Search found 50 matches

by Dietger
Fri Jul 27, 2012 2:40 pm
Forum: General Development
Topic: MIS on the GPU
Replies: 41
Views: 27265

Re: MIS on the GPU

Yep, those are the ones.
by Dietger
Fri Jul 27, 2012 2:12 pm
Forum: General Development
Topic: MIS on the GPU
Replies: 41
Views: 27265

Re: MIS on the GPU

Assuming that f(X) is the measurement function on your path space, the light transport paths X0,X1,X2,X3 and X4 that you drew are identical and thus F(X0),F(X1),F(X2),F(X3) and F(X4) will also be equal. The difference is only in the way they were sampled, so in their sampling pdf's. I think you don'...
by Dietger
Fri Jul 27, 2012 1:14 pm
Forum: General Development
Topic: MIS on the GPU
Replies: 41
Views: 27265

Re: MIS on the GPU

You do not necessarily have to compute equation 3.3 at the very end. You should realize that equation 3.3 also applies to regular path tracing with next event estimation. Each sample consists of a collection of paths (each successful next event estimation results in a path with non-zero contribution...
by Dietger
Fri Jul 27, 2012 12:43 pm
Forum: General Development
Topic: Veach thesis - formula question
Replies: 86
Views: 48709

Re: Veach thesis - formula question

Yes, for the balance heuristic they do cancel. However in the power heuristic they do not cancel out. Also, sometimes it is acceptable (and convenient) to use approximations of the probabilities for computing MIS weights, in which case they no longer cancel. For example, for practical Russian roulet...
by Dietger
Thu Jul 26, 2012 4:21 pm
Forum: General Development
Topic: Veach thesis - formula question
Replies: 86
Views: 48709

Re: Veach thesis - formula question

I assume you are referring to equations (3.3) and (3.7). Note that equation 3.3 does not correspond with your equation a. Equation 3.3 sums over a collection of N different samples X_1 ... X_N, which in BDPT terms is the collection of bidirection connections generated by a single eye and light path....
by Dietger
Thu Jul 26, 2012 4:00 pm
Forum: General Development
Topic: MIS on the GPU
Replies: 41
Views: 27265

Re: MIS on the GPU

I think you misunderstood some parts of the recursive mis paper. It consists of two parts: 1) recursive MIS and 2) a streaming BDPT variation (SBDPT). 1) Recursive MIS. This is essentially just a convenient way of computing Veach's bidirectional MIS weights. The paper uses it for SBDPT, but you can ...
by Dietger
Sun Jun 03, 2012 12:49 pm
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9791

Re: BDPT Debugging

I don't really know what you mean by 'normal' path configurations, but it sounds about right. For normal area lights, you can connect to them explicitly (next event estimation) or you can just bounce around and hit them by accident (implicit light path). Usually, you just do both and combine them us...
by Dietger
Fri May 25, 2012 6:33 pm
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9791

Re: BDPT Debugging

Escaped rays are basically samples that hit a (possibly black) environment light. This is how I like to think of it: When starting a light path from a regular area light, you normally first sample a point on the light source geometry and then an emission direction. Comparing this with environments, ...
by Dietger
Mon May 07, 2012 9:57 pm
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9791

Re: BDPT Debugging

Upon further inspection, something doesn't make sense. Let's say we have an light ray. In PBRT, the conversion to projected solid angle for each light vertex is done to the outgoing angle. For example, let's say a vert has wi going to the light, and wo going away from the light. In PBRT, each light...
by Dietger
Tue Nov 29, 2011 10:38 am
Forum: General Development
Topic: MIS question
Replies: 16
Views: 11450

Re: MIS question

Seems ok to me. Thats more or less what I usualy do. How do you handle Russian roulette? Depending on your RR method, it might be that incorporating the termination probability in your MIS would result in better weights.