r200 xvideo gamma curve

[Edgar Toernig]

Rene Rebe wrote:
>
> Disclaimer: I have no SPEC as well.
> 
> Maybe the effect occurs when someone has xgamma of e.g. 1.8 and the xvideo 
> gamma is always fixed to 1.0 (that is the xvideo overlay gamma is independant 
> of the non-overlay gamma LUT)?

Well, my xgamma is 1.0.  And yes, xvideo gamma is independant of the
regular clut gamma (at least in the r200 hardware).

> Maybe both LUT should be kept in sync in that case?

Hmm... if I had the choice, I would prefer independantly settable
gammas.  If I want to keep them in sync I can already do that.
On some hardware (i.e. hardware which implements xv via yuv textures),
the xv gamma may already be coupled to the regular clut gamma.
But I guess, on those systems the xv-gamma setting is not available
anyway.

However - the patch was more about tweaking the hardware implementation
of the gamma-1.0-curve of the r200 driver.  It seems, with the current
curve the color resolution of the dark areas is severely limited.  Smooth
gradients turn into stripes of distinct colors (you get the same effect
when reducing the bit-depth of smooth gradients without dithering).
With the patched curve, the xv output looks the same as the output
produced from mplayer's yuv->rgb converter. 

Ciao, ET.