[MPlayer-users] <vo_direct3d>Allocating offscreen surface failed

[Greg Bryant]

On 3/23/09, David Bolen <db3l.net at gmail.com> wrote:


> I believe current SVN versions should create the initial off screen
> surface at display size not video size (as of 1/27, r28379), so
> depending on how much earlier your current code base is, you might
> give a shot at a more recent version just for the heck of it.

Sorry, forgot to say that I updated that morning, so I am very svn
current (and missing random.h as of this morning on the linux box :).

>
> Can you see any other differences in the CreateOffscreenPlainSurface
> parameters between the two systems?

The only other interesting parm is a nasty structure that I didn't
feel like dumping out (since it's not a dev machine, no kdbg).

> I have trouble seeing it being a DVI/VGA difference.  Much more likely
> is that it's something to do with the D3D support related to the the
> VIA chipset driver compared to the driver for your other system(s)
> video chipsets.
>
> Have you tried running mplayer with verbose logging (-v) to gather a
> few more of the internal vo_direct3d messages?  Maybe it has something
> to do with the format mapping that goes on between media format and
> display format - e.g., maybe the VIA chipset D3D driver doesn't
> support the same formats, which prevents prev->movie_src_fmt from
> being established properly, which then becomes an invalid parameter to
> the offscreen surface creation call.
>
> Otherwise, not sure what else to suspect.  Perhaps you'll be better
> off with directx and/or gl on that system?

I tried -vo gl, and it worked fine.  The -v doesn't help much - it
shows everything working up to the offscreen surface.  It does show
the backbuffer being allocated at screen size (and working), then the
create_d3d_surfaces, then the failing call to createoffscreensurface.

I'm down to something funky with the via chipset, and we'll just go
with OpenGL on that box.

Thanks for the pointers, I hadn't dug through the vo lib at all before.

Greg