Unlimited Detail Real-Time Rendering Tech

This is probably repost but I found this somewhere and thought it was so amazing it was worth sharing and a little disappointed this wasn't shown at CES this year

Unlimited Detail Real-Time Rendering Technology: Preview 2011 - YouTube

 

It's vaporware.

 

well, I certainly hope they will manage to unleash this tech, it would be awesome to have an atom model instead of a geometrical model in computer graphics

 

Past year two and still nothing of substance arrived. If they produced a tech demo for download and use, I'd have confidence, but it's basically voxel tech that went bye bye nearly two decades ago because the required horsepower was much more than required though use of a dedicated GPU. Heck, show me a demo with interactive and animated objects not static, and I might give it a peek again.

Great in theory but not in practice.

 

Changing the design model wouldn't necessarily be a good thing it all. Changing the rendering model wouldn't make any difference to you as the user, as you are abstracted from the model. You just see the image on the screen. As it stands, I think artists would prefer working with polygons rather than voxels in general, so that much is out.

In the best case, it would make artist's lives easier, as they would just produce one high resolution polygonal model for games.

That said, this technology is still overhyped and misnamed, still in the best case. Vaporware in the worst case.

 

I haven't seen a game that used voxel rendering since the Delta Force series.

 

There are some modern indie games that use voxels, crysis uses *some* voxels, and of course a handful of older titles.

And minecraft of course.

 

Well yeah, Minecraft is an obvious one.

What I meant is, I look at that video and see Delta Force, not exciting ground breaking technology. Looks crappy to me.

 

definitely it is better for the artist to work with an easier model, but I think an advance in the field is needed, after all only so many polygons you can use in an object, at some point we will have sharp corners even the GPU power is increased by 10 or 20..

I am always friends with the new tech I hope they will eventually come up with something of worth

exactly, but that is what advances a field, a.k.a. ridiculous research

 

The point is that this isn't new. There's a reason voxel based rendering never really caught on in mainstream gaming.

 

Granted I hardly know much about this but what keeps be interested is that THIS type of research and technology is what the future of gaming holds...as apposed to mobile gaming, motion controls, etc. While all that stuff is great its technology similar to this that will push us in to the new age of gaming

 

Polygons are basically made of three voxels joined together anyway, and there's really no limit on how small you can make them either.

Secondly, there the fact that if you have something made out of that many voxels, you pretty much can't rotate or move it at all, due to the fact that every single point has to be moved.

I can see the advantage of using voxels in the design phase, and many game maps are in fact made using voxels, however they're then converted over to polygon meshes so they work well.

Then you have the fact that level of detail on geometry doesn't have nearly as much of a visual effect as other things such as shader effects. All they've got to show for themselves is a rather shady presentation which merely blags on about levels of detail in a rather un-convincing manor.

In case you can't tell, I'm not buying it.

 

Related Article - Exploring Euclideon's Unlimited Detail Engine - Features - www.GameInformer.com

tl;dr version - chap from gameinformer went to their studio, played around with the demo, was pretty amazed by it and was convinced by their answers for the criticisms against them

 

hmmmm...very interesting. I hope this is for real and not some bogus hype.

Did someone come up with a powerful voxel compression algorithm? It is very interesting indeed

 

Voxels aren't an advance in the technology. It's a sidestep at best. In the case of BOTH polygons and voxels, you have sharp corners and detail limited by the effort you put into the design of the model. There's no difference on that end in regards to polygons vs. voxels.

Yes, with current hardware, you can only render a limited number of polygons in real time, but the same is true of voxels as well. I could come up with a tech demo tomorrow and declare that I have an engine with the capacity for unlimited polygon rendering, using some special proprietary selective rendering algorithm. It would cause an equivalent amount of head turning. Even if this doesn't end up as vaporware, it's important to understand that voxels themselves don't represent an advance in the field. They are just different.

It's possible that these people came up with some novel rendering algorithm which happens to operate on voxels and that their algorithm represents an advance in the industry. It won't have any effect on how artistic design is accomplished, though. You can easily convert polys -> voxels (because they are so similar, and neither represents an advancement in technology over the other), and so artists will continue to use those tools to take advantage of the animation and modelling features polygons provide, which is completely separate from the rendering.

Again, neither voxels nor polygons allow for unlimited detail. If you laser scan an object, the detail is limited by the resolution of your scanner. If you design an object from scratch by modelling, the detail is limited by the amount of time the modeller puts into their model. In both cases, sharp edges are inevitable if you zoom in far enough. The only modern standardized method we have of actually getting unlimited detail is with NURBS surfaces, and there's no way to render NURBS surfaces with the standard graphics pipeline, so they get converted to polygons first.

