Re: [darktable-devel] branched darktable 1.5 and added some high noise demosaic codes /run time

2014-02-27 Thread johannes hanika
nice, thanks for the timings! (cc-ing dt-dev back, might be interesting to others) -jo On Thu, Feb 27, 2014 at 5:36 PM, Gert van der Plas - Schrama < [email protected]> wrote: > PPG: [dev_pixelpipe] took 0,584 secs (0,428 CPU) processing `demosaic' > [export] > AMAZE: [dev_pixelpipe] took 1

Re: [darktable-devel] branched darktable 1.5 and added some high noise demosaic codes

2014-02-27 Thread johannes hanika
On Wed, Feb 26, 2014 at 6:57 PM, Gert van der Plas - Schrama < [email protected]> wrote: > Hi Johannes, > > I am mostly mobile online. And nearly all of my communications is async. > So irc is not really an option. I mostly snatch here and then 15-30 minutes > coding time. On demosaic algo's, t

Re: [darktable-devel] can't changing roi_out in demosaic algorithm

2014-02-27 Thread johannes hanika
you're only processing the visible region of interest, or worse, region of interest for a certain tile. so it's essential that these pixels all get processed. you can modify the requested input roi though to span more than the output roi that the core requests from you. the inverse case is done in

Re: [darktable-devel] branched darktable 1.5 and added some high noise demosaic codes /run time

2014-02-27 Thread johannes hanika
On Wed, Feb 26, 2014 at 7:04 PM, Gert van der Plas - Schrama < [email protected]> wrote: > Run times. Time complexity is O (1) > in the number of pixels? impressive. i was more interested in output of `darktable -d perf' in milliseconds for example. j. > for stagger. No clue for igv. But it

[darktable-devel] can't changing roi_out in demosaic algorithm

2014-02-27 Thread Gert van der Plas - Schrama
hi, At high iso border-interpolations from some demosaic codes become noticable by being very different from the main demosaic algorithm. Could we have an option where we simply drop these border pixels? I.e. that the current selected demosaic algorithm returns the bordersize and roi_out is proper

Re: [darktable-devel] build error in current git

2014-02-27 Thread Michael Born
Sorry :-( I didn't see this. And didn't expect such thing as it compiled before. Maybe the dt build chain should check if this tool is available. I had to install the intltool package and it works again :-) Thank you, Michael Am 27.02.2014 09:43, schrieb johannes hanika: > [ 18%] Generating da

Re: [darktable-devel] build error in current git

2014-02-27 Thread johannes hanika
[ 18%] Generating darktable.desktop /bin/sh: intltool-merge: Kommando nicht gefunden. nothing to do with rawspeed, you're missing the above tool ^ -jo On Wed, Feb 26, 2014 at 9:48 PM, Michael Born wrote: > Hi developers and users, > > for a few days now I can not compile darktable anymore. > I

Re: [darktable-devel] build error in current git

2014-02-27 Thread Togan Muftuoglu
> "Michael" == Michael Born writes: Michael> Hi developers and users, Michael> for a few days now I can not compile darktable anymore. Michael> I tested it on my main computer and on the notebook (both openSuse 13.1 Michael> 64bit). I used to build the git version with both m