Struggling on raw blue color saturation

Software darktable
1

There seems to be lots of info on the problem I’m encountering, but I’m really having trouble understanding it and how to deal with it. Here are 5 100% crop samples of opening a specific raw file. Left to right are:

  1. DT with default Input Color Profile module (Input profile - standard color matrix; working profile - linear Rec2020 RGB; Gamut clipping off)
  2. DT with adjusted ICP module (all set to Rec2020 RGB)
  3. Lightroom raf converted to DGN using Iridient X-transformer
  4. LR and RAF file
  5. Capture 1 Express RAF file)

DT%20ICP%20default-side
DT%20ICP%20default-side5000×750 2.2 MB

3-5 are what the jacket color is, and note the artifacts in #1 resolving in #2 throws the color off. Photo is ISO 3200; 1/800 sec.

All the other photos look normal in DT - except the ones where my wife is wearing that blue jacket.

I’m getting lost in the tech talk on building or creating icc profiles, or the right way to set up DT to manage this issue. Seems more I dig into it the more confusing it becomes. From what I can tell, I think I need specific icc files for my X-T20 camera? And it needs to be in a specific folder?

I’m also looking for good, simple, plain language sources to help decipher how all this color profile stuff works. Jeez, I’ve got a university degree yet I feel like I’m back in grade 3 trying to figure this stuff out!

Running DT 2.7 1477 on Win10. I have calibrated my monitor.

Apologies as I know this issue has been raised before, but I’m just not getting a solution I understand. Any help is appreciated.

2

It’s hard to say without having access to the raw file but I think this is related to “preserve colors” setting added to base curve in darktable 2.7. So, please try to set this setting in the base curve to “none” and see if it solves the problem. You should use “standard color matrix” for ICP.

3

Thanks - I did have base curve off, but also tried it on with preserve colors set to off. Playing around with base curve on/off and colors just creates slightly different variations of the problem. I also just downloaded RawTherapee and although it does a much better job on the this raw file than DT, it does present the blue as a bit more purple - so kind of in between the LR/DNG/C1 versions and the DT version.

Here is a link to the raw file: Dropbox - File Deleted

4

Seems like the camera white balance is causing the weird blue colors, I suggest setting it to something closer to daylight and then using color toning to shift all non blue colors to something more bluish if you prefer the original toning.

I am quite happy with this starting point in RT, using adobe standard dcp and daylight wb. I removed the automatic base curve and used the dcp base curve.

Long%20Beach-66
Long%20Beach-666018×4012 2.28 MB

The contrast is relatively mild and you can tweak everything to taste

5

IMHO It’s definitly the standard color matrix in Input Color Profile causing this problem. (in combination with basecurve)

Thats the reason I calibrated my camera with darktable-chart and use Rec2020 as Input Color Profile to get nice color:

there are some styles on https://dtstyle.net/ for Fuji cameras. did you try those?

6

Thanks for the suggestions. I did play with WB, then did need to play further with color shifts in the non-blue portions. I got reasonable outcomes, but still don’t understand why its so different in DT compared to the other software - especially since any of the shots without the significant blue pixels come out looking very similar to other software (both in WB and overall color).

I’ll try using some Fuji styles and see how those perform.