The degree of "unlimited" which this company is referring to, is the type of unlimited where it's geometrically detailed enough to render smoothly on our modern screen. It's "unlimited" but not "infinite". It's important to note that the same thing can be achieved easily with polygons. The limiting factor is the GPU's ability to handle it in real time in games (in the industry, there's no issue with the design phase in this area, or with CGI in movies, which don't need to be rendered in real time).

More importantly, image quality is extremely perceptual. Our eyes and brains are very biased in terms of how we perceive detail and reality. As someone else noted, geometry is not a major image quality bottleneck right now. Lighting and shading (together), as well as animation are the major problematic areas in real time rendering.

---

TL;DR

1. Voxels are not an improvement over pixels
2. Both voxels and pixels are subject to the same limits in terms of detail. You can be as detailed as you want, the constraint is how much time and money you can invest into adding geometric detail to a model.
3. It's possible that this company came up with a proprietary real time rendering algorithm which is novel in terms of it's ability to display geometry in real time applications like games, but that won't have any effect on design tools which artists use to prepare models for games.

 

I am not saying it is new or not (quite frankly I don't have enough knowledge in this to talk), I am just saying ridiculous research advances the field, most of the time what you find is nothing, no worth whatsoever, but it helps you to figure out another problem, if not you particularly, somebody else, I am a long time grad student, that's my experience about research, I did a ton of b*llsh*t, which led me to the point where I am, and I learned from the b*llsh*t of other people (to avoid or to learn from them)..

not the geniuses, the cultures / communities advances the science after all..

still, thanks for the information though! I learned from your post quite a lot

 

been seeing threads on this for the past couple of years in nbr. and they still talk about the same thing- a theory they cant put into practice and a promise they cant keep.

 

Voxels isn't new. As already stated it was popular with NovaLogic Delta Force games many years ago. But I think current tech is well beyond what voxels can provide. I like the *idea* especially if you can assign a physical property to each voxel in how it will act or react to the environment. But the practicality of it is not realistic. It would be nice to have an environment with VOLUME instead of hollow shells. But we're a far cry from the computational ability of that.

 

it doesn't need to be new a little example is one of my favorite theorems, it was actually proved at 1894, however the power of the theorem was discovered by 1933.. again I am telling, it may not be easy to see what fruits can grow from research (I am not saying there maybe something into this thing, I absolutely have no idea, but I learned to give the benefit of doubt to people working hard, 2 years of unfruitful research and they are still going at it, that IS something), that is WHY it is tough, every once in a while you look to absolute emptiness to discover beautiful results...

 

There's plenty of research being done in all sorts of fields of graphics, including voxels. They just aren't often used in video games, so consumers aren't quite as aware of them. It's not like UD is the only private company studying voxels.

As far as volumes vs. hollow shells, it really doesn't matter in terms of a virtual environment. The whole thing is virtual anyway. Calculating out the interior of a virtual sphere, and then looking at that sphere from the outside doesn't make it any more or less of a sphere.

Also, the voxel geometry as displayed in these examples is not filled in with extra voxels, either. It's all flat, just like polygons.

In general, the big benefit of voxels is to be able to visualize volume. This generally isn't important in games, but it could be important in medical visualization. Hence, voxels tend to be used in medical visualization. Tada.

So, in short, voxel study is good. It's also important to meter our expectations appropriately, especially in regards to this company. This is still vaporware at the moment. Although they seem to have solved the problem of geometry limitations (which honestly isn't that big of a deal, and will also resolve itself in hardware in a few years) - they have introduced other problems which are a huge deal: namely- lighting and animating. (Keep in mind, lighting includes shading and shadows)

 

definitely! maybe us gamers it won't be a big deal, but other fields of science can profit from it also yes, real time lightning is what led to GPUs right? in their video they also admit that, they cannot do lightning properly right now, let's hope they will eventually come up with something rather than vaporware man, you really know quite a lot in this, are you working at software company?

 

lighting is one of the very early and major GPU features, yes.

 

Is he just storing the map as a giant 3d array, or is there something more complicated behind this? If it's simply a giant 3d array then it doesn't matter if there are atoms inside a space, the memory will be used anyway.

 

If that was just one giant 3D array, your typical game map would consist of tens of GB of data for one "area". The one advantage I can see is with destructible environments now you could have actual destruction because it would be a solid model and not just some animated sequence.

