[MPlayer-cvslog] r28546 - in trunk/libvo: video_out.cvideo_out.hvo_direct3d.cvo_xv.cvo_xvmc.c

[Carl Eugen Hoyos]

The Wanderer <inverseparadox <at> comcast.net> writes:

> >>> To share a fast guess: You need
> >>> Option  "Composite"     "Disable"
> >>> in the Extensions section of xorg.conf
> >> 
> >> I'll have to remember to try that when I next restart X, which
> >> probably won't be soon unless I need to update drivers to try to
> >> address the fullscreen problem. (I can, at least, confirm that
> >> Composite is available in my currently-running session.)
> > 
> > Well, I just crashed X while testing a hypothesis with -vf expand and
> > -vo vdpau, so I've tried this already; it does seem to fix the
> > problem.

I'd like to add the following:
I never had the Option "Composite" in my xorg.conf, but knew about possible
issues from different forums.
Out of curiosity, I added Option "Composite" "Disable" to my xorg.conf, which
produces an incredible amount of tearing for me if I use hardware accelerated
decoding.
I added Option "Composite" "Enable" to my xorg.conf now, and everything works
again (occasional, not-always-easy-to-see tearing on HD material).

> Clarification: it fixes the "looks like a refresh-rate problem" issue,
> not the fullscreen-playback issue.

As expected.
I just tried the following with latest svn on my 1920x1200 (16:10) monitor:
(I added a symlink to point ~/.mplayer/subfont.ttf to a random truetype font on
my system to silence warnings with -noconfig all)
I ran mplayer -noconfig all harrypotter.mov (1920x800) with and without
fullscreen, with and without panscan: Whole picture shown as long as I don't use
panscan which works as expected.
I ran mplayer -noconfig all dvb:// (768x576) and again had no problems with
fullscreen or panscan.

I use nvidia drivers 180.29

Carl Eugen