Hi Dan, My present algorithm (3 pass) takes about 11 to 12 secs on my machine (Intel on-processor graphics, so nothing fancy). Markstein (3 pass) via dcraw takes about 35 secs on my machine. Both for a 16 megapixel x-trans raw.
However, I use openCL not only for speed reasons, but also because I like the modularity that the kernels give me. I’m presently still pretty much in the experimental phase, and I like that I can change order and parameters of the kernels and then see how that impacts the quality of the output. So far, my emphasis is not yet on speed, and I’m quite optimistic that I can squeeze out some more speed once all the design decisions are taken. Most importantly: 1. Presently, I use guided filtering / upsampling not only for constructing the green pane, but also afterwards for red and blue. For red and blue, DCI would be another option once green is there. This could not only improve speed, but also improve quality. I’m presently working on some experiments in this direction. 2. My approach is iterative, starting from a provisional green. Presently my provisional green is created in quite a sophisticated manner. I think I might overdo this a bit - I could likely simplify this part without sacrificing quality, but I don’t expect the savings to be revolutionary. 3. As said, my approach is iterative. The above time is with three passes. If I reduce to two passes, about 2 to 3 seconds can be shaved off. My first impression is that after two passes further visual improvements become marginal, but I want to check this a bit more before prematurely concluding. So the 11 to 12 secs are already a bit of a worst case scenario. Now as to documentation: don’t be too optimistic ;-) , so far I only documented a rough outline of my idea. Concerning a CPU variant: once all design decisions are taken, this shouldn’t be that much of an effort. But even before that, I’d clean up my GPU variant. As said, presently it’s still quite experimental. Cheers, Ingo > Am 08.02.2016 um 23:52 schrieb Dan Torop <d...@pnym.net>: > > Hi Ingo, > > This is quite interesting work to see... A x-trans demosaic algorithm > which is well described, high quality, open source, and fast is > something which I'm sure many people are awaiting. Though of course > having all of these qualities is a lot to ask! It's great to see > continued work on this, and in particular addressing the color > artifacts. > > How does the speed of your code when hooked into dcraw compare to 1-pass > or 3-pass Markesteijn via dcraw? The dt version of Markesteijn is about > 2-3x faster than dcraw's, if I recall right, but dcraw's Markesteijn > could still be a good basis of comparison. > > How much work would it be to make a CPU variant? So far as I know, all > of darktable is built to function on CPUs with the possibility of GPU > speed-up in certain cases. > > I can't speak for the dt core developers regarding their interest & > priorities, of course... > > Best, > Dan > > > > On Mon, Feb 8, 2016, at 03:42 PM, Ingo Liebhardt wrote: >> Hi all, >> >> Congrats to version 2.0.1. >> >> Would you maybe be interested in an alternative approach to the Markesteijn >> x-trans demosaicing? >> >> I see that for Bayer patterns you have a fast one, plus two different >> high-quality ones (AMaZE and VNG4). >> >> The only high-quality one for x-trans seems to be Markesteijn. >> I personally find that Markesteijn is producing very sharp results, but also >> quite some false colour artifacts. >> I’ve been playing around with an alternative approach, and I’m slowly >> starting to get reasonable results. (even images with lots of green - always >> problematic - start looking okay(ish)). >> >> If you want to have a look: >> https://github.com/ILiebhardt/xtrans >> >> And some sample comparisons to Markesteijn, plus a brief explanation of the >> idea: >> https://www.storehouse.co/stories/b8sj2 >> >> Don’t be mistaken by my version number: there’s still a lot of work to be >> done, and I also still have quite some ideas for improvements… >> >> So at this stage I just want to carefully pre-inquire if there could be some >> interest, in principle. >> >> Thanks a lot for letting me know. >> >> Cheers, >> Ingo >> >> >> >> ___________________________________________________________________________ > darktable developer mailing list > to unsubscribe send a mail to > darktable-dev+unsubscr...@lists.darktable.org > >> > ___________________________________________________________________________ > darktable developer mailing list > to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org > ___________________________________________________________________________ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
