[mythtv] mythtv-vid branch needs testing

Doug Larrick doug at parkercat.org
Mon Aug 13 11:56:40 UTC 2007


Stuart Auchterlonie wrote:
> 
> Doug Larrick wrote:
>> 2007-08-12 15:38:39.042 NVP: Video is 8.17415 frames behind audio (too slow), dropping frame to catch up.
>> 2007-08-12 15:38:39.047 NVP: Video is 14.0766 frames behind audio (too slow), dropping frame to catch up.
>> 2007-08-12 15:38:39.051 NVP: Video is 18.2674 frames behind audio (too slow), dropping frame to catch up.
>> 2007-08-12 15:38:39.055 NVP: Video is 21.1824 frames behind audio (too slow), dropping frame to catch up.
[snip]
>> 2007-08-12 15:38:39.620 AO: dropping back audio_buffer_unused
>> 2007-08-12 15:38:39.654 NVP: Video is 4.7699 frames ahead of audio,
>>                         doubling video frame interval to slow down.
>> 2007-08-12 15:38:39.700 AO: dropping back audio_buffer_unused
>> 2007-08-12 15:38:39.720 NVP: Video is 4.54508 frames ahead of audio,
>>                         doubling video frame interval to slow down.
>> 2007-08-12 15:38:39.759 AO: dropping back audio_buffer_unused
>> 2007-08-12 15:38:39.787 NVP: Video is 4.09324 frames ahead of audio,
>>                         doubling video frame interval to slow down.
>> 2007-08-12 15:38:39.820 AO: dropping back audio_buffer_unused
>> 2007-08-12 15:38:39.850 AO: dropping back audio_buffer_unused
>> 2007-08-12 15:38:39.853 NVP: Video is 3.47116 frames ahead of audio,
>>                         doubling video frame interval to slow down.
> 
> 
> And that's an example of the A/V sync code overcompenstating which
> causes stuttering of (alternately) audio & video. Although this is
> a problem which exists in head as well.

Though to be fair, on my system I mostly see this briefly when I begin a
recording or after a seek, and then it settles down.  My theory is that
after a seek we start playing whatever audio and video packets happen to
come along in the stream, regardless of their relative timestamps; in an
MPEG2-TS recording I think that audio is often quite a bit ahead of
video.  In this instance, it appears to be entering timestretch that
kicked it off; I have AC3 Passthru turned on so entering timestretch
would have had to start decoding audio frames.

-Doug

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 252 bytes
Desc: OpenPGP digital signature
Url : http://mythtv.org/pipermail/mythtv-dev/attachments/20070813/4271010a/attachment.pgp 


More information about the mythtv-dev mailing list