Hi, is anyone able to do a better job than this in darktable?
IMG_6208.cr2 (13.4 MB)
This file is licensed under ‘Creative Commons, By-Attribution, Non-Commercial, Share-Alike5
I’ve brought up that topic two or three weeks ago. The color highlight reconstrution doesn’t do a good job. RT is much better. However hanatos started to do some research in that area and I think Aurelien started to look into it too.
Agreed. For this image, I just reconstructed highlights in RT and saved a 32-bit tif to feed darktable.
Good news
I’m starting to see Aurelien’s second video and he says at some point that one of the problems with lab color space is that it doesn’t handle well high dynamic range images. I wonder if there’s any chance that the current highlight recovery module is working in lab and that would be the reason it easily produces artifacts.
No, that’s a different issue.
Indeed.
I see what you did there 
I don’t have better for now.
1 Like
Had good results with ‘reconstruct color’ on the upper arm of the man right to the man with the red shirt, but very strong artifacts in the other overexposed areas. First idea was to use a mask and a second instance with ‘reconstruct LCh’ but in this module you can’t add a second one.
Only went to dynamic range and highlight reconstruction here, didn’t continue to denoise/sharpen and the rest I do otherwise
IMG_6208.cr2.xmp (9.3 KB, darktable 3.0rc2)
1 Like