Greg Ward wrote:

The trouble is, trans doesn't really know WHAT to do with a surface
normal perturbation in the case of a transmitted ray. If an object were
really infinitely thin, then a texture would perturb both incoming and
outgoing normals identically, having no effect at all on the transmitted
ray. I assume that the user wants an effect when they apply a texture,
and this is a semantic difference between a true texture and a
"smoothing operation." Unfortunately, Radiance treats the two cases the
same, so this is the result. I don't have a fix for it. Smoothing
doesn't work with the trans material, and that's the bottom line.


> From: Peter Apian-Bennewitz <[email protected]>
> Date: Wed Apr 10, 2002 05:14:27 AM US/Pacific
> To: Greg Ward <[email protected]>
> Subject: trans+texfunc
> Hi Greg,
> do I get this right ?
> using
> void trans t
> 0
> 0
> 7 .90 .90 .95 .001 0 .95 1
> on a non-convex body, plus texfunc for Phong-interpolation doesn't
> work, because with trans, the surface normals are auto-flipped towards
> the incoming ray, which jeopardizes the Phong interpolation ?

one for the archive ...


pab-opto, Freiburg, Germany,