[MPlayer-users] VDPAU errors, flickering black video display on aspect change in MPEG2 TS

[wm4]

On Sun, 8 Dec 2013 17:04:52 +0100
Reimar Döffinger <Reimar.Doeffinger at gmx.de> wrote:

> On Sun, Dec 08, 2013 at 04:54:21PM +0100, Reimar Döffinger wrote:
> > On Sun, Dec 08, 2013 at 04:41:23PM +0100, wm4 wrote:
> > > > If you're using FFmpeg multithreading I see no way you couldn't see
> > > > that the frames with the new aspect are decoded long before you can
> > > > in any way see the new aspect from your application.
> > > 
> > > My application separates decoder and VO init, though. The decoder is
> > > initialized in the get_buffer callback (which also pretty much
> > > guarantees that libavcodec's use is synchronized with the decoder).
> > 
> > Hm, that might be kind of luck, I don't know if it's actually specified
> > if get_buffer should return the main context the application has or the
> > thread-specific context.
> > I suspect you enabled that libavcodec assumes get_buffer callbacks are
> > thread-safe, because otherwise I'd suspect your solution would in fact
> > break/have exactly the same issue.

I disable multithreading entirely for hardware decoding, since it doesn't
seem to have advantages. If I enable threading anyway, it happens to work
and I still see no issues.

> > > So
> > > that's probably why it doesn't there. Now the question is, why doesn't
> > > it happen on Mesa with mplayer?
> > 
> > Why should it? An aspect change is not a resolution change, there is
> > no sane reason for it to fail.

I agree, and actually the vdpau API never gets the "aspect", just the
real resolution. The aspect is just a result of the display rectangle
passed to the video mixer, and vdpau allows arbitrary parameters here.

> Also, from my side I am wondering why you don't see the issue.
> I mean I can see how you might not have any issues reinitializing
> the decoder on aspect change, but you said you aren't doing that,
> so I would think you should still see the issue.
> Or did you mean you only skip reinitializing the VO on aspect change,
> but do create a new VDPAU decoder?

I still reinitialize the VO on aspect change, but not the decoder
(and VO reinit doesn't do much anyway in my code). The decoder is
reinitialized only on a real resolution change, which doesn't happen on
a mere aspect change. I think libavcodec does flush frames in this
situation (didn't test directly), but does nothing that would require
reinitializing the vdpau decoder.