> > How about use normal vector to surface and assing, that resulting
> > contour of object will be perpendicular to normal vector's projection
> > into image plane? (Just an idea, don't know, if I'am right).
> 
> I have no idea what you are trying to say...

When you have sphere, for example, its controur is circle.
Normal vector to this circle is the same as projected normal
vector to sphere in this point. Thus we have direction of
sphere's controur in this point and we can search next point
in nearly that direction. This fails for edges on rotated cube,
for example (generally on all edges, it is usable for countour
of objects).

> CSG objects work the same as any other objects. Why should they be any
> different?
> 

And it fails also on edge between two parts of some CSG.

Disnel

Post a reply to this message