[MPlayer-dev-eng] [PATCH] Make vo_gl / fragment programs work with Mesa

[Matthias Hopf]

On Mar 11, 07 17:35:07 +0100, Reimar Döffinger wrote:
> > How about attached patch instead? It adds an interpolator that is
> > cubic only in X-direction. Doesn't look too bad IMO.

I thought a bit about the theory. I don't think we are actually allowed
to call that "interpolation" - because when interpolating the values at
sample positions must remain the same. Because we cannot have negative
filter weights, no higher-order interpolation is possible (higher than
the order of 1, that is linear).
We'd better call that "bicubic filtering", because it *does* smooth out
details.

> Applied.
> How about attached patch that avoids the extra lookup texture? (you
> still need to "fix" two swizzles).
> It is a lot more instructions more and thus a lot slower on my hardware
> of course?

Probably.
Looking at the size of the code, this is presumably a really bad idea.

I proposed a different solution, making the texture lookups for the
weights non-dependent, so they don't count for the maximum of 4
dependent texture lookups. We can do so by adding to more interpolants
(fragment.texcoord[] or similar), because the code on does a linear
transformation on fragment.texcoord[%c], which can be done before
interpolation.

I'll do that eventually, but right now I'm pretty swamped with bugzilla,
and there are more pressing issues for me in mencoder :-]

CU

Matthias

-- 
Matthias Hopf <mhopf at suse.de>      __        __   __
Maxfeldstr. 5 / 90409 Nuernberg   (_   | |  (_   |__          mat at mshopf.de
Phone +49-911-74053-715           __)  |_|  __)  |__  R & D   www.mshopf.de