Unlimited Detail... soon.
polycounter lvl 19
Euclideon, after a year out of the limelight, are back with another demonstration of where their unlimited detail technology is at.
Euclideon is showing off their in-progress technology that dismisses polygons as a visual measurement for videogames. As game artists, how do you feel about the possibility of moving away from the polygonal world?
https://www.youtube.com/watch?v=00gAbgBu8R4
Replies
How does animation work?
How does physics work?
A hell of a lot is left out. Its a cool tech demo... but im saying cautiously suspicious.
From his past reputation though I give it about a 10% chance we will see anything technical or an much less an SDK by the end of the year.
And about 5% percent chance we will see a of any game of any value out for at least a year til two yrs after that.
Even the previous video looked like just a whole buch of instancing with just one object, we can do that know especially with one light and not that accurate of shadows and no reflection and no texture but hey maybe thats why it took him another year when he realized that
I'm also wondering about some of the things that have been left out. The biggest one for me is storage. How many megs is that elephant.. or one of those rocks?
Either way very exciting stuff. Can't wait to see more. Time to make Myst again :P
ZBrush also uses point-based data (pixols) which is why it can handle dozens of millions of "polygons" (ever wonder why you can only see the "point" count?), so this tech has proven its performance advantage over polygon based-tech. I mean this is why regular 3D apps have a hard time with hi-def ZBrush exports -- they can only display polygons.
It's too bad though that we don't have more info on the performance side of things: Ok, 20FPS, but on what kind of hardware ? How much memory does a model+texture take ? Is the texture information stored per voxel also (which would improve efficiency) ? How are they planning to do animation ?
The storage and memory issues were the first question I wondered about. So far everything they've shown has been heavily instanced.
Shall wait and see.
point-clouds/voxels don't really fair well for character animation. They would be limited to static geo, moving stuff like characters could still be polygons, but just a higher count. A 100,000 polygonal character in a super detailed environment would look awesome!.
Stuff isn't going to jump to this like tomorrow, but over time it is kinda the way the future is heading. Right now the industry imo, is quite far behind were it should be.
much faster than regular 3D packages.
For those that want to know how voxels actually work, their advantages and limitations : http://www.tomshardware.com/reviews/voxel-ray-casting,2423.html
But like everyone else said, there are still a lot of unanswered questions. Storage, texturing, memory, etc...
Also how would water work?? Seems that they still have quite a lot of work in head of them.
But still awesome to see that people are continuing to innovate!
Still, if it gets rid of grass cards in games it'll be considered a success in my book.
I imagine only environments at the start would use this kind of tech... But indoors? And dynamic lighting? Lets see...
Q: How do textures work?
I would imagine each point has a color. How you color those points is probably anyone's guess. Probably something similar to PTex, or projection mapping.
_______________
Q: How does animation work?
Point Cloud Animation, similar to vertex animation but with a lot more points. It would probably be smarter to mix and match technology. They have point cloud animation tests they've done with small objects. I doubt you could do much on the fly animation with point cloud data, things like animation blending and rag dolls would be insanely complex.
_______________
Q: How does physics work?
Right now they don't... But given he's a bit of a "real atom" zealot I would expect him to take this to the next level otherwise he'll have to fake it like everyone else is doing. There is a serious danger in trying to recreate the world atom for atom in 3D while forsaking everything else that goes into making the world of a game.
Go Notch
OH SH#&!
As far as why they are addressing the general populous and not engineers is... well, if your company was producing the tech of the future, would you want to tip your hand too early or later? They are generating hype in the populous that buy's games. This will steer the industry much more, I feel, than development studio's looking at it and saying, "Eh, it's cool, but it's a lot of changes. We have a way of doing things now that works, so thanks, but not right now." Seems smart to me.
My point is, it's new tech. Every time new tech comes around, it has it's challenges and people have their doubts. Early adopters almost always get burned, so it's nature to hold back. But seriously... how about some applause for progress?
This is forward thinking! what ALL of us should be doing. Yes lots of stuff not answered! BUT, because of this, someone else out there may know how to help them with those issues. We would had never progressed were we are today if someone said "no you can't do that, thats impossible, our hardware can't handle that for another 100 years!"..... guys.. its only be 26 years sense the NES! I dont see why a jump like this is not possible. Keep an open mind! instead of seeing the problems, become a problem solver and see how they could fix those issues. And don't assume anything! we don't know that they are only showing us the tip of the ice burg or not. Lets build them up with encouragement!... not tear them down.... So theres my two cents on this.
Great work Euclideon! Can't wait to see what else you surprise us with!
My problem doesn't steam from anything that they are showing. My problems come from how they are toting this as new tech, when it isn't. It has been done before.
AMD and Intel can and would support this tech if it were viable for what needs to be done for games. They have a lot of money that they can throw around for R&D. These guys are looking for money from outside investors. And instanced geo does not make the game. I do applaud them for what they have done, but saying that their tech is what will change games is extremely naive.
Time will tell, I hope I'm wrong, but I wouldn't call myself an optimist either.
yeah, its for sure not for them to make any claim of this being new tech. I have seen the old videos from what has been done before. But it does seem like they have taken it to a whole new level. As well seems like they are actually able to continue pushing this tech with more funds.
But i think some people are a bit harsh, but the guy does annoy me his whole "Look at this", the lighting is pure garb at the moment it seems. We will see if they deliver on their so called SDK in a few months. lol?.
http://notch.tumblr.com/
Cubic Pixels are all very well, but can they do a better job for a lower computational cost than texture mapped polys? In gun-toting space marine game 97, do you really need every grain of dirt modelled?
Also, if you want UNLIMITED DETAILl, I think you'll find that Tesseractic Pixels ( or Tesels) are the way to go.