Information: Metropolis Light Transport

[Geddart]

Due to some developments in the Renderer Business here some (rather technical) information about Metropolis Light Transport and Techniques beyond Bidirectional Path Tracing.

"The procedure constructs paths from the eye to a light source using bidirectional path tracing, then constructs sligt modifications to the path. Some careful statistical calculation (the Metropolis algorithm) is used to compute the appropriate distribution of brightness over the image. This procedure has the advantage, relative to bidirectional path tracing, that once a path has been found from light to eye, the algorithm can then explore nearby paths; thus difficult-to-find light paths can be explored more thoroughly with the same number of simulated photons."

Pure Bidirectional path tracing

Metropolis Light Transport (same rendertime)


- Original Paper (August 1997)
- Further Ideas (December 1997)
- Simple and Robust Mutation Strategy for the Metropolis Light Transport (2002)
- Metropolis Iteration For Global Illumination (2004)

Renderers with Metropolis Light Transport

- An Implementation
- MetropoLight
- Maxwell Render

Btw, Vray's QMC Sampling is already way more advanced than simple Bidirectional Path Tracing.
However we still might see some changes of that in the future... (Vladimir squeezes everything out... )

I Hope some of you might find this Collection interesting.

[Frances]

I hope nobody held their breath waiting for the MetropoLight download to be available. :P

[natty]

Interesting ... thanks Sascha ..

[ngrava]

I have a book called "realistic image synthesis using photon mapping" that discusses MLT in depth. I was under the impression that it was just an early derivation of Bi-directional path tracing. I get that it's non biased and that's good and preduce more “True” results but that can be pretty wasteful as it can take hours to render simple scenes. Non biased renders are not really supposed to be used in time crucial production environments like movies or commercials. This is what I find so interesting about Vray's Light mapping technique. It appears to very similar to a unbiased solution. I sure would like to know more about it. I met Vlado at Siggraph this year and he tried to explain that it traces light from the light source back to the camera (like Bi-directional path tracing and MLT) and that it simulates infinite bounces... Huh? How does that work?! I've read about MLT doing this. I've wondered before if there was some way of interpolating the results at some adaptive number of stages... Is that what Light mapping actually is; a simple Metro light transport used to gather samples for interpolation? And if so, wouldn’t that make it a lot closer to being un-biased?

-=GB=-

[Frances]

Hi GB,

I just looked at every clip on BentImageLab. The Sustainable Forestry Initiative spot by Chel White is my favorite.

Does Max/VRay play a very large part in your production work?

[ngrava]

Hi Frances,

I'm glad you like our work! We only recently switched to Max/VRay from LightWave and Maya. So, nothing that's currently on the website is done with VRay. But yes, We're using it almost exclusively now. Even the work we do in other software will get converted to Max and rendered in VRay.

-=GB=-

[ngrava]

Oops, I meant to say that Vlado said that Light Mapping traces from the camera into the scene. Sorry for any confusion.

On another note, I've been messing with Light mapping and have discovered that it's probably the closest thing to MLT I can find. If you use QMC as the primary and secondary GI engines, you'll get the closest thing to an unbiased render. If you compare this to the Irradiance map and Light Map combo, it looks very similar.