[MPlayer-users] HD mpeg transport stream native decoder regression.

[Reimar Döffinger]

On Fri, Jan 16, 2015 at 12:47:44AM +0000, Andy Furniss wrote:
> Andy Furniss wrote:
> >This is a crap report as I haven't had time to do anything yet but
> >I'll mention it anyway.
> >
> >UK HD TV transport streams have a playback issue, I suspect due to
> >recent changes, but trying to rest to say r37331 gets a build fail.
> >
> >verbose output would be very long with mplayers ts logging.
> >
> >Symptoms - the timing is wrong = slow motion but not 100%
> >consistently slow.
> >
> >Issue does not happen with -demuxer lavf or h/w decode.
> 
> It turned out this is an ffmpeg regression not demuxer (though ffplay
> doesn't show it).
> 
> The behavior above only applied to multithread decode, with single
> thread the slow mo went away and both native and lavf demuxers showed
> different symptoms = some out of order fields.
> 
> bisecting ffmpeg came up with -
> 
> fe439c20698fc4992b9373cb068efb438fe1a098 is the first bad commit

Did you report this to FFmpeg?
Admittedly since it doesn't show up with ffplay (though did you enable
multithreaded decoding for it and set the SHOW_ALL flag?) it might
be a bit hard to get it accepted, but I don't think I have time
to debug myself currently.