[MPlayer-dev-eng] [PATCH] improve quality of OpenGL output

[Reimar Döffinger]

On Sat, Sep 12, 2009 at 08:05:56PM +0200, wbrana at gmail.com wrote:
> > When does it make any difference? Even the worst precision most GPUs
> > have to offer should be sufficient for almost all calculations.
> 
> I expect sound and image quality should be best possible if there
> isn't any reason for worse quality.

You have given neither proof nor argument that quality is any worse.
Also "faster speed" is not really "not any reason". If it's not any
faster it is reasonable to expect for the graphics driver to generate
the same code for both.

> I tested FullHD videos. GPU temperature didn't increase more than usually.

Probably your graphics card isn't older than 4 years and thus it
probably makes no difference at all, not even in the generated code.
I think around GeForce FX/Radeons before the "HD" variants might be the
last ones where it makes any difference at all (I have not tested much
with these, but I was unable to make out a visible/measurable difference
even with these though - and I have seen quite a few graphical issues,
all of which were graphics driver bugs and this option had no effect at
all on them).

> Is there any reason for using fastest instead of nicest?

I already said that: even the worst precision available should
be more than sufficient for any calculations, so optimizing for
quality even in the best case makes no difference, in the worst
case could be a good deal slower.
If that assessment is wrong, I'd be happy to hear about it.
You did not answer my question though, which is in short:
Is there any reason for using nicest instead of fastest?