Jintor said:
Friend Serrels and KotakuAU have
posted an interview.
Reading this interview gave me some more insight into how the system works.
It's an extremely cool idea. The whole thing isn't really 'unlimited voxel' or 'unlimited atoms' really at all. I mean it is, they're using voxel based graphics, in the sense that it is the crazy super human over the top ultra intensive graphics technology that everyone knows about, but their approach as to how to display that on the screen is totally different.
It's a very interesting an amazing idea actually - the idea that they're just picking the pixels to output to the screen rather than rending the whole scene and then picking the pixels to display, it is ONLY picking the pixels that need to be rendered and that is the entire point of having so many billions of little atoms because that single atom that needs to be displayed becomes the pixel in reference.
It's an amazing concept and an amazing idea. Thinking about how to get around the complexities of animating something like that, is a baffling prospect. I think it could be done, but that's where the horsepower is going to start to come into things. That is where things may start to slow down a bit, and the problem is - they CAN'T mix polygons and voxels using this pixel picking technology so it will have to be ALL voxels because of the way the 'unlimited detail' works by picking just the pixel it needs to render.
So. For what we've seen here, a still scene with amazing levels of detail, that's the easy part, and I fully believe that what they've achieved so far is legit and is doable - it's the next step, making it into a working game engine that is feasible and usable on current technology... well, that's going to be the hard part - as everyone has said.
I really think they are onto something here, it's kind of mind numbingly obvious to just like "FACEPALM" just render the pixel you need to with the colour for the screen, but getting it to work...
I'm just trying to think about how they would do it. If they're picking the right pixel needed from the perspective of the 3D scene to essentially make a 'bitmap' for each frame, then to animate everything you're going to have to pre-render everything - but then that's just going back to regular old graphics rendering techniques...
BAH! How the fuck are they going to do it... Interesting times!