I am finding that the 'denoise (profiled)' when switched from NLM to wavelets is doing a fairly good job in particular on my sky areas .... it is a bit aggressive but can be toned down as needed.
On 03/10/2018 05:34 PM, Robert Krawitz wrote: > On Sat, 10 Mar 2018 17:04:45 +0100, David Vincent-Jones wrote: >> When I apply the 'denoise (profiled)' onto a fairly raw image the >> results, on my data, look quite acceptable. I have been trying to see >> which modules that I am using are creating more noise than others. >> >> In another test I have used the para. mask to eliminate some processing >> from sky areas ... it is a bit tedious but it appears to help. I still >> feel that the blue is the trouble-maker. > > Tedious is something I cannot abide when I'm processing a few hundred > game photos (much less 3 weeks ago, when I had 600 frames). I need > the most efficient workflow I can get. It's already wasteful to me > that I can't simply hit space to move to the next photo and be able to > crop right away; I need to click on the crop settings to be able to > crop. > > For my use case, I don't need absolute elimination of noise, but I > like to cut down the noise some for the very high ISO settings I use. > If I were able to shoot at ISO 1600 I wouldn't even both bother with > NR at all. But I don't want artifacts or complete removal of detail. > If I didn't have an alternative I'd accept the noise. > >> On 03/10/2018 04:56 PM, Robert Krawitz wrote: >>> On Sat, 10 Mar 2018 16:44:24 +0100, David Vincent-Jones wrote: >>>> Your examples interest me since they are shown on a blue subject. My >>>> experience with a fully profiled sensor is that the 'basic' Denoise >>>> (profiled) works quite well by itself EXCEPT for blue sky areas. >>>> >>>> My sky areas tend to form into rosette clumps whenever some reasonable >>>> degree of processing is applied. It has led me to believe that it is the >>>> blue color that is causing the greatest problem. >>> >>> That's not my experience -- I'm seeing clumps like this in plenty of >>> other places. > > ____________________________________________________________________________ darktable user mailing list to unsubscribe send a mail to darktable-user+unsubscr...@lists.darktable.org