Re: [AGG] Blending colors, associated alpha, etc

[Stephan A. <sup...@gm...>]

Hi Maxim,

as a curious thing, there is virtually no difference if I use >> 8 or / 
255. I have double checked. I'm on Pentium-M, maybe divisions are not 
that expensive there?!?
As it turns out, my worst case blending (0 < top and bottom alpha < 
255) is much much slower than my usual case (bottom alpha == 255). So 
I'm really looking for an optimization of this piece of code:

t[1..3]	top color				on 0..25500
ta			top alpha				on 0..255

b[1..3]	bottom color			on 0..25500
ba			bottom alpha			on 0..255

d[1..3]	resulting color		on 0..255
da			resulting alpha		on 0..255


	uint8 alphaRest = 255 - ta;
	uint32 alphaTemp = (65025 - alphaRest * (255 - ba));
	uint32 alphaDest = ba * alphaRest;
	uint32 alphaSrc = 255 * ta;
	*d1 = kInverseGammaTable[(b1 * alphaDest + t1 * alphaSrc) / 
alphaTemp];
	*d2 = kInverseGammaTable[(b2 * alphaDest + t2 * alphaSrc) / 
alphaTemp];
	*d3 = kInverseGammaTable[(b3 * alphaDest + t3 * alphaSrc) / 
alphaTemp];
	*da = alphaTemp / 255;


Best regards,
-Stephan