Am 04.05.2016 um 18:09 schrieb William F Pokorny:

> So! Depending on the angle of the ray relative to the lathe, it looks
> like the x<0 control point does indeed sometimes cause a flip in normals
> at one end.
> 
> In other news the ordering of X as in Y can also flip normals - suppose
> expected mostly.

Outright flipped normals are not that much of a problem (they only make
themselves known if you use an interior_texture) as entirely distorted ones.

Compare for instance the shading of the spindle-shaped "core" inside a
self-intersecting torus (so-called spindle torus) between versions 3.7.0
and the current development versions.

Post a reply to this message