Search found 89 matches

by friedlinguini
Thu Apr 24, 2014 11:45 am
Forum: General Development
Topic: Full Circle
Replies: 0
Views: 2715

Full Circle

In the past, ray tracing was mocked as practical only if you want to draw glass spheres on infinite checkerboards. Given all the advances in global illumination over the last few decades, what is the state of the art in general-purpose global illumination algorithms that can draw a glass sphere on a...
by friedlinguini
Sun Apr 06, 2014 10:30 pm
Forum: General Development
Topic: Measure the convergence speed
Replies: 20
Views: 13859

Re: Measure the convergence speed

Following your post, I also implemented the algorithm as outlined in "Progressive Path Tracing with Lightweight Local Error Estimation". However, I found that using (AllSampledImage(x,y) - OnlyEvenSampledImage(x,y))^2 to compute the variance didn't work too well. The issue I had is when a firefly a...
by friedlinguini
Sat Mar 22, 2014 1:34 am
Forum: General Development
Topic: Efficient sampling of many lights
Replies: 22
Views: 22477

Re: Efficient sampling of many lights

Googled a bit. Before ingenious yells at me I should point out this paper: http://cs.au.dk/~toshiya/misc.pdf

It explains things more thoroughly than I did and also points out start-up bias in metropolis light transport.
by friedlinguini
Sat Mar 22, 2014 1:10 am
Forum: General Development
Topic: Efficient sampling of many lights
Replies: 22
Views: 22477

Re: Efficient sampling of many lights

Is my new understanding closer to correct? Am I still missing something? There's probably a FAQ somewhere about this. "Unbiased" means that the expected value for a random sample equals the correct value. I.e., if you look at every possible sample you might generate with a rendering algorithm, weig...
by friedlinguini
Fri Mar 21, 2014 1:12 pm
Forum: General Development
Topic: Efficient sampling of many lights
Replies: 22
Views: 22477

Re: Efficient sampling of many lights

From my own reading of the paper, the choice of representative light, while random, is fixed during the tree construction stage. So while repeated renderings, with different fixed trees, may eventually average to the correct solution, the technique will never converge to the correct solution within...
by friedlinguini
Fri Mar 07, 2014 4:40 pm
Forum: General Development
Topic: Measure the convergence speed
Replies: 20
Views: 13859

Re: Measure the convergence speed

It's a good read, but I think there have been improvements in the basic approach since it was published. Any pointers to those improvements? :) http://www.cgg.unibe.ch/publications/2011/adaptive-sampling-and-reconstruction-using-greedy-error-minimization http://www.cgg.unibe.ch/publications/2012/ad...
by friedlinguini
Thu Mar 06, 2014 10:33 pm
Forum: General Development
Topic: Measure the convergence speed
Replies: 20
Views: 13859

Re: Measure the convergence speed

The last paper references this one: http://www.cs.berkeley.edu/~ravir/Overbeck2009AWR.pdf I think it is a "to read" paper. That one is more about adaptive sampling and noise reduction, and I think the adaptive sampling part is what bridges into the Hierarchical Automatic Stopping one. It's a good r...
by friedlinguini
Wed Feb 26, 2014 6:41 pm
Forum: General Development
Topic: Curve bounding box - fast computation
Replies: 7
Views: 5001

Re: Curve bounding box - fast computation

The bounding box isn't minimal, but you can get closer by cutting the Bezier curve in half via de Casteljau subdivision and then taking the bounding box of all of the resulting control points. You can get arbitrarily tight bounds by subdividing recursively. You don't need to subdivide curves if the ...
by friedlinguini
Tue Jan 14, 2014 1:57 pm
Forum: Tools, Demos & Sources
Topic: WebGL path tracer
Replies: 4
Views: 9986

Re: WebGL path tracer

See, for example, this tidbit from Inigo Quilez: http://iquilezles.org/www/material/nvsc ... /rwwtt.pdf
by friedlinguini
Thu Jan 09, 2014 10:12 pm
Forum: General Development
Topic: Population Monte Carlo Sampling
Replies: 14
Views: 11087

Re: Population Monte Carlo Sampling

hmm... doe's it mean that any adaptive MC algorithm is hopeless in general as it introduces bias? Is it (at least theoretically) possible to have a priori bounded bias in adaptive MC? Define "hopeless". I've heard that all adaptive MC is biased, but not an actual explanation for the claim. Heck, ev...