[FFmpeg-devel] lavfi state of affairs

Michael Niedermayer michaelni
Sat Feb 7 03:28:11 CET 2009

On Fri, Feb 06, 2009 at 11:39:00PM +0000, Robert Swain wrote:
> > Imgconvert is _complete_, simpler, easier to code on, albeit a lot slower.
> To my eyes, swscale does look a mess. But, I haven't looked at it in
> depth and I haven't looked at it in about 2 years. And I'd rather have
> significantly faster code but much harder to understand code than not.

> I think my complaints about using the API can be easily solved by
> documentation. There's nothing to flame about here, it's just a case
> of doing it.

tell me what is unclear and ill see what i can do

> Had I joined this thread earlier, I might have been more aggressive,
> but as it is, I think we should focus on doing stuff and making
> progress so we can drop these deprecated APIs and transition to the
> new ones. TODO list anyone?

kostya is IIRC correctly going to rewrite the table generator this weekend
other things to do

split asm code out cleanly, maybe similar to dsputil maybe not
(its important that its clean, has a cleanly documented API betwen th
C/ASM on one side and the code calling it on the other, and of course
it must be fast)

split/put under #ifdef the optional GPL optimizations so swscale can be
compiled in a pure LGPL mode. (this is the only thing AFIAK except the table
generator that is needed for droping the old scaler)

see the related past gsoc project there where many other ideas of what
could be done

Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

There will always be a question for which you do not know the correct awnser.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090207/6e3ca999/attachment.pgp>

More information about the ffmpeg-devel mailing list