[MPlayer-dev-eng] [PATCH] try to enable XV_SYNC_TO_VBLANK

Rich Felker dalias at aerifal.cx
Tue Jun 28 04:26:13 CEST 2005


On Mon, Jun 27, 2005 at 04:51:39PM -0700, RC wrote:
> On Mon, 27 Jun 2005 12:14:24 -0400
> Rich Felker <dalias at aerifal.cx> wrote:
> 
> >  (most tvout devices do their own scaling, btw..), 
> 
> I don't know what you mean by that.  If you playback a 4/3 video with
> "-noaspect" on any Nvidia/ATI cards with TV-out, it will look just as
> squished on the TV as it would on a monitor.

This has nothing to do with what I was talking about. TV out scales
the WHOLE picture. Even if you set the video mode to 480 scanlines,
most crappy tvout devices will still resample the picture because of
overscan considerations.

> > The motion is indeed smoother; you'll have full 60fps. And the
> > aliasing is no worse than the aliasing already inherent in interlaced
> > video. 
> 
> If you were here, I could show you what I'm seeing first hand.  Since
> you're not, you'll probably have to take my word for it.  The 60fps
> video playback is rather jumpy, and any fine details (straight lines,
> glasses, pinstripes, etc) get pretty well chewed-up.  I can easily watch

This happens on the tv too. Your eyes just don't see it because it's
being refreshed so quickly.

> the actual channel, then flip over and watch mplayer's damaged output of
> the same segment.  Even someone with poor eyesight could tell you
> there's a problem.

Perhaps your monitor's refresh rate is bad. Unless it's a multiple of
60 it will look choppy. 120 is the best refresh rate for videophiles
to use since it's a multiple of 60 and 24.

> > Note that for tfields to work, you MUST preencode a 60fps file
> > and then play that. The filter will not work correctly at all if
> > called directly from mplayer.
> 
> Yes, I know.  I tried each of tfields=[0-4], and I even tried each with/
> without inverse telecining.  No improvement (not that I was expecting
> any).

2, 3 and 4 are the only ones that work.

Rich




More information about the MPlayer-dev-eng mailing list