[FFmpeg-devel] [PATCH] issue251, xvid within .ogm will not remux to .avi try2

[Måns Rullgård]

Michael Niedermayer wrote:
> Hi
>
> This patch fixes "according to spec" timestamp association for theora
> it also as a sideeffect fixes the file from issue 251
>
> quote of the theora spec: (A.2.2)
>    Frame data pages MUST be marked with a granule index corresponding to
> the display time of the last frame/packet that finishes in that page.
>
> Note ive not tested this extensively, i thought our users can do this better
>
> Ill apply this in a few days if i hear no objections
>
>
> Index: libavformat/oggdec.c
> ===================================================================
> --- libavformat/oggdec.c	(revision 13799)
> +++ libavformat/oggdec.c	(working copy)
> @@ -523,11 +523,25 @@
>          return AVERROR(EIO);
>      pkt->stream_index = idx;
>      memcpy (pkt->data, os->buf + pstart, psize);
> +    if (s->streams[idx]->codec->codec_id == CODEC_ID_VORBIS){
>      if (os->lastgp != -1LL){
>          pkt->pts = ogg_gptopts (s, idx, os->lastgp);
>          os->lastgp = -1;
>      }
> +    }else{
> +        int segp= os->segp;
> +        int nsegs=os->nsegs;
> +        while (segp < nsegs){
> +            if (os->segments[segp] < 255)
> +                break;
> +            segp++;
> +        }
>
> +        if (segp == nsegs && os->granule != -1LL){
> +            pkt->pts = ogg_gptopts (s, idx, os->granule);
> +        }
> +    }
> +
>      pkt->flags = os->pflags;

Can you please explain a little more why this change is needed, and
why it doesn't apply to Vorbis?

-- 
M?ns Rullg?rd
mans at mansr.com