On Tue, Nov 22, 2011 at 12:35:20PM +0100, Janne Grunau wrote:
> On Tue, Nov 22, 2011 at 12:16:19PM +0100, Anton Khirnov wrote:
> > 
> > On Tue, 22 Nov 2011 12:01:33 +0100, Janne Grunau <[email protected]> 
> > wrote:
> > > On Tue, Nov 22, 2011 at 07:19:55AM +0100, Anton Khirnov wrote:
> > > > Also simplify the code so it's the same for video/audio/subs.
> > > > ---
> > > >  avconv.c |   10 ++++------
> > > >  1 files changed, 4 insertions(+), 6 deletions(-)
> > > > 
> > > > diff --git a/avconv.c b/avconv.c
> > > > index 3ba8534..b30dad4 100644
> > > > --- a/avconv.c
> > > > +++ b/avconv.c
> > > > @@ -1738,6 +1738,8 @@ static int transcode_video(InputStream *ist, 
> > > > AVPacket *pkt, int *got_output, int
> > > >                                  decoded_frame, got_output, pkt);
> > > >      if (ret < 0)
> > > >          goto fail;
> > > > +    pkt->data += ret;
> > > > +    pkt->size -= ret;
> > > 
> > > I think it was assumed that there's only a single frame per packet for
> > > video streams. This doesn't harm if it's still the case and it makes
> > > sense to assume the same behaviour for all decode functions.
> > > 
> > 
> > I think the API implies that if only a part of the packet is consumed,
> > then the rest should be passed to the next decode call, as is done for
> > audio.
> > If any decoders don't expect this, then I think it makes more sense to
> > fix those decoders.
> 
> I think you misunderstood my remark. I was just saying that no demuxer
> produces video packets with more than 1 frame (not sure how packed
> b-frames in avi are handled). So this is a situation which will not
> occur. Decoders can probably handle it, timestamps might be a problem
> but nothing to worry about since this it's safe to assume that it does
> not happen.

There is such thing as PB-frames in H.263++ with both P and B-frame bundled
together but that beast is extremely rare.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to