[MPlayer-dev-eng] [PATCH]VO_VDPAU, round 6

[Reimar Döffinger]

On Tue, Feb 03, 2009 at 01:07:59AM +0000, Carl Eugen Hoyos wrote:
> Reimar Döffinger <Reimar.Doeffinger <at> stud.uni-karlsruhe.de> writes:
> 
> > On Tue, Feb 03, 2009 at 12:17:30AM +0000, Carl Eugen Hoyos wrote:
> > > I forgot to mention that I tested -fixed-vo -vo vdpau with different video
> > > format files with different resolutions before sending the last patch not
> > > seeing any difference to -vo gl and -vo xv.
> > 
> > With different resolution even? E.g. vid_width is not adapted to the new
> > value...
> > Did you try
> > mplayer file -aspect 0.4 file -aspect 2
> > and see if the aspect changes correctly?
> 
> I believe they don't change correctly (the first file uses "2", the second
> "0.4"), but since it is not different from -vo gl, I don't see anything wrong
> about it.

Sorry, I forgot this is handled in common code nowadays...

> > How about a file that uses hardware-decoding and then one which does not
> > use it but instead the YV12-features of the vo?
> 
> There is no hardware-decoding in my current patch and I believe you explicitly
> asked to remove every trace of it (like switch statements).

Yes, yes. I am trying to find code in config() that actually has a
function and find little.

> mplayer file -flip file -noflip
might have worked except that -flip is not supported at all, which makes
this:
> +    flip_flag = flags & VOFLAG_FLIPPING;
pointless.
Not sure if this works, but I think it is handled in common code, too:
> mplayer file file -fs
(first file should play normally, the second in fullscreen)
I can also say that
vdp_video_surface_create, create_vdp_mixer and vdp_video_mixer_create
are only ever called with the size of the first video.
This seems clearly wrong to me, but if it works (including OSD) when
you first play a small and then a much larger file, I can't really
"proof" it (though I then wonder a bit what these functions are needed
for anyway)...

Greetings,
Reimar Döffinger