> Your next best bet is to fake it by way of running the lighting check
> yourself (cf. the demo on how to write a raytracer in the SDL, contained
in
> the help file) and use the intersection points to produce a mesh that
> contains only what you want, namely the areas lit by these light sources.
> This would get rather messy eventually, especially if you want things to
> look right; however, you do have the "advantage" of being able to speed up
> the render by lowering the resolution of the lighting checks as needed.
> (The worst part of this is that it'd be a forward step, as with the photon
> map, rather than the standard reverse method.  Ah well.)

A much simpler and faster approach would be to render the image with only
the specific object in white, everything else in black, and the special
lights set to white as well. You'd get an image where white represents the
visible object, and black the background.
Then do another render of the scene, this time with the appropriate colors
and all lights switched on.

You the black/white image as a function, multiply it with the colored image
(as a function) and place that close in front of the camera when rendering
the scene without the object. Voila! Composite image!

If you're using focal blur though, that might be a problem, so then I'd
suggest using trace() to find the 3d-Locations of the pixels, and color them
using the colored image.

All together not very suitable for animations, but nontheless, that could
still be automated rendering the animation three times, using the frame
number as reference which image-maps to load.

Regards,
Tim
-- 
aka "Tim Nikias v2.0"
Homepage: <http://www.nolights.de>

Post a reply to this message