adjoint photon tracing back
(L) [2006/04/18] [fpsunflower] [adjoint photon tracing] Wayback!now that you've got your tracer shooting millions of rays per second to achieve realtime ...
... shoot trillions of rays and make it all slow again!
[LINK http://www.cs.utah.edu/~shirley/papers/gi06.pdf]
(L) [2006/04/18] [tbp] [adjoint photon tracing] Wayback!The worse is that it makes sense.
I was wondering how i was going to justify raising up this room temperature this summer...
(L) [2006/04/23] [Ono-Sendai] [adjoint photon tracing] Wayback!Strange.. that's almost exactly how Indigo works [SMILEY Smile]
Except indigo handles spectral calcs more efficiently [SMILEY Smile]
oh.. and indigo has more efficient importance sampling
_________________
[LINK http://indigorenderer.com/]
(L) [2006/04/24] [tbp] [adjoint photon tracing] Wayback!And we're supposed to take your word for it?
Where are the papers or source code, eh? [SMILEY Razz]
(L) [2006/04/24] [Ono-Sendai] [adjoint photon tracing] Wayback!Heh good point TBP.
I usually spend all my time coding (and working), and little writing down my ideas.
It's interesting that they're publishing stuff similar to what I'm doing tho, perhaps i'll try and get something published at a not-too-hardcore conference somewhere (e.g. not siggraph [SMILEY Smile])
Anyway the main reason i say my importance sampling is better, is that basically the paper is describing a pretty minor variation on basic path tracing. Indigo uses metropolis light transfer (MLT), and i'm working on bidirectional MLT, both of
which are well established improvements in terms of importance sampling paths over basic path tracing.
nik
_________________
[LINK http://indigorenderer.com/]
(L) [2007/05/24] [gargamel] [adjoint photon tracing] Wayback!Hi, does anybody have source code implementation
of Adjoint Photon Tracing? I want to include some
results of this algorithm in my bachelor thesis, but
i'm not sure that i completely understand some
points in the article [SMILEY Sad]
Thanx!
(L) [2007/05/25] [gargamel] [adjoint photon tracing] Wayback!Sorry, i was misunderstood. I don't want to copy/paste foreign source into my bachelor thesis.
I've just wanted some basic sources to explore, just to be sure that while i'm reading tha article
and "playing" with source it will help me to understand it better.
After all, I have to defend my thesis in the presence of my examinators, so copypasting is
not an option and i'have to be sure that i understand everything perfectly.
That's why i'm asking for help with some basic implementation.
(L) [2007/05/28] [JuNC] [adjoint photon tracing] Wayback!I've been playing with one in C, I'll upload to: [LINK http://www.atomictiger.net/adjoint.zip]
Warning that it is VERY rough but does work [SMILEY Smile]
(L) [2007/05/29] [dr_eck] [adjoint photon tracing] Wayback!You piqued my interest, so I added adjoint photon tracing to my tracer. As I understand the algorithm, it just means that you eliminate the shadow/direct lighting rays from path tracing. (I don't yet have participating media, so I haven't worried about ray marching. I'm also still in a state of confusion regarding the pertinent difference between Monte Carlo integration and simulation, so I may be missing something there.)
My results make it clear that you need to trace a lot of rays to get low noise results. It's also necessary to have a much faster BRDF calculation than I do. (I hope to get Mersenne Twister figured out this week, but interruptions usually mess with my plans.)
_________________
Opinions? Those are *facts* son.
back