Too bad there's no informtation on the page on how they accomplish this. It's impossible to tell whether or not the method is suitable for realtime use. Their claim of 100hz is useless without an actual demo with characters in an actual environment at the level of detail needed for a real game.
Hopefully though, this is all it's cracked up to be, and not patented. Radiosity though, is expensive. They say infinite light source bounces, but that's sounds like BS to me. As far as I know, each bounce would require another render, or at least another calculation in a shader, even if they've found a way to accelerate the process by deteremining for each point ahead of time all the texels which contribute to its brightness. I don't see any other way to do it. You have to know how much light surface A has falling on it before you can calculate how much of that light is contributing to the illumination of surface B.
Hopefully though, this is all it's cracked up to be, and not patented. Radiosity though, is expensive. They say infinite light source bounces, but that's sounds like BS to me. As far as I know, each bounce would require another render, or at least another calculation in a shader, even if they've found a way to accelerate the process by deteremining for each point ahead of time all the texels which contribute to its brightness. I don't see any other way to do it. You have to know how much light surface A has falling on it before you can calculate how much of that light is contributing to the illumination of surface B.