Dear friends, I installed the “Adobe DNG Converter” which comes with several camera profiles, including specifically my Sony A6300 camera. I would like to know how to use these profiles on Darktable.
You really can’t…they are DCP files and DT is icc based… ART and RT support them
1 Like
Lensfun can convert the lens correction profiles for use in darktable, but the color correction profiles can sadly not be used. DCP support in darktable would be a dream come true.
https://lensfun.github.io/manual/latest/lensfun-convert-lcp.html
Friends, I converted the dcp to icc formats using the dcamprof program and then placed the icc files in the C:\Users\Marcio\AppData\Local\darktable\color\in folder. However, it became very strange after I opened the DT: 1) the names appear as unknown camera (figure 1) and the colors are very wrong (figures 2 and 3)
You also have a few modules after that input profile which influence colours (2 copies of color balance RGB and a copy of Lut3D. I suspect the LUT might not work as expected with a different input profile.
I reset the photo to raw without editing, but it didn’t work. As you can see in the attached image, nothing has changed. Another thing that I found strange besides the name “unknow camera”, is that when I switch to other profiles with that name, the colors do not change.
Adobe DCP is an alternate workflow encapsulated in the camera profile. Bits and pieces of a DCP can be extracted and used, e.g., RawTherapee uses Adobe DNG Converter ColorMatrix1/ColorMatrix2 for about 28 cameras (ref: dev branch camconst.json), and any software using lensfun for lens correction can incorporate ‘acm’ model corrections using the lensfun-convert-lcp script.
Bottom line is that using Adobe information in other software is a bit fiddly, or you may already be doing it and not knowing it in the case of appropriated ColorMatrix data.
- https://github.com/Beep6581/RawTherapee/blob/dev/rtengine/camconst.json, I used grep and wc to do my count.
- https://lensfun.github.io/manual/latest/lensfun-convert-lcp.html, overviews the LCP migration process and gives some discussion on why it may not be the optimal solution.
1 Like
You see the name tag which is part of the profile not the file name in the dropdown. When you convert you had to give it a name … I really think it’s a waste of time. Just work on making a good ICC profile. You might need to invest in a color checker card if you don’t have one… DCP are not simply a slightly different icc. IMO it not going to give you anything to work with trying to convert these…
2 Likes
I stopped trying to do this. Never got results I liked, and DT results without any were always better.
Anyway.
- The ‘unknown camera’ : you need to name the ICC file you make when you use dcamprof. If you don’t, you get this.
- The ICC files can be created two ways. One with a Lut inside it. The other with a matrix.
- If you don’t see a lot of difference, you are probably making matrix profiles that don’t differ DCP to DCP. Matrix in this case, it sort of makes a few very simple adjustments to set a ‘direction’ of the alterations. You sort of set the black and white point, but everything in between must be linear.
- If DCP profiles have the same matrix but different Luts, and you make a matrix icc file , you will see no difference. You ignore the Lut part of the DCP.
- If you make a Lut ICC file, you get way more and precise alterations . The problem is that the lut stops at a certain point (100% value) and everything above is clipped. The darktable pipeline is meant to place the profile later when image data is already exposure compensated , so might already be above 100%. And then applying an icc profile clips the data. You have to work around this by moving darktable modules around, and/or lowering exposure first, then raising it after the profile is applied.
- A DCP file can contain way more than an icc file can handle , and is made for the adobe style pipeline. And my conclusion is it just doesn’t fit or work well in darktable. A DCP file can contain multiple matrices , Luts and curves that all depend on white balance , and the all that data gets interpolated based on the white balance of your shot. An icc just can’t contain that data and darktable can’t (simply) blend between two icc profiles.
Every time I start playing with profiles in DT I give up a few days later. Or it seems to work well, and weeks later you get a few shots that are certainly ugly with the profile that worked well on the other shots before.
This isn’t worth the hassle and the gain I get is minimal, so I just use DT without and i het good results.
1 Like
I found that a ICC LUT profile made from a target shot didn’t do much better than the matrix profile. Spectral data for the camera informed a LUT that did much better.
The alterations can be more precise compared to a matrix. Of that actually yields to better results , indient say anything 
. In my experience it doesn’t , just like you. But I also think that this might differ per camera and per person (and expectations ).