(Stochastic) Progressive Photon Mapping

Discussions related to GPU Acceleration in LuxRender

Moderators: Dade, jromang, tomb, coordinators

(Stochastic) Progressive Photon Mapping

Postby Dade » Tue Jun 01, 2010 8:54 am

This topic is the result of a split from the previous one, just to keep all SPPM discussion in one place.

Stochastic Progressive Photon Mapping is an extremely interesting rendering technique. It is in the same league of MLT/BiDir Path tracing (i.e. some time even better, some time worse). Anyway it is vastly superior to simple Path tracing. (Stochastic) Progressive Photon Mapping related papers are:

http://graphics.ucsd.edu/~toshiya/ppm.pdf
http://graphics.ucsd.edu/~toshiya/sppm.pdf

yolao wrote:Those are Wonderful news Dade, SPPM vastly superior than Path Tracing..AMAZING... Do you already have some test made with this Stochastic Progressive Photon Mapping technique?..if so can you show it please.. ;) ?


Oh, I'm just splatting the first photons around, I'm still far from a complete PPM (the Stochastic version will arrive later):

i1.jpg


and after few seconds:

i2.jpg


I enjoying looking at photons bouncing around and splatting colours over the surfaces :D
User avatar
Dade
Developer
 
Posts: 8298
Joined: Sat Apr 19, 2008 6:04 pm
Location: Italy

Re: SLG v1.5beta3 (The SunSky release !)

Postby yolao » Tue Jun 01, 2010 9:05 am

Fantastic Dade!

I don`t know all the tech.

Can you help me with some answers please so i know a bit better about all this technologies..

1-So MLT works together with Bidirectional Path Tracing..or are kind of fusion on the same technologY?..most of time i see them mention in the same sentence...

2-Renders like Maxwell use MLT/Bidirectional Path Tracing..right?...Is this what is able to achieve such a complex lighting/material scenes..?

3-You will have SPPM in SmallGPU which is on the same league as MLT/Bidirectional Path Tracing....so this means that SmallGPU will be able to render any kind of ultra complex lighting/materials situations with great speed?

Thanks and congratulation for you work, is AMAZING..!!
yolao
 
Posts: 28
Joined: Thu Apr 22, 2010 8:11 pm

Re: SLG v1.5beta3 (The SunSky release !)

Postby straaljager » Tue Jun 01, 2010 9:27 am

Amazing indeed! :D This could give SLG the edge over other GPU path tracers. A preview in video form would be more than welcome ;)
Last edited by straaljager on Tue Jun 01, 2010 12:28 pm, edited 1 time in total.
straaljager
 
Posts: 62
Joined: Mon Feb 01, 2010 5:02 am

Re: SLG v1.5beta3 (The SunSky release !)

Postby losbellos » Tue Jun 01, 2010 10:03 am

Damn! Thats fast!
So where is the build and what are the params? ;) Let us test, maybe we find some bugs too.
losbellos
 
Posts: 783
Joined: Wed Jun 24, 2009 4:59 am

Re: SLG v1.5beta3 (The SunSky release !)

Postby straaljager » Tue Jun 01, 2010 12:56 pm

Just found this amazing demo while searching the web for "SPPM": http://graphics.ucsd.edu/~toshiya/gpusppm.zip It's written in GLSL and runs on ATI (R600 and up) and Nvidia (G80 and up) cards. It's really fast even on my feeble 8600M GT.

Left image is the result after half a second and right is after 8 minutes (on newer cards this should be just a few seconds):

Image

Caustics converge very fast! I now understand why this is considered superior to pathtracing. Having SPPM in SLG will be crazy awesome :D
Last edited by straaljager on Tue Jun 01, 2010 1:31 pm, edited 1 time in total.
straaljager
 
Posts: 62
Joined: Mon Feb 01, 2010 5:02 am

Re: SLG v1.5beta3 (The SunSky release !)

Postby losbellos » Tue Jun 01, 2010 1:05 pm

it takes one minute with hd4850 and quad core processor.
Fantastic, even dof is integrated into this small program.
losbellos
 
Posts: 783
Joined: Wed Jun 24, 2009 4:59 am

Re: SLG v1.5beta3 (The SunSky release !)

Postby SATtva » Tue Jun 01, 2010 1:09 pm

Please forgive my ignorance, but is SPPM an unbiased method? From what I can understand, it is not.
Linux builds packager
聞くのは一時の恥、聞かぬのは一生の恥
User avatar
SATtva
Developer
 
Posts: 7162
Joined: Tue Apr 07, 2009 12:19 pm
Location: from Siberia with love

Re: SLG v1.5beta3 (The SunSky release !)

Postby losbellos » Tue Jun 01, 2010 1:30 pm

"In computer graphics, unbiased rendering refers to a rendering technique that makes no assumptions or approximations in computing an image. In the real world, light interacts with the surroundings in complex ways, and light rays can reach the viewer via an infinite number of paths"

So it is not the method but the utilization. In a very strict meaning brute force core would be the base of true unbiased rendering method.
losbellos
 
Posts: 783
Joined: Wed Jun 24, 2009 4:59 am

Re: SLG v1.5beta3 (The SunSky release !)

Postby SATtva » Tue Jun 01, 2010 1:37 pm

In my mind, "rendering technique" == "rendering method". So the question still holds.
Linux builds packager
聞くのは一時の恥、聞かぬのは一生の恥
User avatar
SATtva
Developer
 
Posts: 7162
Joined: Tue Apr 07, 2009 12:19 pm
Location: from Siberia with love

Re: SLG v1.5beta3 (The SunSky release !)

Postby rikb » Tue Jun 01, 2010 1:41 pm

SATtva wrote:Please forgive my ignorance, but is SPPM an unbiased method? From what I can understand, it is not.

I looked at http://graphics.ucsd.edu/~toshiya/sppm.pdf and apparently PPM is supposed to remove the bias from photon mapping, so I'm assuming this carries over to SPPM.
I could easily be wrong though.
rikb
 
Posts: 81
Joined: Mon Feb 01, 2010 5:22 am
Location: Finland

Next

Return to GPU Acceleration

Who is online

Users browsing this forum: Bing [Bot] and 1 guest