Real-time radiosity?

There's probably a catch somewhere. Maybe the lights are dynamic but the geometry isn't or something. It's pretty unbelievable that after years of research in the field, someone comes up with a solution thats orders of magnitude faster. Not impossible, just unbelievable.
 
Inane_Dork said:
There's probably a catch somewhere. Maybe the lights are dynamic but the geometry isn't or something. It's pretty unbelievable that after years of research in the field, someone comes up with a solution thats orders of magnitude faster. Not impossible, just unbelievable.

Still if they can get this to work as a pluggin for 3d apps like Max/Maya that would be sweet!

Would have been nice to see a bit more in depth read on it though.
 
The solution is real time, achieving frame rates of up to 100Hz on common graphics hardware, and allows for fully dynamic lighting including point spot-lights, area lights, texture-based lighting, glowing objects and the projection of video onto geometry. The full indirect lighting calculation includes the soft-shadows that characterise real-world scenes. Geomerics' technology also integrates smoothly with classic techniques such as normal and gloss mapping, and fully supports HDR (high dynamic range) lighting effects.

Now if we can only get them to release a demo on the PC... the video looks nice, now I want to know all their hacks.
 
Geomerics has applied its core technology, based around geometric algebra, to produce this new radiosity solution.

So am I to understand they have achieved with math what could not be done in real time to day via polling environment data with bounced rays? (I'm really asking is this a valid guess as to what Geometrics is doing)

This would remove the bandwith problem associated with ray traced radiocity and place the load on just how much horsepower a GPU has with respect to it's ALUs if I understand the quoted statement correctly.

I'm making a assumption here but does this suggest this method is more of an approximation than ray tracing due to results coming from math more than taking in environmental data first and then computing results?

Sorry I can't find just the right wording for what I'm asking but hopefully someone gets what I trying to ask.

Anyway, I hope this is a robust and flexible means to radiosity in games as that would be pretty sweet.
 
From what I can tell they are doing PRT with an improved basis function system instead of using SH. This is probably something similair to Haar wavelets. You can check their site, they give some basic info and hints. It's really nothing amazing though, I thought of the general concept a while ago, but no time to really work it through.

If you notice there is no dynamic geometry, so it still has significant limitations.
 
Looking at their website, it seems they invented realtime anything simply using Algebra. Colour me skeptical.
 
DudeMiester said:
If you notice there is no dynamic geometry, so it still has significant limitations.
Yes, another of these solutions with a very limited presentation. Great for architects wanting to walk customers through virtual buildings with dynamic lighting tests, but not much use for everyone else...
 
Not really. You can still do tricks to light the dynamic objects based on the illumination of the surrounding environment. Sort of a dynamic IBL. There was a Ruby demo from ATI sort of showing this idea. Basically, you create a set of points in the space the dynamic object will move through, and create functions that represent the light passing through the surrounding space at each point. Then you just sample the nearby points when rendering the object. It won't allow the object to contribute to the scene radiosity of course (other then maybe in the form of dynamic ambient occlusion), but you should still be able to cast shadows without issue. It's a huge hack either way, but I imagine it could somewhat work.

Just thinking about how it would look, and I'm reminded of both anime and those old 2D adverture games where all the dynamic objects are lit more flatly or just a bit off from the environment. Oddly, I find this actually quite nice. In practice though, I think it would be sufficently realistic.
 
Last edited by a moderator:
Back
Top