[Ffmpeg-cvslog] Re: r5598 - in trunk: Changelog MAINTAINERS doc/ffmpeg-doc.texi libavformat/Makefile libavformat/allformats.c libavformat/allformats.h libavformat/gxf.c

Reimar Döffinger Reimar.Doeffinger
Wed Jul 5 18:57:14 CEST 2006


Hello,
On Mon, Jul 03, 2006 at 06:55:58PM +0200, Baptiste Coudurier wrote:
> reimar wrote:
> > [...]
> > +        case 13:
> > +        case 15:
> > +            st->codec->codec_type = CODEC_TYPE_VIDEO;
> > +            st->codec->codec_id = CODEC_ID_DVVIDEO;
> > +            st->codec->codec_tag = MKTAG('d', 'v', 'c', ' ');
> > +            break;
> > +        case 14:
> > +        case 16:
> > +            st->codec->codec_type = CODEC_TYPE_VIDEO;
> > +            st->codec->codec_id = CODEC_ID_DVVIDEO;
> > +            st->codec->codec_tag = MKTAG('d', 'v', 'c', 'p');
> > +            break;
> 
> I don't know if it is "right" to assign those tags to dv50. I suppose
> tags are there to help demux_lavf.c to assign correct codecs, am I right
> ? I know lavc dv decoder can probe dv50 on its own, but still...

Well, also for remuxing into another format I'd guess. What would be the
alternative? I'm open to suggestions, esp. as I do not have any sample
files for that format.

> Also, would it be better to assign CODEC_ID_NONE and CODEC_TYPE_DATA to
> unrecognized streams ? Instead of -1.

?? CODEC_ID_NONE and CODEC_TYPE_UNKNOWN is currently assigned. It seemed
like the most appropriate choice. Why would CODEC_TYPE_DATA be better?

Greetings,
Reimar D?ffinger




More information about the ffmpeg-cvslog mailing list