[vlc-devel] Questions about VLC's avcodec video decoder

[Etienne Membrives]

Hi,

I'm working, with two other French students, to build a VDPAU decoder
for VLC. We are building this decoder upon the ffmpeg VDPAU support, and
we have run into the VLC avcodec, which we want to use as a work base.

Thus, we have a few questions about video decoding, and more
specifically about the management of references frames in
modules/codec/avcodec/video.c :
- How VLC knows about which frames are referenced, and which aren't ?
We've seen that decodec_(un)LinkPicture seems to manages that, but how
and when these functions are called ?
- What corresponds to the "direct rendering" mode ? It should be a mode
where libavcodec passes directly to video output the picture data, but
which functions of modules/codec/avcodec/video.c are called in this mode
and which aren't ? And when "direct rendering" is deactivated ?
- It may be linked to the previous question : what are doing
ffmpeg_GetFrameBuf and ffmpeg_ReleaseFrameBuf functions ?

Then a more general question :
- When we are filling-in decoded data into a picture_t structure, why
there is different strides for source and destination in a vast majority
of codecs ?

Any help or hint would be greatly appreciated.

Thank you,
-- 
Etienne Membrives
Kevin Keraudren, Tilo Van Ekeris

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.videolan.org/pipermail/vlc-devel/attachments/20090418/5642fb76/attachment.sig>