[MPlayer-users] Re: tdfxfb scaling bug

Alban Bedel albeu at free.fr
Tue Sep 16 10:28:06 CEST 2003

Hi Georgi Petrov,

on Mon, 15 Sep 2003 09:54:38 -0700 (PDT) you wrote:

> [Automatic answer: RTFM (read DOCS, FAQ), also read
> DOCS/bugreports.html] First I want to say that I __HAVE__ been RTFMing
> last 4 days. I have been also looking at the sources for
> hours, trying to locate the problem, but I can't. 
> The problem: when using tdfxfb as output driver, it
> must scale the image (when -fs or -xy 640 for example
> are used). And this must be hardware scaling. BUT when
> I see the upscaled image, it appears less than
> optimal. For example **on the same machine, same
> resolution, same movie** when I use -vo xv,
> everything's alright. The scaling is perfekt, but when
> I look at tdfxfb's output, I have the impression that
> the image is scaled by cheap software method. Of
> course this is not the case, because SwScaler stays
> turned off and everything's done by the tdfxfb driver.
> This is _very_ important thing to fix, because not
> everybody can see it and I think that it IS a bug.
> Just compare tdfxfb and xv output when upscalling
> small, say, 300x200 image to 640x480 and then xv's
> output.
This don't think this is a bug. This is rather a limitation of the
banshee hardware. Afaik the sacling can only be filtered when using the
overlay. tdfxfb doesn't use the overlay but some 2d operations, however
xv probably use the overlay.
Now if xv is too slow for you can try tdfx_vid. It use the overlay and
is a lot faster (even faster than tdfxfb in fact). But it requiere an
AGP card along with an AGP chipset supported by the kernel.

Everything is controlled by a small evil group
to which, unfortunately, no one we know belongs.

More information about the MPlayer-users mailing list