colors different in fit to screen and 100% view

#1

Hi,

Screenshot
Screenshot.png1920×1200 1.95 MB
Screenshot-1
Screenshot-1.png1920×1200 4.17 MB

As you can see the fur of the animal looks different, it is quite a bit darker in fit to screen view.

I have set AdobeRGB as display profile and preview profile.

Here are the raw and the sidecar file. After exporting it appears that the fit to screen preview is more accurate.P3280299.ORF (13.5 MB)

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Is this another bug in dt 3.0rc1? (Maybe a blackpoint issue?)

Thanks in advance for checking

Anna

Edit: the right sidecar file is on one of the comments below

#2

can you try to remove each module one by one and see if one in particular is causing it ?

Also the sidecar file you gave is PP3 (Rawtherapee format), not XMP.

#3

Hi
This is probably caused by denoise (profiled). It is very hard to make a perfectly accurate preview in the case of denoising, so it is better to look at the image at 100% when denoise profiled is activated.

2 Likes
#4

ups, sorry

#5

oh no, I re-edited and I think I lost the original sidecar file

#6

The combination XMP + raw you gave works fine with me:

Capture%20d%E2%80%99%C3%A9cran%20du%202019-11-13%2021-27-45
Capture d’écran du 2019-11-13 21-27-45.png3840×2104 9.54 MB

#7

its a different edit, forget it

#8

P3280299.ORF.xmp (6.4 KB)

I think I was able to re-edit it quite similarly. I think the phenomenon occurs when I go above detail 300 in the local contrast module.

#9

Anyway, as I said, the exported image looks more similar to the fit to screen view, not to the 100% view.

#10

OK. I did a few more test. Apparently, this “bug” isn’t there without OpenCL. But with ocl it is even more apparent when I go up to 500% with detail in local contrast.

#11

Why is everybody so quiet here?

#12

I guess, darktable folks are busy with getting the release ready, judging based on the activity on github. Maybe you already did, but if not, and you still consider your observation a serious bug (after considering the input you got in this thread), think about filling a bug report.

#13

I did a few days ago. One user could reproduce. I guess the dev of this module is occupied with other things… I hope he will fix this eventually, not expecting an instant fix