Hi all,
I’d like to try to contribute my frequency domain chroma approach to x-trans demosaicking.
See also the chain of e-mails on Redmine bug #10333, where I’ve also got some links to sample images.
However, for this I’d still need a helping hand on the makefiles, where I don’t have enough experience.
In my fork, I introduced a new dependency on fftw3, and I have difficulties to adjust the make files properly. I managed to build & run properly locally, but it’s a (more than) dirty hack.
Demosaicking quality is pretty decent with this approach especially for higher iso, and with fftw3 even the speed starts getting acceptable (still not blazingly fast).
Alternatively, I have a code version without dependency on fftw3, implementing the convolution filters as nested loops, but that’s around 45% slower on my machine :-(
I’d be glad to hear your opinion on feasibility of this. . .
My feeling is that it's challenging, on the other hand there’s still some time left until winter.
Cheers,
Ingo
Op 22 juni 2017 om 22:26 schreef Ulrich Pegelow <ulrich.pege...@tongareva.de>:
Am 20.06.2017 um 11:43 schrieb Tobias Ellinghaus:
> Hello there,
>
> we are currently thinking about having the next feature release (2.4.0)
> earlier than Christmas.
> In order to plan for that we need to know if anyone is currently working on
> something that should go into the release. We would also need some hands on
> deck to help review and polish pull requests.
Short remark on the user manual. I will be able to start work on the
2.4.0 update early August and should have finished the English version
round-about early September.
Ulrich
___________________________________________________________________________
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
