On Tue, May 27, 2008 at 11:20 AM, peter boeykens <peter_b(at)skynet.be> wrote:
> true enough -
> but that's not so different from say specular highlights.
> They will also differ between rendering in a viewport and rendermapping.
yeah - was just making sure it was known incase it was expected/required to be
identical :)
Rendermapping is tough as soon as you go beyond anything that isn't
view dependant.
Shader need certain pieces of information, and in the case of
rendermapping they don't
exist and so you have to substitute them somehow. They're just
limitations inherent to
the way things work.
What could be a very cool alternative is to do something in CG akin to
a technique they
use for digital doubles. We stored to a custom file a surface response
data. For each
point (pixelish equivalent) we store a series of values for a single
white light from a set
of directions both of the light and the camera. Do it twice - once
with shadows on, once
with shadows off.
Then write a custom surface shader which just has a file input
pointing to the illumination
cache. The shader runs through the scene lights (or even samples an
environment map
if you want to be super tricky). And returns an interpolated lookup of
the illumination cache.
Not quite as fast as a straight rendermap of a texture, but it would
allow you to alter lighting,
and still be blazingly fast compared with a complex shader on the surface.
Cheers,
Alan.
---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi
