[FFmpeg-devel] FFmpeg 2.1
nfxjfg at googlemail.com
Mon Sep 9 00:26:34 CEST 2013
On Sun, 8 Sep 2013 20:22:39 +0200
Reimar Döffinger <Reimar.Doeffinger at gmx.de> wrote:
> On Sun, Sep 08, 2013 at 05:34:11PM +0200, Michael Niedermayer wrote:
> > Hi all
> > the FFmpeg 2.0 release is already 2 month ago so it soon becomes
> > time for another release. My plan is to make the release within the
> > next 4-6 weeks or so
> > If you want something in, dont hesitate to work on getting it in
> I'd like to get VDPAU (using the new API) working properly, can
> someone help out?
> I can't even test it properly unfortunately.
> The issues I know about:
> 1) packed B-frame MPEG-4 videos seem to decode without B-frames.
> Possibly just adding a "s->mb_y = s->mb_height;" to the hwaccel case
> of decode_slice will work.
> Am I correct that mb_y != mb_height at the end of that function should
> only happen in case of errors?
> Either way there is quite a mess, for example vaapi_mpeg4 has this code:
> > if (avctx->codec->id == AV_CODEC_ID_H263)
> > s->mb_y = s->mb_height;
> However that means for != AV_CODEC_ID_H263 mb_y will stay 0 _forever_,
> which seems like completely nonsense.
> Could it be that all this was implemented by pure trial and error?
> 2) I have had reports of decode artefact with VC1 which only happen
> with the new decoder code. Unfortunately I couldn't reproduce it so far.
> This might be because I either miss the right sample, or because or
> try running on AMD hardware (which in turn seems to have hardware hangs
> when trying to decode WMV3, but for the moment I assume that is not
> a FFmpeg issue).
3) Some mpeg files fail to decode correctly with the libavformat
demuxer, but work correctly with mplayer's mpeg demuxer. (Trac
Well ok, that got nothing to do with the new vdpau API, just vdpau
in general (and possibly other hwaccels).
More information about the ffmpeg-devel