Johannes Hubert wrote:
> 
> You mean you want to have a "beam" of light going through the prism,
> changing it's course two times (coming in and out of the glass) and then
> lighting up something that does not lie in the straight path of the original
> "beam"?
> 
> Sorry, but that is not possible in POV-Ray.
> 

I am a beginner at this stuff, so take what I write
as worth what you paid.... :)

My understanding is that to determine the illumination of 
a given point, the tracer checks for interfering objects
along rays sent directly to all point light sources...
Area lights are simulated as a superposition of a small
number of point sources.  Thus effects like reflection
and refraction which perturb the direction between a point
and a light source are not captured.

Another method would be to send out infinitely many rays
in all directions and count those which pass within a threshold
of a light source.   This would work, but tracing infinite
rays would take awhile :).

A final method would be to determine perturbed paths to
light sources, doing the calculations off reflecting surfaces,
index of refraction, etc.  But this is impossible -- imagine
an object with a fractal surface and with an interior
with an index of refraction specified
as a fractal field...  there may be infinitely many optical paths
through such an object connecting any point on one side with
any other point on the other.... asking "given a ray from point
A in direction V, where does it intersect infinity" is
easily approximated numerically, but asking
how to connect point a with a given point P at infinity
is much harder due to the potential multiplicity of solutions.


Dan

-- 
http://www.flash.net/~djconnel/

Post a reply to this message