Quote
Originally posted by: nemo_83
It could prove less taxing to load six frames of a room than to render it in real time allowing the power to be focused on rendering the character in real time, streaming AI, physics, and lighting.
Imagine for example, Samus' ball, prerendered using cube mapping, perfectly round. Imagine her arm cannon prerendered (as if it came right out of a hollywood movie) in six dimensions all at once allowing you to see it in 3D.
Cube mapping doesn't do this. I don't know where that idea came from, but it looks like the interpretations everyone seems to love on Gamespot's forums. "Panoramic cubemapping" only allows for what is essentially a skybox. Creating true 3d objects out of 2d images is just BS through and through. The closest thing to what you've described is displacement mapping (a whole new can of worms), that in a nutshell converts textured bumps and depth into ACTUAL bumps and depth (ie more polys), allowing for the creation of very complicated detail on models quickly and accurately. For instance, if a displacement map of a rough, pockmarked brick wall was used on a flat surface, it would would modify the geometery of the wall to actually have pocks, bumps, and spaces between the actual bricks. It's quick and efficient compared to actually modelling the individual details in the wall by hand, but it is very taxing on the CPU and GPU because of the huge increase in polygons.
I've seen this theory of getting enormous detail at no cost many times before, just applied to a different name. First it was applied to NURBS, this cubemapping bit, and now displacement mapping -- everyone seems to latch onto the idea that these are some sort of magic bullet that will allow for the rendering of complex, fully 3d objects without the use of system resources. This is, of course, utter bollocks. Everytime someone shoots the rumor down, it just gets shifted to another CG technology that people don't fully understand.
ANYWAYS, this is an example of panoramic cubemapping:
http://www.panoramas.dk/fullscreen5/f14a-pope.htmlYou need quicktime to view it, and it ought to give a pretty good idea of what the limits of this technology entail.
You had the right idea when you described it as a more dynamic take on what goes on in Resident Evil. I *think* that Nintendo's patent is a step forward in that it eliminates the need for a polygonal mesh for the interaction of realtime objects and the pre-rendered scene, somehow through the use of a corresponding depthmap for each face of the cubemap.
I'm by no means a 3d modeling guru, but I have a decent understanding of the basics. Also keep in mind that if any of these fantastic rumors/interpretations of patents had any weight behind them, you'd see all sorts of stuff about it on anandtech, gamasutra, etc, and NOT a casual gaming forum like GS or PGC.
PANORAMIC CUBEMAP