Open KIRSJ opened 3 days ago
I think the Haze removal module is not being written or read properly to/from the xmp-file.
Another round of "haze removal" issue so basically a duplicate.
jenshannoschwalm, Duplicate of which report, could you be more specific?
@KIRSJ see above PR. If i misunderstood your issue please provide more precise information including at least the rawfile and xmp . Also i was not sure what you were comparing exactly.
jenshannoschwalm, Thank you. The #17650 explanation goes over my head, as I am not directly involved with the Darktable code. My observation is a shift in contrast and colors using this module when the (ARW) raw-based output from Darkroom is exported to a JPEG. If the technical explanation in #17650 covers that then good, otherwise I can't provide further documentation. Sorry about that, but wanted to ring the bell anyway should others have had a similar problem. I rarely use the Dehaze removal module. But as far as I understand is has not been flagged as problematic when using the scene-referred (filmic) mode.
Anyway, thanks for reporting and thus reminding me to fix this :-)
jenshannoschwalm, Thanks for the clarification. When do you think the fix will be available from nightly builds?
Pretty soon i think :-)
Good, thanks.
Describe the bug
Go to 'additional context'... I am not able to provide files. I just want to describe the phenomenon based on more than one year of daily editing in Darktable without this problem - has this problem been reported by others. Normally using Darktable and lately version 4.9.0-549-g0fbd178e3 the Export function typically to 8 bit JPG at 80 percent quality creates JPG copies close to be visually identical to the source raw file (ARW). The latest image shows a discrepancy in color and contrast. The original and often slightly cropped raw file, i.e. almost original size, is exported with the following parameters: JPEG (8-bit), 80, auto, 2048x0, no, yes, sRGB (web-safe), relative colorimetric, none. Thumbnails use the raw file instead of the embedded JPG. This is my suspicion: With the actual image having a yellow-green foreground with grass isolated from the background of more grey-blueish tree trunks using a masking path in the Color equalizer module, it seems as if the separation is lost in the copy and the editing making the grass more warm with a saturation - bleeds to the background - making it more warm and less contrasty.
Steps to reproduce
See the description above.
Expected behavior
The JPG copy is normally largely identical viewed in Lighttable.
Logfile | Screenshot | Screencast
No response
Commit
The Export function in Lighttable.
Where did you obtain darktable from?
GitHub nightly
darktable version
4.9.0-549-g0fbd178e3 and 4.9.0+776~g6667649af8
What OS are you using?
Windows
What is the version of your OS?
Microsoft Windows 11 Pro, version 10.022631
Describe your system?
Dimensioned for image editing.
Are you using OpenCL GPU in darktable?
Yes
If yes, what is the GPU card and driver?
AMD Radeon RX 6600
Please provide additional context if applicable. You can attach files too, but might need to rename to .txt or .zip
The same behavior for the mentioned versions of Darktable. A TIFF file from editing the same ARW-file in Affinity exported to the actual Darktable folder - and stripped from the xmp-file - can be exported to an identical JPEG using the same settings in Darktable. Fresh edit: The problem seems to be introduced with the Haze removal module here used at 0.51 strength.