[FFmpeg-devel] [PATCH] yadif port to libavfitler

[Michael Niedermayer]

On Sat, Sep 25, 2010 at 05:09:23PM +0200, Stefano Sabatini wrote:
> On date Saturday 2010-09-25 15:56:02 +0200, Michael Niedermayer encoded:
> > On Sat, Sep 25, 2010 at 11:11:12AM +0200, Stefano Sabatini wrote:
> > > On date Wednesday 2010-09-08 17:50:59 +0200, Stefano Sabatini encoded:
> > > > On date Saturday 2010-09-04 12:45:08 +0200, Stefano Sabatini encoded:
> > > > [...]
> > > > > Updated against the av_get_cpu_flags() patch.
> > > > > 
> > > > > Still missing documentation (Baptiste feel free to continue to work on
> > > > > this, I don't even know how did you test it).
> > > > 
> > > > Updated again. Documentation for the second param should be extended,
> > > > (Mplayer docs is outdated as well), also maybe ASM code should be
> > > > placed in /ARCH dirs.
> > > > 
> > > > I'll let someone else work on this though, so don't expect other
> > > > updated patches from me.
> > > 
> > > Updated, I moved the x86 code to a new libavfilter/x86 dir, and added
> > 
> > now it doesnt compile anymore
> > libavfilter/x86/yadif.c: In function ?ff_yadif_filter_line_mmx?:
> > libavfilter/x86/yadif.c:86: error: ?YADIFContext? undeclared (first use in this function)
> > libavfilter/x86/yadif.c:86: error: (Each undeclared identifier is reported only once
> > libavfilter/x86/yadif.c:86: error: for each function it appears in.)
> > libavfilter/x86/yadif.c:86: error: ?yadif? undeclared (first use in this function)
> 
> Fixed.
> 
> BTW, can someone suggest some way to test it? What about adding an
> interlacer, libmpcodecs/tinterlace would be a good option?

your code looks great, but its broken, dont bother further ive a fixed version
locally, ill commit it after some testing (and possible fixing).
We also need a pts interpolation filter for the case where a frame is produced
for each field (this case doesnt work in your code but in mine ;)
and its related to the missing continue_buffered_image() and uau commiting
trash to vf_yadif.c in MPlayer, so i dont blame that people didnt port
continue_buffered_image() that code borders on insane

of course if you want to work on tinterlace that is welcome

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

He who knows, does not speak. He who speaks, does not know. -- Lao Tsu
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20100925/3e0628e0/attachment.pgp>