Up to and including 766, the module used a fixed rotation matrix I found in some Python code. After that, we started using the output profile.
I copied the matrix code from sigmoid, but I don’t use a user-selected profile, but rather the ‘output’ profile. I thought that was the profile set in the output color profile module, but it is not.
I’ll see if there is a way around that. Stay tuned.
You mean from the Gamut Checking toggle? Yes, if I change it to sRGB away from “System display profile”, the problem goes away.
“System display profile” is the default setting, so this is an interesting little issue.
Ya sorry I said input profile but you knew what I meant…off-line I noted to @kofa that is a very old icc profile I think dating from 1996 and I wish I knew more about icc profiles, but it had a few weird values in it that were not in an updated version that you can find on color.org…in anycase you would likley be better served with the proper icc for your monitor if you can find it over that one…
For sure, and I’m happy to do this at some point. But for the purposes of testing AgX, I think it’s good to use this as a test case for compatibility. We have a very small sample of users, and we’ve already found this little issue, so for sure there will be many more when AgX is released to the outside world.
I don’t believe there was ever an official ICC profile released for my monitor because I remember looking for it some years ago. There is one that can be downloaded from TFT Central, but it’s not official, just the result of those guys calibrating it.
Acer xr341ck by the way.
1 Like
Well that profile you are using is actually outdated and just left sort of floating around in Windows as a fall back…at least IMO…
If you go to color.org you can see that some changes were made to update “the srgb” profile…
Found here
Hi,
I hit
https://tech.kovacs-telekes.org/dt-agx/Darktable-5.1.0%2B803~g4b359c70db-x86_64.AppImage
and What I ended up with is
Darktable-5.1.0+803~g4b359c70db-x86_64.AppImage ie it does not have the 2B before the 803
Which still has a checkbox for the duplicate experimental gamut code.
I am not clear which numpty error I have made. Any suggestions much appreciated.
BW
F
%2B is the web’s way to write the + that’s part of the file name.
Please don’t fight over the profile. It’s a bug in my code, a misunderstanding on my part. I’ll fix it, just give me time.
5 Likes
No fight on my part and maybe something to learn why a certain profile would then be affected by a bug and another would not or would mask it…
Just curious for the most part
Some profiles just have a matrix to convert between RGB and XYZ. Other are more complicated. That’s all I know. 
I’m ready for fisticuffs whenever you are Todd! I have very sharp elbows recently with all this Canadian pride going around…
As I’m sure you’re aware, I’m not fighting you at all. I always appreciate your efforts to help people on this forum.
3 Likes
The respect is mutual. I look forward to seeing how it all comes out in the wash… nice and white like with some bleach…
There.
@MStraeten : you’ll need more than agx.c this time for the Mac build. Please also incorporate src/common/iop_profile.c and src/common/iop_profile.h.
@priort or @Dave22152 , please update the Windows build.
https://tech.kovacs-telekes.org/dt-agx/Darktable-5.1.0%2B804~g5faede5077-x86_64.AppImage
I was unable to build with the current master, some problem upstream, I think.
5 Likes
recent macOS arm64 build:
darktable-5.1.0+713~g63784cc139_arm64.dmg
same comments as in [Blender AgX in darktable (proof of concept) - #181]
1 Like
Ok, looks like we’re up to build 813, now. Here’s the Win 11 executable:
Ah, you built from the tip of my branch, with the latest master changes. On Linux, I got a build failure with that. Good that it worked for you. I didn’t have time to investigate what was wrong, so I built the AppImage with only my changes, excluding those from the latest master.
I used the same process that I’ve done before. I just invoked git checkout -f kofa/blender-agx-poc per the directions you gave me a few weeks ago.
It seems to work (and you fixed the Lady in Red… congrats), but I can check further if you’d like.
1 Like
No, I figured it out, see above. You replied quickly, before I saw what happened. Thanks for the build, as always!
1 Like
Got it. Yeah, I went back and saw that the latest commit on my build matches yours.
Glad to help in my own limited way.
3 Likes