On Fri, 31 Aug 2018 15:27:53 -0400 Michael <bmi...@gmail.com> wrote: > follow the thread? > https://www.dpreview.com/forums/post/60637838
I hadn't, but did now. There is a second suggestion with denoise non-local means, which I also tried unsuccessfully. I believe that those solutions are for a different problem: in my case, one pixel of the original image yields about 8x8 pixels in the photo, producing rather high contrast stripes. But the photograph reproduces each original pixel perfectly shaped; there is no sensor noise. The camera just did perfectly what it is supposed to do. The screening is in the original. The moiré is not visible in a 1:1 view. But looking at the original image, the stripes are invisibly small; it looks just like a plain color (with a subtle gradient). The Gimp plug-in descreen works in the frequency domain by means of a fourier decomposition. I think, darktable has means to do this too, but I don't know how or if the pixel patches aren't simply too large. BTW, I didn't try the Gimp plug-in, as I wasn't able to install that (the plug-in repository is down, and the files I found elsewhere wouldn't work in gimp 2.10). ____________________________________________________________________________ darktable user mailing list to unsubscribe send a mail to darktable-user+unsubscr...@lists.darktable.org
