[FFmpeg-devel] [RFC] libmpcodecs libavfilter wrapper?

Michael Niedermayer michaelni
Thu Nov 25 20:41:04 CET 2010

On Thu, Nov 25, 2010 at 08:20:06PM +0100, Luca Barbato wrote:
> On 11/25/2010 05:03 PM, Michael Niedermayer wrote:
> > And i told already that such wraper needs to be written by someone who has
> > more experience with porting code between projects. For such person it should
> > not be much pain to do it.
> > You surely will have to agree that mplayer is able to use the filters from
> > libmpcodecs.
> > Now you can surely take vf_*.c from libmpcodecs and check that into
> > libavfilter/libmpcodecs
> > and then write the surrounding code to translate libavfilter API to libmpcodecs
> > API.
> > You dont want to implement it that way maybe, i dont know but iam not aware of
> > any problem in this approuch
> There isn't a problem per-se beside the fact mplayer codebase is quite
> scattered and the faster way to achieve a wrapper might be fix the
> missing function/struct one by one through the gcc output...
> > Also ill wager a bet that arpi could implement a wraper in less than a day
> > arpi has very extensive experience with porting code between projects and he
> > wrote large parts of libmpcodecs.
> 24h? Probably, maybe 30h and I guess we'd have some restriction in its
> usage.

i think we should ask arpi, the foundation could pay him for it

> > That didnt work out as noone was willing to do it so now after 3 years
> > we are better in some ways (arbitrary filter graphs) than libmpcodecs and
> > still worse in others (like actual number of supported filters)
> Well now we can improve adding filters in the slower way =).
> That said _probably_ would be faster having libmp using the avfilters
> and compare that way.

libmpcodecs is not able to support all libavfilters as libmpcodecs does only
support 1 input and 1 output link. That said yes its easy to add a libavfilter
support in libmpcodecs. Ideally we should support not just 1 filter but a full
libavfilter graph in libmpcodecs, that way mplayer could within limits use
more complex graphs as long as the whole graph has only 1 input and 1 output

Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The educated differ from the uneducated as much as the living from the
dead. -- Aristotle 
-------------- 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/20101125/87e91a3a/attachment.pgp>

More information about the ffmpeg-devel mailing list