Quantization error/layering of gradient and dx9



there is node in dx11 - noise dithering, which should improve “quantization error”, is there a way how to improve this error in dx9, thus have nice smooth gradient without “layers”

i am using quad with png texture, white gradient to RGB=1,A=0

or should i just go dx11?

(to be honest i dont want just solution, but also understand why is this happening)


color banding


that tfx should be easily portable to dx9, but to get it working you need to process all your rendering in floating point texture format, like “A16B16G16R16F”


ok texture format on dx9texture solved the problem, thanks people!