Search found 12 matches

by jameszhao00
Tue Nov 06, 2012 7:17 pm
Forum: Links & Papers
Topic: Ph.D. thesis
Replies: 21
Views: 24737

Re: Ph.D. thesis

Saw the section on converging and temporal reuse. Have you taken a look at the Real Time Reprojection Cache paper?

http://dl.acm.org/citation.cfm?id=1180080
by jameszhao00
Mon Oct 22, 2012 4:13 am
Forum: General Development
Topic: Eye path reuse
Replies: 1
Views: 2454

Eye path reuse

I've been working on a CUDA path tracer, and thought it would be great to reuse eye paths from threads in the same warp. For example, pixel B creates a path of EDDL, and intuitively pixel A can use its own ED and connect to pixel B's DL. This was inspired by the 'Light path sharing' concept mentione...
by jameszhao00
Sat Oct 20, 2012 5:28 am
Forum: General Development
Topic: Veach thesis - formula question
Replies: 86
Views: 48714

Re: Veach thesis - formula question

Quick question: I was writing out and simplifying the path integral formula and noticed an extra cos term appearing near the sensor parts. Am I simplifying something incorrectly? I've always seen eye path throughputs start at 1. For example, for a ED-L path with 1 light vertex, we have (all pdfs are...
by jameszhao00
Thu Jun 07, 2012 3:31 am
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9803

Re: BDPT Debugging

Thanks for the help!
by jameszhao00
Sun May 27, 2012 3:08 pm
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9803

Re: BDPT Debugging

Thank you for the response! I think I'm kind of getting it. I essentially have to add a black environment light (visualized as infinite sphere around the origin) and consider that as a light source (in addition to my area lights). Thus, the normal path configurations as well as ones with ones that c...
by jameszhao00
Sat May 19, 2012 9:48 am
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9803

Re: BDPT Debugging

Ah I see. Thank you for the clarification. I should've said earlier that I'm trying to deal with escaped rays (dealing with RR termination is in the Veach thesis :D ) I've theorized about escaped rays as a set of vertices at infinity... thus a G of 0 and a P(i+1)/P(i) of 0. Escaped rays are physical...
by jameszhao00
Fri May 18, 2012 9:59 pm
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9803

Re: BDPT Debugging

Thank you for the help! Hmm... so that means that I can just ignore those probabilities when summing Pi. I definitely like this solution :D However, I'm still confused by the intuition. What if I set P(a->b)G(a<-->b) to be 1 for all the i's? So in essence we're uniformly weighting all the techniques...
by jameszhao00
Fri May 18, 2012 3:25 pm
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9803

Re: BDPT Debugging

If anyone happens to have some tips on Multiple Importance Sampled BDPT: Sometimes we can only partially connect eye and light paths. For example, with 10 eye vertices and 2 light vertices it would be impossible to create a 8 vertex path with 4 eye vertices and 4 light vertices. What probability are...
by jameszhao00
Sat May 12, 2012 11:24 pm
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9803

Re: BDPT Debugging

Thanks for the previous posts! I'm currently looking at the BDPT MIS weight calculation and have come to some intuitions. Can someone confirm if they're correct? - The connecting edge's probability/geometry is not taken into account in p or p[i+1]. I'm confused as to why this is so. - p[i+1]/p only ...
by jameszhao00
Tue May 08, 2012 2:54 am
Forum: General Development
Topic: BDPT Debugging
Replies: 15
Views: 9803

Re: BDPT Debugging

Thanks for the help!

I narrowed it down to the light path generation function earlier and spent a few hours debugging it. Didn't make any progress, so I rewrote it (without looking at the original code). And voila it matches the reference!