[Radiance-general] Solid acrylic cylinder

[Georg Mischler]

Barbara Matusiak wrote:

> I am trying to simulate a cylinder made of solid acrylic material, lying on
> the table by the window. I defined the sky as "sunny sky with sun". The
> sunlight is incident on the cylinder. I used the "dielectric" material
> description and the primitive "cylinder".
>
> You can see on the picture, that the cylinder is transparent, but there is
> no any spot of the sunlight that comes through the material. I gues the
> reason is that the dielectric material is not a secondary light source.
>
> Has anybody an idea which other "material" should I use?


As Peter already mentioned, you'd need a forward raytracer to
do this "right".

If you're really desperate, then you could try to cranc up the
numbers of diffuse rays to really insane values (-ad >10000 ?)
and hope that you generate enough rays on paths that actually hit
the sun through your cylinder to get some visible effects. As
long as the receiving surfaces are close enough to the cylinder,
something *might* become visible, but it certainly won't be
realistic on a numerical level. The sun is much to small for that.

You can see that *in principle* it does work when all elements
are very close together, and the light source is very big, on
this page:   http://www.q-bus.de/Blender/test009.html
The first picture made with Radiance is the only one where the
highlight inside the shadow of the glass sphere has a square
shape, which is what you'd expect given the square light source.
But this really rather happens almost by chance (and of course
because Radiance follows physical models in the core). The
backwards raytracing method used by Radiance makes caustics
extremely inefficient in almost any other situation, eg. with
smaller light sources, greater distances, and especially with
non-trivial scenes that prohibit excessive numbers of diffuse
rays.


Have fun!

-schorsch

-- 
Georg Mischler  --  simulations developer  --  schorsch at schorsch.com
+schorsch.com+  --  lighting design tools  --  http://www.schorsch.com/