more like 15-20 years if we keep making advances at the same rate

flymike, But the rate of development has slowed in recent years. But who knows it may pick up again soon.

someday...biological computers powered by genetically engineered brains?

Quote from tristancliffe :

In 200 years will all graphics be polygon based? I doubt it.

not all and neither is it now but its quite likely that it will stick with non flat polygons for quite a while
even if its only for no better reason than that splines and other forms of curved polygons have been the most sensible way to create surfaces for a long time

is a hugely beneficial and efficient way to store 3d data with an arbitrary amount of detail

It's only 'highly efficient' because we haven't thought of anything better yet. Don't misunderstand me; I think polygons are fine, and the level of detail achieved with them is pretty staggering (with various surface tessellation and bump mapping techniques etc etc), but to suggest they are great simply because it's the best we have is to say that Nuclear Fission is the best sort of nuclear power because we haven't made Fusion work yet.

no its highly efficient because its basically what happens when you apply the sampling theorem to 3 dimensions and make it even more efficient by allowing you to adjust the sampling frequency freely to reduce data in smooth areas even more

theres a mathematical optimality there between the amount of surface detail you can represent and the amount of data necessary to do that

Some interview with guys from Euclideon:
http://www.youtube.com/watch?v=JVB1ayT6Fdc
At 22:00 you can see them playing the realtime demo with a gamepad on a laptop.

Instead of video camera a screen recording would have been much more fun. Looks impressive but at the same it's just too good to be true. I'm sure there are some people who could make an optimized scene like that.

I want Randi to take a look

Quote from Crashgate3 :

I'd say polygon tesselation is the next quantum leap as far as surface representation goes.

+1. Displacement mapping still bogs down rendering times on a number of commercial renderers so to see it happening in realtime is quite something. Theoretically you could start to see detail on characters in games that look very similar to the original zbrush sculpt (instead of using a normal map on low poly geometry)- so the future is kinda already here and it's bright.

Seen point cloud rendering years ago. Not so much of a news.