PDA

View Full Version : Still Life by Monte-Carlo Renderer


Kelemen Csaba
07-09-2003, 01:53 AM
Hi everybody,

Images have been generated by my Monte-Carlo renderer. Check the latest results on web.axelero.hu/kcsi .

http://web.axelero.hu/kcsi/SteelAndGold III.small.jpg

KendoJedi
07-09-2003, 04:16 AM
oooo, I feel like running my hand along the tops of those. Simple design and I like it.

GateKeepeR
07-09-2003, 04:46 AM
very nice and simple, good work :thumbsup:

snowflake71
07-09-2003, 08:32 AM
Eszméletlenül tetszik, főleg, hogy saját rendererről van szó.
Csak gratulálni tudok! :)

(Folks, sorry for the Hungarian language, I feel weird when writing to a fellow Hungarian in English. :) )

Kelemen Csaba
07-09-2003, 04:38 PM
Thank you for your words. Here is my favorite. Very simple.

( Köszönöm a jó szavakat.)

http://web.axelero.hu/kcsi/Cones II.small.jpg

emtilt
07-09-2003, 08:46 PM
So you wrote this rendering engine? Is it gonna be available publicly?

DogmaD
07-09-2003, 09:20 PM
I must say this looks great! So you programmed your own monte carlo huh! That is some hard work, are you working alone on this?

I really like the cones, look real to me, could be some ikea product :p

So, how long did it take to program this and how long does it take to render these images?

Kelemen Csaba
07-09-2003, 11:50 PM
“So you wrote this rendering engine?”

Yes. It is mine and relatively simple.

“Is it gonna be available publicly?”

Sorry, it is under construction. In this moment, creation of scenes is feasible via C++ interface. Not too user friendly. I think, useless for others in this way. I’m looking for a generous CAD developer team, to be a host.

There is two article with technical details. ( maybe hard to understand)

1) Universal Metropolis sampling: See: http://www.iit.bme.hu/~szirmay/metroimp4_link.htm
2) Relatively simple, microfacet based reflection functions for metals and dielectricums ( paper, marble, plastic etc): See: : http://www.iit.bme.hu/~szirmay/scook_link.htm

“That is some hard work, are you working alone on this?”
Yes, almost. The code is not too difficult but the theory of bi-directional path tracing and Metropolis sampling and appropriate reflection models are really complex.

“So, how long did it take to program this and how long does it take to render these images?”

I’m working on this 14 years ago. J
The beginning was my first path tracer. It was very simple, but the development of the next version equipped with bi-directional path tracing was a nightmare.

I can create a simple scene under one day. The first noisy image needs only minutes but final version with high resolution and completely eliminated noise needs days. The computation time is proportional to pixel number and samples per pixel, shortly proportional to total number of samples. On the other hand, there is no considerable sensitivity to number of geometrical elements and lighting condition.

See a noisy ( early stopped) result. The computation time was circa 20 minutes. It is very simple, because it shows lambertian ( diffuse) surfaces only.



http://web.axelero.hu/kcsi/noisy.jpg

Brett H.
07-10-2003, 02:22 AM
Not a criticism, but this very much resembles Lightwave radiosity (and that's a good thing, IMHO). I like the realistic results.

Brett

endi2
07-10-2003, 08:42 AM
(Sorry, not english)
Nem akarod ezt beepiteni a Blenderbe? :)

Kelemen Csaba
07-10-2003, 09:09 AM
It is a pure raytracer, radiosity, photon-map, and environment map free. It is unbiased that means the result is pixel precise, simulates exact light transport on physical base.

Radiosity needs patch subdivision, the shadows and shades are patch precise in the best case only and can’t evaluate the general reflection functions.

We can see light only, in this wise history of light traveling over space, bounced on material and reach our eyes. Pure raytracing simulates that, theoretically, without compromise.

CGTalk Moderation
01-15-2006, 02:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.