Internal color calculation should
always be done in floating point, unless for realtime purposes. Usually channels (regardless of the color model) are considered to range from 0 to 1 (
exception: YCC, where chroma is -.5 to .5). Only when performing the actual drawing, the float colors are scaled into the target image's color format. That way you would not have the problem, and also minimize quantization errors across the entire application, that are quite prone to emerge when doing multiple and different processes on the same image.
I do not however understand your issue with the (C)CW gradients. CCW -> blend Color1 to Color2. CW -> blend Color2 to Color1, and reverse the resulting list of colors.
"When one person suffers from a delusion, it is called insanity. When a million people suffer from a delusion, it is called religion." (Richard Dawkins)
Zitat
Baum-Darstellung
