__int16 color16;
__int8 red = (color16 & (0x1f << 11)) >> 8;
__int8 green = (color16 & (0x3f << 5)) >> 3;
__int8 blue = (color16 & (0x1f << 0)) << 3;

...beware the 5-6-5 16 bit cards...
...use the ddpixelformat struct to find out where the bits are supposed to go/how long they are...
something along the lines of:
_int32 rmask = ddpixelformat.???RMask (don''t know the exact name, sorry.)
int i =0;
while(rmask & 1 != 1)
{
rmask >> 1;
i++:
}

while(rmask & 0x80 !=1)
{
rmask << 1;
i--;
}

_int8 R;
if(i > 0)
R = (tempBuffer & ddpixelformat.???RBitMask) >> i;
else
R = (tempBuffer & ddpixelformat.???RBitMask)<< -i;

So same thing for the other colors (you can just do the i,j,k (or whatever) calculation once upon initialization, and if most(all?) the cards just have the 565 or 555 code, even do away with the if(i>0) (should be >0 for R, <=0 for green (just treat it like it''s <0) and B < 0)) Now, use the 24bit macro.

Hope it helps.

Alexandre Moura
alexmoura@hotmail.com

Also, you''ll get banding if you just cut off the least significant bits of the 24-bit color. Go read up somewhere (read: Gamasutra) about pixel-conversion and dithering methods.

- Splat