[FFmpeg-user] question about "Buffering several frames is not supported. Please consume all available frames before adding a new one."

littlebat dashing.meng at gmail.com
Sat Apr 2 09:48:16 CEST 2011


On Fri, 1 Apr 2011 22:11:25 -0400
Leo Izen <leo.izen at gmail.com> wrote:

> On Fri, Apr 1, 2011 at 9:44 PM, littlebat <dashing.meng at gmail.com>
> wrote:
> 
> > On Fri, 1 Apr 2011 17:10:58 +0200
> > Nicolas George <nicolas.george at normalesup.org> wrote:
> >
> > > Le duodi 12 germinal, an CCXIX, Stefano Sabatini a écrit :
> > > > Can you share the sample (and possibly file a bug-report)? That
> > > > would help.
> > >
> > > It's just the way output-based seeking (-ss after -i rather than
> > > before) is implemented: the frames are decoded but not encoded.
> > > Obviously, now that there are filters, they are fed to the input
> > > of the filter chain but not taken out.
> > >
> > > Any sample with video should show the problem. Even -f rawvideo -s
> > > 42x42 /dev/zero does, as long as there is a -ss after the -i.
> > >
> > > Regards,
> > >
> > > --
> > >   Nicolas George
> >
> > Put -i after -ss solved the problem. thanks.
> >
> 
> I still recommend putting -ss after -i. If you put it before, the
> audio and video will almost always get misaligned. The error isn't
> really an error, so it should be fine.

Add "-async 1" will avoid syncing problem of audio and video in my
case, for example: 
ffmpeg -ss 02:40:00 -t 00:10:00 -i yyzs.rmvb -async 1 yyzs17.mp4
Although I don't understand "-async" parameter clearly.

On my Core(TM)2 Duo CPU E7400 machine, if I put -i before -ss I will
spend about 11 minutes to finish the job above while only 1 minutes by
putting -ss before -i.

littlebat


More information about the ffmpeg-user mailing list