[FFmpeg-devel] [Patch] [avfilter] refactor floating point based interpolation and introduce in vf_lenscorrection

[Michael Niedermayer]

On Wed, Aug 20, 2014 at 01:00:29PM +0200, Daniel Oberhoff wrote:
[...]

> > 
> > 
> >> 
> >>> of course if someone wants to use the filter as a quick correction
> >>> for a random video downloaded somewhere thats likely in YUV space
> >> 
> >> For me going to YUV made everyting around 4x faster..
> >> 
> >> Anyhow, this starts to feel a little like “perfect is the enemy of good” and would lead me to drop this refactoring and instead hack interpolation into vf_lenscorrection directly...
> > 
> > what you write sounds a bit like you try to avoid making any
> > improvments to the filter. Not sure i understand that
> > 
> > Of course its work to add rgb support, of course its work to get
> 
> I can do this, I simply did not understand that all this was about that.
> 
> > gamma correction right, of course its work to implement a fixed point
> > bicubic filter, …
> 
> this will not help me at all, since the filter can not use fixed point. I would rather leave this to someone who is more motivated.

patchset posted that removes floats from the filter and makes it
fixed point

i think its better if further work is based on it and doesnt introduce
more floats, that is if my patchset is ok for you ?

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Why not whip the teacher when the pupil misbehaves? -- Diogenes of Sinope
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: Digital signature
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20140820/01917ce2/attachment.asc>