Hi,

my problem is slightly off-topic but I guess you're the best people to ask
for help...

I am currently developing a physical simulator for gas transport and surface
chemistry is sub-micron features. At such scales gas-transport is pretty
much collisionless, i.e. particles collide with walls much more often than
with each other -> ballistic particle flux. To my understanding this is -
apart from the nature of particles - identical to photon tracking...?

While my code gives correct results (cross-checked with view factor), I
don't know how fast my tracer is, compared to the state of the art. Now I
have several questions:

1) is Povray capable of photon tracking and, if yes, is it a good benchmark
(=fast)?
2) if yes, is there anyone out in the povray universe, who could help me
setting up the input files - PLEASE!?
3) if not, what is the fastest photon tracker publicly available?

Ah yes, two more things before you all volunteer ;-)  The structures are
closed, except for a small opening -> MANY reflections (several 1e4-1e5).
And is there a way to read in geometries as a list of triangles?

Thanks a lot in advance for your help!!!

Greetings from Bavaria, Georg Icking-Konert

Post a reply to this message

nomail@nomail wrote:
> 
> 1) is Povray capable of photon tracking and, if yes, is it a good benchmark
> (=fast)?

If you'd describe what 'photon tracking' is someone might be able to 
answer your question.  POV-Ray uses a feature called photon mapping to 
generate reflective and refractive caustics.

Christoph

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 06 Jul. 2004 _____./\/^>_*_<^\/\.______

Post a reply to this message

Hi Christoph,

sorry for the ambiguous question. Unfortunately I am not very familiar wit
RT, so please let me explain it in other terms.

To my understanding, in 'photon-mapping' (or 'tracking' or however it is
called... ;-) ) photons are emitted by a light source (in contrast to the
camera as in 'usual' RT) and traced over several reflections and
refractions. Is this correct? If yes, this is VERY similar to my problem...

I track gas particles which are emitted from a source (the ambient, an ion
source, etc.) and determine the first collision with the surface. At the
surface the particle is absorbed, scattered, or reflected. If it is not
absorbed, the next collision with the surface is determined... This
procedure is continued until the particle is absorbed at a surface or until
it leaves the simulation window. To my understanding this is pretty much
what is done in photon mapping...??? Only in my case I sample gas flux
where RT calculates light intensity.

There is a slight difference to RT which however won't be an issue for a
benchmark. I just mention it here for clarity. The microelectronical or
-mechanical structures I study are closed, i.e. particles can ONLY leave
the simulation to the top. Particles reaching the bottom are discarded
(usually that doesn't happen) and particles reaching the in-plane
boundaries (x and y) are reflected back into the simulation (reflecting
boundary condition) or translated to the opposite border (periodic BC). As
I said this is no problem, since my benchmark structures are already closed
(i.e. don't utilize the periodicity). Also, the light/gas source is located
1e-3mu above the structure so there is very little space above the
structure.

Hope this clarifies my problem!? If you have more questions, please don't
hesitate to ask! Thanks a lot in advance for your help!

Ah yes, I almost forgot. I also have initiated contact with a group at a
university who claim they can do real-time RT. I am still waiting for the
benchmark results but this is your chance to benchmark Povray not only
against my code but also against bleeding edge technology (provided I am
allowed to release the results)... ;-)

Georg

PS: nice homepage, Christoph! :-)


Christoph Hormann <chr### [at] gmxde> wrote:
> nomail@nomail wrote:
> >
> > 1) is Povray capable of photon tracking and, if yes, is it a good benchmark
> > (=fast)?
>
> If you'd describe what 'photon tracking' is someone might be able to
> answer your question.  POV-Ray uses a feature called photon mapping to
> generate reflective and refractive caustics.
>
> Christoph
>
> --
> POV-Ray tutorials, include files, Sim-POV,
> HCR-Edit and more: http://www.tu-bs.de/~y0013390/
> Last updated 06 Jul. 2004 _____.//^>_*_<^/.______

Post a reply to this message

nomail@nomail wrote:
> Hi Christoph,
> 
> sorry for the ambiguous question. Unfortunately I am not very familiar wit
> RT, so please let me explain it in other terms.
> 
> To my understanding, in 'photon-mapping' (or 'tracking' or however it is
> called... ;-) ) photons are emitted by a light source (in contrast to the
> camera as in 'usual' RT) and traced over several reflections and
> refractions. Is this correct? If yes, this is VERY similar to my problem...

More or less this describes the shooting step of photon mapping but this 
is only one part of the whole process.  The result (the photon map) is 
not a very good approximation of the distribution of light in the scene 
in general so it is only used directly for caustics (where the photon 
density is high enough for good results).

So concerning the original question: photon mapping in POV-Ray is 
certainly quite fast in what it is designed for (i.e. distributing 
photons in a complex scene and using the generated photon map for 
rendering caustics.

Christoph

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 06 Jul. 2004 _____./\/^>_*_<^\/\.______

Post a reply to this message

So I understand that Povray is too specialized to do what I would like it to
do? Or, in other words, there is no easy way to benchmark the photon
tracing algorithm only...? Too bad :-(

Georg

Post a reply to this message

nomail@nomail wrote:
> So I understand that Povray is too specialized to do what I would like it to
> do? Or, in other words, there is no easy way to benchmark the photon
> tracing algorithm only...? Too bad :-(

I am not aware that i said anything like this.

Christoph

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 06 Jul. 2004 _____./\/^>_*_<^\/\.______

Post a reply to this message