Unlimited Detail Real-Time Rendering Technology Preview 2011

[Ares]

http://www.youtube.com/watch?v=00gAbgBu8R4&feature=player_embedded#at=17

link- http://www.youtube.com/watch?v=00gAbgBu8R4&feature=player_embedded#at=17

[ximmy]

It will be fun when we can download ourself into a virtual world

[vacuum]

Simply amazing.

I'm sure there is a solid mathematical theory to back it up. Its probably one of those things that the only limiting factor previously was lack of an advanced theory. Like compressed sensing (is this using compressed sensing theory??). Previously everyone thought that you had to sample at twice the highest frequency component of a sound or image to be able to perfectly reproduce that sound or image (Nyquist theorem). Then one day a professor was randomly screwing around on matlab and perfectly reconstructed an image based on a low resolution sample. He published it and now the scientific community realizes that while Nyquist's theorem is true, it isn't necessary the best you can do under certain conditions, given your data is sparse and satisfies other criteria.

Come to think of it, these types of graphics probably do satisfy that criteria. They probably are using compressed sensing. They are probably sampling the atoms at a sample rate that is dependent on the distance from the object, and using compressed sensing to reconstruct the image. The "quality" of the image is perfectly proportional to its distance, so it doesn't matter how much is on the screen - what needs to be rendered is enough to fill up the pixels on your screen, no more, no less.

Read about compressed sensing here:
http://www.wired.com/magazine/2010/02/ff_algorithm/

If that's not what they're using I'd be even more interested in learning about their technology.

[vacuum]

In this video, they say it's based on a 3D search algorithm.


http://www.youtube.com/watch?v=Q-ATtrImCx4&feature=player_embedded



Perhaps their search algorithm, coupled with compressed sensing, is the key.



Here is some research on compressed sensing for computer graphics:
http://agl.unm.edu/researchAreasTest/CSProject.php
http://www.flintbox.com/public/project/7021
http://cvcweb.ices.utexas.edu/cvcwp/?p=139