Archived

This topic is now archived and is closed to further replies.

gunslinger77

are there any algos for implementing

Recommended Posts

global illumination (ie soft shadows, self shadows and so on) other than radiosity, ray tracing and harmonic spheric lighting? Mmm maybe something nice and gritty without all those incomprensible integrals? I admit, i am no mathematician at all...

Share this post


Link to post
Share on other sites
Well.....there is a reason why there are all those incomprehensible integrals. Sounds like you are looking for a procedure for generating art content that can feed already-developed code for global illumination, and NOT an algorithm at all. You might check out ATI and nVidia''s developer sections of their websites. There are quite a few presentations and articles there----some that focus on the artists view of creating content for global illumination. Hopefully that will help.

Graham Rhodes
Senior Scientist
Applied Research Associates, Inc.

Share this post


Link to post
Share on other sites
>>Sounds like you are looking for a procedure for generating
>>art content that can feed already-developed code for
>>global illumination, and NOT an algorithm at all.

Well, if that's the case then I guess I don't have further to add.

But anyway, if you're looking for an easy solution, then I personally suggest monte carlo ray tracing for your first GI code. Sure, it may not be the fastest (for simple and mediocre scenes at least) but it's VERY easy to add if you already have a "standard" ray tracer that traces mirror surfaces (ie. generates a reflected direction as v'=v-2n*(n*v)) - just change the reflected direction from perfectly specular to lambertian (randomly over the hemisphere above the surface and scale properly, or using cosine-weighted sampling, the formulas are available) and take an average of a large number of rays.

While radiosity might be faster for simple scenes, the problem is that it's not too trivial given an existing rasterizing code (opengl or directx in practice), because it requires coding a hemicube renderer and most of all, a meshing system for the patches.

>>Mmm maybe something nice and gritty without all those >>incomprensible integrals?

Monte carlo ray tracing and radiosity are very simple algorithms once you understand their ideas. You don't really even need to know that you're solving an integral problem (just check out Hugo Elias's site about "patch conversation" in radiosity!). Harmonic spherical lighting isn't all that simple, imho, and for that you do need some background of the math.

- Mikko Kauppila

PS. of course simple monte carlo raytracer with lambertian sampling isn't a *full* GI solution, but at first that's easier to add to a standard recursive raytracer than arbitrary BRDFs.

[edited by - uutee on January 16, 2004 2:14:31 AM]

Share this post


Link to post
Share on other sites