[Ffmpeg-devel] [RFC] VC1 in EVO support
Sat Feb 3 07:54:16 CET 2007
On Sat, Feb 03, 2007 at 12:12:41AM +0000, Ian Caulfield wrote:
> On 03/02/07, Michael Niedermayer <michaelni at gmx.at> wrote:
> >On Fri, Feb 02, 2007 at 08:40:49PM +0200, Kostya wrote:
> >> This patch enables decoding VC-1 stream (tested on MAININTRO.EVO)
> >> It consists of three parts:
> >> * support for initialization with extradata present in first frame
> >> * parser for reconstructing frames
> >> * a few hacks to make it work now (two conditions commented out)
> >> It also requires EVO demuxing support enabled (still pending,
> >> you can find the very patch at
> >> Please test and give tips on implementation as it's my first attempt
> >> to use parser and to make decoder work without extradata.
> >> If it works as supposed I will clean it and divide before commit
> >> Thanks
> >ive just quickly looked at it but it seems that you could solve the
> >extradata in first frame issue simpler by implementing
> >AVCodecParser.split() which simply identifies which part of the frame
> >is the extradata and then moves that into AVCodecContext.extradata
> >if the AVCodecParser.split() API is not flexible enough for EVO-VC1
> >then the API should be changed to be more flexible ...
> What happens if the stream parameters change during the stream? Will the
> codec pick up on the change in extradata?
Depends on what is changed. Entry points will be parsed. Sequence end and
new sequence header will be ignored (they may be supported in future but
now I see no need to do this).
More information about the ffmpeg-devel