Yes but that's an interface problem. Ofcourse you can't provide more decimals than the number of bits used by the float. EDIT: 0.999 * 255 (8 bit) = 255

Yes but that's an interface problem. Ofcourse you can't provide more decimals than the number of bits used by the float. EDIT: 0.999 * 255 (8 bit) = 255 0.9999 * 255 (8 bit) = 255 0.999 * 4096 (12 bit) = 4092 0.9999 * 4096 (12 bit) = 4096 So for lower color depths the number of decimals isn't that important.