Tif which uses Linear rec2020 RGB seems to save color profile as sRGB

[pablo-dra]

Is there an existing issue for this?

• I checked and did not find my issue in the already reported ones

Describe the bug

I'm currently trying to use Tif as Export
I used the Linear rec2020RGB for both input and output color profile.

But there seems to be something wrong with the image in some softtwares.

• Ok ones:
• Affinity
• Windows preview

Not working as expected:

• Nomacs
• PTGui

This is the DarkTable preview which seems ok...

This is Nomacs preview

And this is PTGui preview

Affinity shows it properly and shows the color profile that was saved. (However it is weird that it reads it as a HDR 32 bit...)

Inspecting the file details, it shows that the color profile says sRGB. Should Darktable embed that the image is linear rec20RGB there?
Could this be a bug?

Steps to reproduce

Just Export image as Tiff 16-bit in Linear Rec2020 RGB

Expected behavior

No response

Logfile | Screenshot | Screencast

No response

Commit

No response

Where did you obtain darktable from?

darktable.org / GitHub release

darktable version

5.0.1

What OS are you using?

Windows

What is the version of your OS?

Windows 11

Describe your system

No response

Are you using OpenCL GPU in darktable?

Yes

If yes, what is the GPU card and driver?

No response

Please provide additional context if applicable. You can attach files too, but might need to rename to .txt or .zip

No response

[kmilos]

Please report issues w/ Nomacs and PTGui then (or whatever 3rd party software), this is not a dt problem. You can check by importing back to dt.

[kmilos]

And please stop using dt GitHub as a forum, feel free to ask questions at https://discuss.pixls.us/

[pablo-dra]

And please stop using dt GitHub as a forum, feel free to ask questions at https://discuss.pixls.us/

@kmilos
I was asking because the exported image didn't work well in other softwares, and because I noticed that the color profile in the details says sRGB when I didn't use that one (I used linear rec2020 RGB)...

For me (a standard user) it seems like a bug when exporting... as well as the other openEXR which doesn't save the EXIF data...

Thanks for the suggestion of the forum anyway.

[kmilos]

Why do you assume it is a dt problem and not the other software's broken color management on import? Did you equally open issues in their bug trackers already, or you like to single out dt for some reason?

[kmilos]

I suggest you ask questions in the forum first, try to identify the source of problems w/ help from other enthusiasts, and only then report issues where appropriate.

[pablo-dra]

Why do you assume it is a dt problem and not the other software's broken color management on import? Did you equally open issues in their bug trackers already, or you like to single out dt for some reason?

@kmilos
I'm using Darktable as a standard user, and as I said (I quote myself):

"because I noticed that the color profile in the details says sRGB when I didn't use that one (I used linear rec2020 RGB)..."

If the exported image already says another color profile since the export moment then this may be the cause. Hence the report here.

I feel that this is getting a bit out of tone... I just thought that this was an issue, again like the Exif data... I can't certainly know when a bug is completely like so. That is why there are some steps in bug reporting like triage. Users report things that may or may not be an issue.

[kmilos]

Apologies for not giving this a deeper thought initially - there is maybe one small detail dt could do better: the "sRGB" you see in the properties dialog maybe comes from the Exif ColorSpace tag that has been carried over from the CR2 preview image (you can verify this in the exiftool or exiv2 output). dt should either it to "Uncalibrated" (or leave it if we can detect the export profile is sRGB at that point).

The exported file should still be valid and has the correct embedded ICC profile as shown by Affinity. I don't know if any other apps actually use and prioritize Exif ColorSpace over embedded ICC.

[kmilos]

Hm, we seem to try to do this already:

darktable/src/common/exif.cc

Lines 2732 to 2736 in bc62056

	// Write appropriate color space tag if using sRGB output.
	if(sRGB)
	exifData["Exif.Photo.ColorSpace"] = uint16_t(1); // sRGB
	else
	exifData["Exif.Photo.ColorSpace"] = uint16_t(0xFFFF); // Uncalibrated

[pablo-dra]

@kmilos

Thanks for the follow-up.
I can confirm you that the .CR2 file says sRGB in its Exif data. (I used EXIF tool to check this)

It also shows sRGB in the Tiff 16-Bit exported using linear Rec2020 RGB.

As far as I was chatting with ChatGPT (because I'm not an expert) some programs read what color profile has the image, and that may disrupt the workflow if there is no embeded color profile in the image. But as you comment the color profile is embedded (it is nice to know that).

I guess that still may be conflictive because that data says something and then the color profile embedded is another.
Affinity seems to have this polished, and PTGui definitely doesn't give the same result as Affinity (or loading it again into Darktable as you suggested).

I managed to set an "ok workflow" with PTGui if I leave the output color profile as default (which is sRGB, and which matches the written Color space data of the Exif data).

Maybe the cause is the "profile" option in the export section? I checked that it says "image settings"(the default) in mine so it should match the one that I assigned when editing. Maybe that option overrides some info at the export moment in the backend despite changing it in the image? Just trying to find possible clues