H'mmm, sound almost to good but when you think about it, at my screen resolution of 1280 by 1024, I only have 1.3 mega pixels to use.
In essence it sounds like they've taken the swapping down to the pixel level. As an object gets closer and fills more of the screen, show more points belonging to that object.
I wonder how easy it will be to make the objects themselves. It sounds like a subtraction system where the master object contains all the points it could ever need based on minimum distance and then as you back away, fewer and fewer are used, like a continuously variable LOD.
Things could get interesting if and when the next generation of games start using this.
