Do you work for nVidia or something? Tessellation is completely different to the concept presented by the OP. This is actually building virtual objects from tiny "atoms", something that is usually not possible due to the processing power that would require. These guys claim to have invented a way of getting around that problem. However I am still questioning whether their claims are real or not, because the implications are quite huge. I hope they aren't lying. I can't really see why they would.
I'm not really that impressed. Tessellation is already being used in many modern games if you have a quick direct x 11 capable graphics card - which practically eliminates the blockiness inherit to using polygons.
You can scan in things and it will look exactly like what you scanned in with no poly count consideration...it runs on even a weak computer and makes all things smooth and breathtaking...not to mention the ability to use the new infinate geometry in practical environmental settings (dust storm is literal bits of sand picked up randomly from the scene, could even to proper destruction.
This looks to be nothing more than a reincarnation of the voxel idea, which (for those of you who have been coding long enough) remember preceded polygonal modeling. It is a very cool idea, dont get me wrong, but those of us who have been doing this a while already knew that voxels would overtake polygons eventually, the hardware just had to catch up.
Originally posted by Laokin
reply to post by SaturnFX
Realistic destruction is still a no go. Also, it doesn't run on weak computers... he gets a pretty low frame rate on the computer he is demoing it on... and even says the frame rate is still pretty rough and that they have advanced versions that have better lighting models that weren't ready to show yet because they aren't finished and also have low frame rates.....
Realistic destruction is a no go because there would still have to be a separate physics engine. Remember these different "game atoms" need to have different physical properties to them like real atoms. I.E. A wooden table would be made up of millions of atoms... in which, when they fuse they need to act like wood. That means the cpu would have to track the physical interaction between millions of digital atomic particles vs a couple hundred polygons which are then grouped into a low amount of "hitboxes" which in props can be as low as 1...
In reality, rag doll physics will be about the same, but destruction is going to end up being harder to do if attempting to do it off of any modern physics engine of today and trying to scale it to millions and billions of atoms. These guys didn't make an unlimited detail physics engine to go with it... In the end, the physics will be identical to now... they will have atoms grouped into hit boxes and do canned animations out of the atoms like normal physics now.
Ya dig? Clipping is fundamentally a different problem separate from rendering techniques. This isn't going to help cgi either... At all. Every art asset in the engine that runs on atoms is still built in polygons... The engine converts it to atoms.
They would need to make a different renderer for suites like maya and 3dsmax that converts all assets upon completion into atoms before animation in order for that to improve CGI. It probably still won't help CGI too much anyhow... since this has nothing to do with texture constraints. All it would do is marginally speed up the render time. Something that used to take 5 weeks to render might only take 2 now.... This would largely depend on the efficiency of the version built for the suites though.....edit on 2-8-2011 by Laokin because: (no reason given)
Originally posted by Nick_X
This is apparently not real at all and a scam and has been debunked here:
Notch Debunking of Unlimitedness
