[MPlayer-dev-eng] a-v sync fixes/cleanup backported from g2
D Richard Felker III
dalias at aerifal.cx
Fri Aug 8 00:54:32 CEST 2003
On Thu, Aug 07, 2003 at 09:55:18PM +0200, Arpi wrote:
> > On Thu, Aug 07, 2003 at 09:35:18AM -0700, Mark Guptill wrote:
> > > On Tuesday 05 August 2003 03:32 pm, Arpi wrote:
> > > > Please tets it, before commit. I think it's ok, but should be tested
> > > > on very slow system (with and without -framedrop, i wanna know if A-V
> > > > starts drifting away from 0 then).
> > >
> > > i tested your patch with a mpeg2 video without framedrop A-V sync was reported
> > > to be in sync but the video was actually slower than the audio. with
> > > framedrop A-V sync was in sync. logs are attached.
> > I haven't tested this patch, but this might very well be the case if
> > it makes behavior more like -autosync. With -autosync enabled, mplayer
> > always reported A-V: 0.000. Actually this was kinda nice, since
> > mplayer didn't massively overshoot in correcting it and make A-V:
> > -5.xxx, but it's annoying from a user perspective...
> Hmm, I didn't know that -autosync was already so.
> Anywat that is the expected behaviour, as A-V sync value reports
> the streams sync, not the audible delay. We should do teh same as in
> g2, so either display 2 values (streams and audible sync) or fake it,
> ie. add timer underflow and inaccuracy values to the displayed A-V delay.
> At least it have an advantage: we can split the YourSystemIsTooSLow
> message to multiple parts, depending on it's really too slow sys
> (timer underrun), buggy soundcard (inaccurate timing), or broken file
> (streams desync).
Thank goodness... If you add that, can you also add
-disable-lame-system-too-slow-message to cfg-mplayer.h? :) Knowing
about the other two situations might be useful, but if my system is
too slow, I probably already know that, and the last thing I need is
lots of ugly spam scrolling the useful console messages away...
More information about the MPlayer-dev-eng