[MPlayer-dev-eng] ATI + vidix scaler quality

[Nick Kurshev]

Hello, Alex!

On Sun, 7 Jul 2002 16:40:11 +0200 you wrote:

> <20020707162649.A21307 at localhost.localdomain.hu>; from alex at naxine.org on Sun, Jul 07, 2002 at 04:26:49PM +0200
> > Hi,
> > 
> > > Alex mentioned some time ago that vidix scaling on radeon looks ugly.
> > > Now i'v tested vidix on notebook with mach64, and the vidix scaling is
> > > simple nearest-neighbour type (YUY2) :(
> > > GATOS drivers (Xv) has nice scaling on the same machine, looks like
> > > bicubic or so.
> > hmm, i will look at the gatos source (as soon as i put my radeon back ;)
> gatos xv's default values are:
> OVERLAY_SCALE_CNTL	0x80000000
> OVERLAY_EXCLUSIVE_HORZ	0
> OVERLAY_EXCLUSIVE_VERT	0
> SCALER_H_COEFF0		0x00002000
> SCALER_H_COEFF1		0x0D06200D
> SCALER_H_COEFF2		0x0D0A1C0D
> SCALER_H_COEFF3		0x0C0E1A0C
> SCALER_H_COEFF4		0x0C14140C
> 
Everything what is related with these coeffs are described n radeon.h of vidix.
They are coeffs for SPLINE interpolation but nobody from opensource developers
has formula for these coeffs.
OTOH - the are disabled by default on rage128-radeon chips. (These chips are configured
to use HARDCODED coeffs).
Monitoring of windows driver behaviour show me that these coeefs are computed depend on
dimensions of source and dest rectangles.
> and for scaling:
> OVERLAY_SCALE_CNTL	0x4000001 | (3 << 30)
> 
> hm, i will immediately swap my cards, the gatos driver seems to be very perfect
> -
> alex
> _______________________________________________
> MPlayer-dev-eng mailing list
> MPlayer-dev-eng at mplayerhq.hu
> http://mplayerhq.hu/mailman/listinfo/mplayer-dev-eng
> 


Best regards! Nick
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/attachments/20020708/f404b24d/attachment.pgp>