Bug #11913

Haze module bug?

Added by Bob Tregilus 14 days ago. Updated 10 days ago.

Target version:
Start date:
Due date:
% Done:


Affected Version:
other GNU/Linux
hardware architecture:


Hi -

Looks like there maybe a bug in the "haze removal" module and/or it is conflicting with "exposure" and "shadows and highlights" modules?

What I noticed before the bug showed up:
When I was adjusting the exposure black levels slider I saw a message appear at the bottom of the screen saying something about "warning inconsistent output".

After that I started seeing the following bug (see attached images):

Image #1 "haze" is off and "exposure black" is set to 0.0000 and image looks okay.

Image #2 "haze" is on (set to 0.20 / 0.250) and "exposure black" is 0.0000 and image looks okay.

Image #3 "haze" is on (same setting) and "exposure black" is set to 0.0010 and you can see the big black patches in the sky. If I turn the haze off the image looks fine.

Image #4 "haze" is on (same setting) and "exposer black" is set to 0.0010 and if I switch "shadows and highlights" from "gaussian" to "bilateral filter" you can now see big white patches in the sky where the black patches were.

So if "haze" is on and "exposure black" is turned up the sky gets big black patches with "gaussian" and big white patches with the "bilateral filter".

Upon closing and restarting DT the image looked correct with "haze" turned on and "exposure black" turned on until I switched from "gaussian" to "bilateral filter" and back which caused the bug to reappear.

Also, I deleted the .xmp file associated with this image and rebooted the computer and got the same results upon loading the image into DT.

However, I have an image taken shortly before this one and I cannot replicate the bug. Additionally, I have another image taken much later in the day (a much brighter image) and I can't replicate the bug on that one either.

So this bug seems to be associated with this one image and the fact that the bug did not appear until after I got the message "warning inconsistent output" at the bottom of my screen.



hazebug.tar.gz (11.4 MB) Bob Tregilus, 01/09/2018 07:24 AM

DSC00464.tar.gz (25.5 MB) Bob Tregilus, 01/12/2018 09:50 AM


#1 Updated by Bob Tregilus 14 days ago

Update: I have a second image where this bug appears. I never saw the "warning inconsistent output" at the bottom of my screen though. I could have missed it however.

#2 Updated by Ulrich Pegelow 11 days ago

The warning indicates a problem that is only transient, i.e. it will typically go away with the next action that changes the image (panning, zooming, changing parameters etc.). "Inconsistent" means that there is some difference between center display and navigation display.

This is usually caused by timing related side effects of the two pixelpipes that process the center image and the navigation image, respectively.

Are the messages always related to one specific module in your case?

#3 Updated by Bob Tregilus 11 days ago

Thanks for the explanation of the warning, which I only saw once, but which preceded the "bug" on that single image.

So no, the message is not specific as it was only seen once.

As mentioned, I have a second image with this same problem, and now just a few moments ago, I found a third image from a set shot two years earlier. However, this image looks fine when manipulating the photo in darkroom, only when I switch to lightable does the problem occur.

I'm beginning to think it has more to do with the image than it being a bug in DT. Or, perhaps, just a limitation on how much I can tweak the image.

So far, it only occurs on images with extreme dynamic range bordering on being blown out on the high end. Perhaps there a few pixels that are blown out and for whatever reason after applying the 'haze' module DT just looses it and blows the entire area out?

I'm not sure what exactly 'haze' does, I rather like applying a little, ~0.10 to 0.20, to my images. It seems to brighten colors and remove haze a little bit at that strength.

But as mentioned, so far I only seem to have the problem on images with extreme dynamic range.


#4 Updated by Ulrich Pegelow 10 days ago

Please attach the RAW+XMP to help others reproduce the issue.

#5 Updated by Bob Tregilus 10 days ago

Attached below.


Also available in: Atom PDF