Re: [Algorithms] The state of the art of (real-time / non-precomputed) Global Illumination?

[Chris B. (BUNGIE) <cbu...@mi...>]

Yes, the Geomerics technique uses precomputed form factors, and it is mainl=
y designed to handle complex environment-mediated diffuse transport. In oth=
er words they are not trying to solve for dynamic occluders, but they can h=
ave an infinite number of light sources in a static environment. There is a=
 significant precomputation time requrired to build and encode form factors=
.


You might like to read this paper from SIGGRAPH last year by Alex Evans of =
Media Molecule (the developers of Little Big Planet): "Fast approximations =
for global illumination on dynamic scenes" - ACM SIGGRAPH 2006 Courses, Ses=
sion: Advanced real-time rendering in 3D graphics and games.

I'm not sure if you have access to the ACM full-text website but here is a =
link if you do. http://portal.acm.org/citation.cfm?id=3D1185834


When you talk about the GPU Gems technique that you liked, is that Michael =
Bunnell's chapter on "Dynamic Ambient Occlusion and Indirect Lighting"? If =
so, yes that is a per-vertex technique as written but from my initial readi=
ng there is nothing that stops you from dicing your geometry or using a sur=
face texture based approach if that's what you want for your game. However =
because it's a direct surface-to-surface technique it is unlikely to scale =
to large environments like those you've mentioned in previous posts to the =
list.


There are as yet no general purpose realtime global illumination algorithms=
 that will solve your problem. You're going to have to decide what kinds of=
 visual effects are most important to you and then select a technique that =
exhibits those characteristics.

--
Chris Butcher
Technical Lead
Bungie Studios
bu...@bu...


-----Original Message-----
From: gda...@li... [mailto:gdalgorithms-=
lis...@li...] On Behalf Of Megan Fox
Sent: Wednesday, March 28, 2007 09:21
To: Game Development Algorithms
Subject: Re: [Algorithms] The state of the art of (real-time / non-precompu=
ted) Global Illumination?

Going by the lack of dynamic objects in the scene, and the lack of
claims regarding the proper GI influence upon the scene as given by
dynamic reflecting objects, it looks like this is another pre-comp
method.  Assumedly, SH-related, given their mentions of experience
with it and their second PRT method.

Very pretty, but useless for us - it looks like it would really only
be appropriate in an FPS where everything was nailed to the ground.

On 3/28/07, Emmanuel Deloget <lo...@fr...> wrote:
> Hello,
>
> Geomerics have claimed to have developped novel GI algorithms (using geom=
etric
> algebra). I'm not good enough in the mathematic fields to understand if t=
his
> algorithm is really a GI one (although they claim it's a real-time radios=
ity
> algorithm) - and certainly not good enough to understand the principia be=
hind
> the algorithm (having problems to understand geometric algebra :))
>
> http://www.geomerics.com/index.php?page=3Dlighting
>
> --
> .Emmanuel Deloget
>