Re: [Gimp-developer] wavelet decompose

[Miroslav Talasek <miroslav talasek seznam cz>]

this implementation spend a lot of time, i cant see to end of way.
Everything will be perfect but it will never happen.
I mean that would be better to go through iterative way.

1. Patch GEGL to multiple output
2. Write operation Wavelet decompose with build in-algo
3. Make generic wavelet code for share betwen ops, polish .... etc

GIMP may grow up  faster i mean

So which IRC we use to talking about this ?



On 12/13/2012 09:22 AM, Alexandre Prokoudine wrote:
> I would suggest joining IRC channel and discuss it there :)
>
> Øyvind is so much easier to talk to online :)
>
> OTOH, I would suggest to stop for a moment and think about bigger
> picture. What GEGL really needs is some sort of generic code for
> wavelets and operations in frequency domain (I would suggest having a
> look at recent versions of darktable and edge-avoiding wavelets too).
> So that new ops, whether decomposition, noise removal or anything
> else, could use that generic code.
>
> (As a side note, in 2009 or so we had a GSoC project for operations in
> frequency domain that we couldn't merge because of a license mismatch
> with FFTW.)
>
> Of course, it's a lot larger task, but also a lot more rewarding in
> terms of future development. Maybe porting the plug-in to a GEGL op
> could be a start towards that.
>
> Alexandre Prokoudine
> http://libregraphicsworld.org
>
> On Thu, Dec 13, 2012 at 12:10 PM, Ville Sokk<ville sokk gmail com>  wrote:
> > You could do an op that has one output and call it repeatedly from a
> > gimp plugin. But an even better idea would be to consult pippin and
> > see if you can change GeglNode and GeglOperation to support multiple
> > outputs. There's even a FIXME comment in gegl-node.c so someone has
> > thought about this already.
> _______________________________________________
> gimp-developer-list mailing list
> gimp-developer-list gnome org
> https://mail.gnome.org/mailman/listinfo/gimp-developer-list