Nathan Bradshaw wrote:
>
> At 2001-06-05 11:32:58, Justin Schoeman <justin@xxxxxxxxxxxxxxxxxxxx>
> wrote:
>
> > After many comments/complaints about my existing capture engines, I
> > finally sat down and wrote a new one. After two evenings of furious
> > coding, I have a very alpha program just waiting for a few suckers to
> > test. There are a number of problems documented in KNOWN_BUGS (and
> > probably quite a number that aren't too!).
> >
> > The capture core features perfect A-V synchronisation, irrespective of
> > frame drops/video loss, etc. It also has dynamic time warping of the
> > audio stream to ensure that EXACTLY the right amount of audio data is
> > written for the number of video frames (at least averaged over a couple
> > of frames). This removes the problem of synch drift over long
> > recordings (without any clicks/pops).
>
> Hi Justin,
>
> Your code is the first I have used under Linux that will synch
> audio and video properly during capture! That's great, but I do have one
> problem using it. After a few minutes of capture, I get the following
> error from DIVXrec:
>
> VQ: 29.476 Offset: 0.005 Slipage: 0.003 Warp: -7 Drop: 0
> v4l2 core next - new frame out of sequence - ignoring-7 Drop: 0
> NVrec: error getting more video data. Trying to continue.
> v4l2 core next - new frame out of sequence - ignoring-7 Drop: 0
> NVrec: error getting more video data. Trying to continue.
> v4l2 core next - new frame out of sequence - ignoring-7 Drop: 0
> NVrec: error getting more video data. Trying to continue.
> v4l2 core next - new frame out of sequence - ignoring-7 Drop: 0
> NVrec: error getting more video data. Trying to continue.
> v4l2 core next - new frame out of sequence - ignoring-4 Drop: 0
> NVrec: error getting more video data. Trying to continue.
> [...loops a million times until I hit Control-C...]
>
> At this point it stops capturing video but keeps on capturing audio.
> The resulting DivX AVI file is valid, but the video freezes at this
> point.
>
> My load never goes above 0.5 on this P3-833 768Mb RAM. The capture
> file is on a separate (fast) disk, and I can't really find any system
> bottleneck that might be interfering with a successful capture. If I
> do create large load on this box, the DIVXrec program drops a few
> frames and continues OK (until the above error).
Oops a slight logical error there (I think...), try edit v4l2_core.c,
and change
} else if(vct->new_seq < vct->last_seq)
to
} else if(vct->new_seq <= vct->last_seq)
Otherwise the first time it looses sequence, it will carry on until it
looses again! Sorry for the (slightly stupid) error.... Please try this
fix and let me know if it works for you.
Thanks,
-justin
P.S. I will make a new release on Monday, with this fix, and a slightly
cleaner, more capable, architecture.
