I've generated profiled denoise data for darktable, essentially following the procedure described by Andreas Schneider (
PIXLS.US - How to create camera noise profiles for darktable), but did deviate a little.
I found that using a printed black to white gradient against my screen did not produce sufficient intensity in the intermediate brightness values.
After a lot of experimentation I ended up simply displaying the black to white gradient image and placing the black card cutout over that.
I have done some limited testing, and think it would be good if others can also do some testing to ensure these profiles are OK before I submit them to the darktable team.
The following description on how to test is for a Linux system.
Save the attached file 'presets.txt' to directory (for example) /home/noise. Rename the file to 'presets.json'.
To test you need to start darktable from the command line in a console window, as follows...
darktable --noiseprofiles /home/noise/presets.json
and then in darktable import your test images as normal and apply denoise (profiled).
Note that using darktable this way will only provide profiled denoise for the K-3 Mark III images, when started this way darktable only has access presets.json noise profile data.
Please let me know if you have any problems, and any comments on the quality of the denoising using this data.
Cheers,
Terry