The future of graphics, THE FUTURE.

Sooo, i saw this today. In simpler words, instead of using Polygon shapes to make game environments, they can instead use little "atoms" which can allow amazing detail and color.

Watch the video. Really. Do it. Now. Please.

I've seen the video, and there are a few comments to make.

Right now, there is no way to animate this. Right now, Polygons move by applying a skeleton that pulls on Vertices. For these tiny atoms, it would take your entire GPU just to process a single jump, let alone a dynamic surrounding. It's not feasible. Lighting this would be a nightmare because once again, you're dealing with billions of tiny atoms trying to move with light.

Second point, this isn't his first time. He's been doing this since 2003. Pops up, talks against the Polygon-Demons, gets his hands on some sweet investment cash, and disappears.

Not to be negative of course. This is still some great detail, and I really hope he delivers soon with a way to create better animations.

Yeah that video is pretty old already, it sounds interesting and all but it's impossible to use in real world situations.


As nerkel said, animations are done via skeletons, as you pull and drag each joint in certain directions to create motion, kind of like a puppet. Second, the tiny atom thing isn't actually possible without MASSIVE performance drop - that amount of detail doesn't just simply happen without the CPU and GPU working in conjunction, and when the CPU and GPU have to render that much detail, without some kind of model/texture swap/pop-in, it becomes extremely taxing.

But I'd say the biggest hurdle is that EVERYONE is using polygons, we've been doing it for over 10+ years, and it works just fine.

Hell, even Unreal Engine can make this:

Samaritan Demo

Albeit it took 2 GTX 580's running in SLi, but still.

I imagine it might be possible to animate that if what he had said about a polygon-atom converter was true. Keeping the skeleton underneath, but even that sounds shaky and like it would eat your processing power.

Does look nifty though, and here in another 20 years I'm sure we'll be enjoying it.

loling forever at the narrator's herpaderp voice. Imagine holding a conversation with him irl without dying a little on the inside.

That... voice. I just tried to ignore it.

Also, the Wisdom of Notch shall see us through! It's a SCAM!

http://notch.tumblr.com/post/8386977075/its-a-scam

Well, you guys are still thinking of static polynomial/rasterized rendering. This video is claiming using "3D atoms" or "voxels"... which is vector based rendering. That means that it basical 3D pixels are dynamically "drawn" at a specific resolution. So as you zoom in, those points are re-calculated to fit. It's exactly like AI or Flash - you draw a line at 300 ppi. Whether you're at 3% zoom or 30000% zoom, that line is still 300 ppi on your screen... just voxels are volumaic instead of just 2D.

All this means is that you don't need a supercomputer since the 30,000,000,000 polygon object is never 30,000,000,000 polygons.

Sadly this video is more snake oil than truth. Looked on their site and they're definitely fiishing for funding rather than truth. Voxels have been around for a long time and are not utilized for two reasons. First is the inaccuracy. You're not conrolling the rasterization, so specific details may never get drawn or objects emphasized that you don't want to be, etc. This reason alone is enough to shoot down most desire to utilize this in movies and games where precise detail matters. Then of course is the line drawn where scalar calculations would exceed the computational power needed to render an X amount of polygons. The more complex and object is, and the more of them there are, you begin to cross the line. This happens to be far, but the RAM requirements would be passed with only a few objects -- all those points have to be stored in RAM. I've had AI art take GIGS of RAM before. XD

Further proof that these guys are all talk and no show is the simple "too good to be true" factor. While they are indeed working with an already existing technology that has a sound theory that could theoritically make up for its disadvantages, they have no explanation how ANY of their coding actually works. They have no explanation of at what point rigging takes place (on the polygon object or the point data object), how anything is translated, nothing. Plus why is this only a team of 9 and no major company taking them over? Why is it just a youtube video instead of an actual live demonstration to all these companies they say their technology is courting too? Actually having seen them actively pan the camera and showing the environment instead of something that could have been potentially pre-rendered is a world of a difference.

All this just boils down to that I'm not holding my breath. :p

Theres also the issue that everything they make seems kinda....well, ROUND.

Eh... I find the bigger problem to be the repeated objects. They claim their trillion points of data... but that's just bull. Every clump of grass is repeated. They only had like 5 different rocks. The leaves were repeated. The trees were repeated. After all is said and done, I'd be surprised if there was more than a dozen different objects in the entire thing. As a duplicated object can be referenced rather than individually rendered, it saves immensely on computing power. Their "laziness" is also a convenient excuse to throw unrealistic scale capabilities.

But maybe it's all legit... ;)

Lets see them make a proper, playable game demo with it then the world will judge properly.


For now, it looks like crap to me and is probably fake.

Notch's (minecraft guy) analysis of it:

http://notch.tumblr.com/post/8386977075/its-a-scam

Video games should go back to sprites. Those were cool.

Just wondering what the point of this unlimited real time renderer is when skeleton articulated polygons does the job just fine and, as it stands, is only gettin' better?

Just sayin'.


It's like trying to invent a whole new way of opening a can when the old can opener in the kitchen drawer works just fine.

Polygon based graphics will always be limited by computational power. As computers/consoles get better, then they can handle more polygons. Sadly it is inevitably an Nth curve for this (calc II binomial formula reference anyone?) - you'll never reach infinity, and the improvements are going to slow down. Eventually, though, we'll reach a point where polygons will be so small that they could represent a curve that is perfect to our eyes if not technically perfect. This will be... maybe 10 years from now? But why wait 10 years? Scalar graphics are already using a perfect curve and can draw pixels/voxels at whatever density you want. As systems can easily handle drawing at a dpi that can fool our eye, and redraws to the same dpi no matter how far you zoom, perfectly smooth graphics are already possible.

If these guys somehow magically solved the problem of moving from polygonal graphics to vector based and made it where animation is not a matter of 50 billion caclulations a second (having to redraw each frame) then they have something that could possibly easily make up for even impresise redraws. Sadly they're only claiming the conversion process and that it can somehow be magically applied to animated graphics. I'm willing to bet that even if they have had developed the formulas to convert an item, it would be a static frame or object.... making it useless for the purposes they claim.

Kotaku recently managed to interview the guy in the video about it:

Linkage

A lot of spritework had real charm that I am yet to see with 3d graphics.
So, I'm inclined to agree to an extent.

Disagree, though most of the 3d stuff with charm is back in the old days when stylization was required as OMGSOREAL graphics weren't even a possibility. Still think Dreamcast had some of the best 3d stuff to be had.


But of course being a sprite artist I couldnt argue against a return to 2D :B

I still think FF7 looked great. Those boxy low-poly-count little sprite-esque models were awesome.

I can agree with that, those were freakin' stylish.
Using the limits of the console to their advantage?
That's freakin' classy.

Currently i'm playing Battlefield III really
this amazing game. Because according
to me this game gives a great graphics
& much more feature.