Very kind to mention me, though I wasn’t involved in the darktable implementation at all. I just implemented it (though a bit different) in RawTherapee a while ago.
Anyway, I’m glad, that @hannoschwalm ports some RawTherapee stuff (the details maks, rcd demosaic and dual demosaic, maybe also capture sharpening) to darktable 
That’s still your idea and it’s available to everyone. Also, your willingness to share, help and be open to people from other projects is for me the best example of how to treat each other in Free Software community and that alone is worthy of praise.
Ingo @heckflosse work (and lots of mails) inspired me to take the road in the right direction 
Yes capture sharpening is one of the algos on my agenda although there is so much to do for dt and my time is limeted too
@s7habo BTW a big thank you from me showing what the details mask can be used for and how to do it. All showcases you gave are just perfect. If you are planning to do any sort of “teaching” or “training” stuff/tutorial i would certainly help if you need any sort of more technical background…
I think some examples may add to this discussion. On the following picture from Sensor Heatmaps, you can clearly spot noise contributions at lower frequencies than pixel frequency.
Also, the following pattern shows low frequency noise:
There is a difference between the technical description of the noise and the appearance to the human eye I think, as much as it is there for audio. Most humans perceive white audio noise as “high frequency”, while the spectral power distribution is flat. I think the same probably holds for spacially distributed noise. You may be familiar with the following optical illusion:
(from Wikipedia). To me, this is some hint that there is a frequency threshold under which you may not observe spacial noise as noisy, when it is “hidden” by image contents. however, I had (and probably have) to deal often with very “blotchy” noise from older digital cameras, which is really annoying and hard to get over.
In my opinion this should be a different discussion (a new topic). Just my 2 cents…
Sorry for the noise (SCNR) 
.
No need to apologize. Hijacking threads by accident also happened to me sometimes 
I think I mentioned this two years ago in Saarbrücken, didn’t I? I am just irritated that Hanno apparently joined this forum a few days after that. Funny coincidence. I wasn’t quite aware though how powerful this really is.
The power of german open source contributors
But … there appears to be a problem … there is some sort of conflict when exporting images to a jpg either on disk or email.
I am running the daily git/Manjaro/Arch/XFCE
First indications were that my system locked-up and/or crashed
But … there appears to be a problem …
Hmm, can’t reproduce right now. Could you please open a darktable issue at GitHub - darktable-org/darktable: darktable is an open source photography workflow application and raw developer providing both the raw file and your xmp file please? Also some precise steps to reproduce? And system ram, opencl, graphics card please?
Please note there have been some other fixes lately that might lead to such crashes notably there was a problem with rcd/orientation/opencl …
May I proudly amend, that even against some wind, I pushed this thru via this issue
[FEATURE] parametric mask by edge detection · Issue #7628 · darktable-org/darktable · GitHub
I just built from github source, on Mint linux, a few minutes ago. I don’t see any new mask options. What am I missing?
Is this new detail mask not pushed to the master branch yet?
It takes the form of a new “slider” in the “mask refinement” option.
For example, go to Local Contrast module, and click the “drawn mask” button - then the new slider called “details threshold” appears just below “mask refinement”
Oops. Sorry. I’m blind. This looks very interesting.
Nice level of refinement for sure. You can slide it without specifying a mask just to see it on the whole image and then of course the real power is to use it with a mask of course…
Currently I have filed a report due to an apparent problem with this function that is seen when exporting images to disk and email. Please be careful.
Thanks for warning but I have not had such problems so far.
