William F Pokorny <ano### [at] anonymousorg> wrote:
> It comes to the fact the pattern - like aoi - needs the surface
> intersection and associated normal to work. You don't have this - or
> have a bogus normal input - when used with functions (or densities). The
> evaluations are happening apart from any surface in 3D space.
This makes me wonder if there's a way to chain the object pattern into the
Anyway, I got it to work - not well, mind you. I need to hunt down where the
errors in accuracy are coming from where I use the normal values to modulate the
color values of the seams (The coloring is intentional, the banding is not.)
I don't think it's my delta-h in the partial derivatives - I went from 0.01 to
0.000001 / image_width without much improvement. I'll probably have to start
graphing the gradient as a function of delta-h for x, y, and z to get a better
feel for what's going on.
Anyway, it's good enough to play with for now. But ... slow. :O
Post a reply to this message