"Tim Gosnell" <gos### [at] lanlgov> wrote in message
news:394A8432.C5DE0D14@lanl.gov...
| Thanks Bob!  I'll give your script a look.
|
| I'm trying to model an object involved in a physics experiment. It starts
| with a hollow shell of glass (of nonzero wall thickness), but the shell
| is then filled with a lumpy layer of ice that coats the inside wall of
| the shell. Inside the layer is just air. I'm having pretty good success
| with this but I have some uncertainties about how some of the optical
| physics is being handled. Are you a good person to ask about this?

In a word, no.  Mainly because I think someone that has a good knowledge of
both optics and the POV-Ray inner workings is needed to answer questions
regarding this sort of thing.  It has always been said though that POV-Ray is
not to be considered a truly accurate system to work with far as science and
reality goes.  Says so some place in the docs I believe.
The major consensus about the type of thing you are doing seems to be to have
the separate parts intermingle their surfaces for the best results, since this
can prevent an "air" layer from occurring at each ray intersection.  I think
however this idea applies to intersection with a non-transparent object such
as a drinking glass set upon a tabletop whereas the transparency of several
surfaces such as you are doing could be an altogether outcome.  Example of the
overlapping method: the tracing ray goes through outer surface of glass, outer
surface of ice, inner surface of glass, inner surface of ice and then through
air and back out in reverse order.  So the thing to do perhaps is to use CSG
'merge' but then you lose the inner surfaces altogether and get only a outer
glass surface then inner ice surface.  Anyhow, it's something I don't know
about as far as mimicking reality and which method to choose over one or the
other.  I've always gone for what looks right at the time and if you have any
real object to compare to you'll no doubt be better off.

Bob

Post a reply to this message