[FFmpeg-devel] [PATCH] Add negate filter.

Aurelien Jacobs aurel
Mon Nov 29 14:54:06 CET 2010


On Mon, Nov 29, 2010 at 11:35:41AM +0100, Stefano Sabatini wrote:
> On date Monday 2010-11-29 02:11:23 -0800, Jason Garrett-Glaser encoded:
> > On Mon, Nov 29, 2010 at 2:06 AM, Stefano Sabatini
> > <stefano.sabatini-lala at poste.it> wrote:
> > > On date Monday 2010-11-29 00:00:06 -0800, Jason Garrett-Glaser encoded:
> > >> Just a thought here -- there's a practically infinite number of simple
> > >> filters like this (i.e. filters for which every pixel is operated on
> > >> independently). ?Would it not be better to port something like
> > >> Avisynth's MT_LUT filter, which allows the user to specify a
> > >> mathematical operation to create a LUT from, which is then applied to
> > >> all pixels?
> > >
> > > Uhm, something like the applyfn filter I posted yesterday?
> > 
> > Doesn't use a LUT as far as I can see, so it's way too slow to be
> > usable for much of anything.
> 
> It can't use a lookup table because the computed value is a function
> of (X,Y), which allows interesting applications,

s/It can't ... because/It can't ... when/

What I mean is that the applyfn filter should use a LUT unless the
function provided by the user contain X or Y.

> but we could write another variant of the filter.

I'm not sure exactly what you mean but I feel like you want to write a
new filter with every small variation (eg. rotate vs. rotate90).



More information about the ffmpeg-devel mailing list