Splitstream and constant sequence issues
cerevant@y...
cerevant at y...
Thu, 18 Oct 2001 13:21:13 -0000
--- In ExtractStream@y..., joe666boxer@y... wrote:
> A few bad chunks are always there. I think extractstream is
> overzeleous and extracts more than it should, especially near power
of
> 2 chunk numbers (which is consistent with the way data seems to be
> organized on the tivo).
Sounds reasonable...
> > Stream 2:
> > 1st convert sequential failure
> > - Audio 22k, Video 284k
> > 2nd convert *also* had a sequential failure.
> > - Audio 6.8M, Video 125.7M
> > 3rd convert again, seq fail
> > - Audio 19k, Video 259k
> > 4th starts with missing first frames & audio failure, then seq
fail.
> > - Audio 6.8M, Video 125.5M
> >
>
> This is the case we have to really concentrate on. My tivo doesn't
> seem to create any of these.
*sigh* Of course I concentrated on the other :)
> I found that reseting the sequence number does more
> harm than good. I don't reset it precisely because of the
> additional chunks that extractstream throws into the mix.
> So if there is a sequence: v1, v2, v3, vxxx, v4, v5, ...,
> which happens alot, we'll just ignore vxxx and
> keep on going on our merry way. If you reset sequences,
> then you're liable to getting bad data into the stream,
> which more often than not, crashes players or causes very
> noticable video/audio problems.
I was sure this was the case, but is there some way we could validate
the sequence numbers at the beginning of the next chunk? Maybe if
you get a bad audio sequence, keep checking the video sequences in
the chunk (still leaving it out of the output) until you get to the
next chunk? Maybe save the last audio & video record in a chunk
(even if the chunk is discarded) for validation of the beginning of
the next chunk? I know we are getting into some pretty ugly logic
here...
> Let's continue this in the other post where posted the results of
> displaystream.
Indeed :)