"Slime" <fak### [at] emailaddress> wrote:
> I imagine the gradient of an image is calculated differently than the
> gradiant of a function. Try setting the normal's accuracy to a low value
> (like accuracy .000001 or whatever works best) to improve the sampling of
> the function.

That was it; thanks! (An accuracy of 0.001 was sufficient)

Post a reply to this message