas-shotā€ white balance and addition of simple HSL saturation. Hereā€™s a screeshot:

In the screenshot is the exiftool dialog, filtered to pull out the PictureProfile tag. To the left is the HSL Saturation tool pane, with the setting used (1.00 is ā€œno-saturationā€).

Duplicating the SOOC JPEG is benefitted by inspection of the metadata to divine what the cameraā€™s JPEG engine is directed to doā€¦

1 Like

My versionā€¦

20221006_0080.ARW.xmp (16.1 KB)

1 Like

This has been fixed in recent builds ā€¦ at least it should be. There is a rawspeed commit a while back to use the maker notes while opening ARW files . So for Sony , it should be ok.

I didnā€™t need to adjust the white level for my files with recent commits .

1 Like

it might have more to do even with the build environmentā€¦ I was only a day or so behind at most wrt current commits and I recall the discussion wrt a Sony fix but I still got the wrong wp ā€¦or maybe it was because I used the xmp of the OP?? I will have to try and remove and load and see what happensā€¦

1 Like

Or you didnā€™t update the git submodules ā€¦ Or the commit is merged in rawspeed , but DT didnā€™t update the submodule .

1 Like

I do init and update every time I buildā€¦

Still not picking that upā€¦will have to look into thatā€¦ I donā€™t own a Sony so in the end no big dealā€¦

1 Like

Rawspeed module not updated yetā€¦

1 Like

20221006_0080-4.jpg.out.pp3 (13,7 KB)

1 Like

Just a thought which I havenā€™t tested - if you wanted could you process your RAWs with the manufacturer supplied software? Last time I tried my Olympus software it let you develop using any of the in-camera settings. (and not that much elseā€¦)
It just occurred to me that if one uses Linux that might not be an option though.

It means that you need a build with the latest rawspeed ā€¦ Or with the branch merged that contains this fix, but i thought it was already in.

Darktable hasnā€™t updated the rawspeed commit that is used with a submodule , but you can checkout the latest head / develop / master / whatever .

It seems to be this PR ARW: Get black level and white point from EXIF by artemist Ā· Pull Request #385 Ā· darktable-org/rawspeed Ā· GitHub

1 Like

What is the best way to find out the rawspeed commit/version pulled to final official DT release? (Tags Ā· darktable-org/rawspeed Ā· GitHub seem to be not used any more)

Commit 154081c

1 Like

Itā€™s a submodule. Thatā€™s a git feature to sort of link to another git repository (keeping it simple ).

In Visual Studio Code with the git-graph extension installed you get a nice graphical view which supports submodules.

The low-level way is to take look in the .git submodules file (or whatever itā€™s called ).

Basically , after you check out darktable and do a ā€˜git submodule initā€™ and ā€˜git submodule updateā€™, you can cd into the external/rawspeed folder. Thatā€™s a git tree just like darktable itself. You can checkout a branch , make commits , push, pull, look at the branches , etcā€¦

A submodule records a commit , not a tag or a branch (well technically itā€™s all the same in git ). So in the hidden gitsubmodule file is a commit hash, and thatā€™s the commit thatā€™s linked . Doesnā€™t mean you can switch to any other commit or point in time.

1 Like

I see it now (Emacs user here). So Commits Ā· darktable-org/rawspeed Ā· GitHub definitely does not have the aforementioned Sony fix. And both 4.0.1 and master branch refer to this. I guess I should not hold my breath to have it included in upcoming 4.2 release then.

PS. I do not mind building DT myself, but I have too many other projects in flight at that point.

As a Sony user this thread prompted me to look into better understanding how things work under the hood. I have read that Raw Speed only reads the minimum data to demosaic the image, so not all the EXIF fields.

The following selected items are from an Exif Tool dump from my (ILCE-7RM3 v3.10) ARW file.

* The first three items have corresponding Raw Speed tifftag.h defines for tag number the others do not so pretty sure they are not read by Raw Speed.

Weirdly, there are two ā€œWhite Levelā€ fields with different values?

0x828d CFA Repeat Pattern Dim  : 2 2	*
0x828e CFA Pattern 2           : 0 1 1 2	        *
0xc61d White Level             : 16383	        *

0x7310 Black Level             : 512 512 512 512
0x7800 Color Matrix            : 1177 -211 91 -54 1267 -159 72 -232 1216
0x787f White Level             : 15360 15360 15360
---- Composite ----
     - CFA Pattern                     : [Red,Green][Green,Blue]

I also stumbled upon the Raw Speed cameras.xml which has similar data.

I have not yet worked out what data comes from where during normal dt & RS operation.

Separately, the EXIF color matrix values above do not match the cameras.xml ones below. Again I do not yet know if these are the same thing or not, or if either are even currently used by dt.

<Camera make="SONY" model="ILCE-7RM3">
	<ID make="Sony" model="ILCE-7RM3">Sony ILCE-7RM3</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="-20" height="0"/>
	<Sensor black="512" white="16383"/>
	<Aliases>
		<Alias id="ILCE-7RM3A">ILCE-7RM3A</Alias>
	</Aliases>
	<ColorMatrices>
		<ColorMatrix planes="3">
			<ColorMatrixRow plane="0">6640 -1847 -503</ColorMatrixRow>
			<ColorMatrixRow plane="1">-5238 13010 2474</ColorMatrixRow>
			<ColorMatrixRow plane="2">-993 1673 6527</ColorMatrixRow>
		</ColorMatrix>
	</ColorMatrices>
</Camera>

Hope this does more than add to the confusion.