 

oh no, a 1000^3 component 3d array would take 1000^3 * 8 bytes (integers), not a wise choice, nothing can store it, yet alone process it, it is about 8GB

 

So essentially he's not really using voxels at all. I'd liken it more to a very high mesh polygonal model, just without the links between points.

Despite my previous critiscism, I'm impressed that's he's managed to find a way to trace each pixels path given that he's not working with a basic array.

 

No. He's explained this many times. It's not a typtical graphics engines and it does not have ANY polygons. There are no polygons. There are no voxels.

It's a search algorithm. In the limitless points, his search algorithm only searches for the points that are to be rendered. It's like google search he said. It's much quicker and faster than you would think and that's why it's unlimited detail. He's not limited to voxels or limited to polygons. That's also why it's not a shader process on a GPU. But he is looking into utilizing GPGPU power for it though. But again the GPU will just be running search algorithms.

 

That doesn't work the way you think it does. You have to separate the idea of damage modelling, including destruction (difficult) from rendering that destruction (relatively easy). They aren't the same thing. It takes a lot of calculation to model destruction physics, and it isn't a good idea to tie that in with the rendering model. Without getting too into the detail, the physical model depends on the interaction of objects in every instant. You have to calculate the state of all the objects relevant to the physics for every state to properly approximate the physical model. You don't need to know the state of all objects for the rendering model. That's the whole point behind this, and that's how UD is supposed to work. If you have to calculate all points for every frame, you wouldn't be able to run the thing in real time. They have some novel method of avoiding that problem with RENDERING, but it's unavoidable with physics. You would have to maintain a separate physics model to have physics in a game with this type of RENDERING engine. Moreover, physics engines in games are already built on interactions between objects, not polygons. Don't forget that polygons are a tool used to show us objects which we can see (render). That doesn't mean that the computation of physics has to operate on the polygons themselves. Objects can be DRAWN with polygons and modeled with mass and physics. Just because you draw an object with a polygon on a screen, doesn't mean that object is empty. It's a virtual object anyway. "Empty" has no meaning in that sense. You're thinking in the box exclusively.

You may or may not know it, but you're describing voxels. All the voxels which potentially may be selected have to be stored in memory. Whether it uses the GPU or CPU doesn't really matter. What is important is whether or not this type of engine will be able to handle complex dynamic lighting and shading, which is where we struggle with graphics currently.

 

Moreover, although some games do include precomputed physics (which is reduced to an animated sequence, just like you said- good job) - not all damage and destruction in games has those same limitations. Moreover, the games which do have those limits have them because of the time consuming nature of calculating physics. It has nothing to do with polygons. Calculating physics on a bunch of tiny atoms is going to take an astronomical amount of time, and because each state is dependent on the previous state, you can't use an amazing search algorithm to avoid doing those calculations.

 

Hey guys, theres this company called Steorn that developed a perpetual motion machine. Scope it out.

 

They are not the same. You do not work with unlimited technology or work for Euclideon. You do not know what you are talking about, neither do I. So the best I can do is just tell what Euclideon has said. It is NOT voxel technology. They make that clear, repeatedly.

 

Phantom Entertainment - Wikipedia, the free encyclopedia

Steorn - Wikipedia, the free encyclopedia

History is littered with people making grandiose claims with little or no grounding in reality.

The truth of it is that if you are talking about volumetric particles as opposed to polygons, you are talking about voxels. It does not matter how small those voxels are, they are still voxels. Also, their screenshots are unimpressive and until the public sees models interacting with each other or SOMETHING happening besides grainy prebuilt environments, this "technology" is irrelevant.

 

Yes Grandiose statements without the proof to back it up does not lead to the largest government grant provided by the Australian government to fund software technology.

You don't know what you are talking about. No one does here. None of us are investors who were provided the hard facts about what Unlimited detail is. None of have read the grant proposal that the Euclideon was awarded for. So stop talking as if you know better than Euclideon and you know what the technology is.

The arrogance in this thread is just mind boggling. Especially since I doubt any of you have any experience whatsoever in creating a graphics engine.

I'm not here to defend Euclideon. I'm just flabbergasted how any of you can claim to know what Unlimited detail technology is and say Euclideon are liars and wrong.

What you are all doing is slandering and spreading libel about Euclideon. What you are doing can be prosecuted by law. What you are doing is accusing Euclideon of fraud. Euclideon state this is not voxel technology. You are saying Euclideon are wrong and are liars, that they are fraud. Just because this is the internet doesn't mean you can say whatever you want. Even if Euclideon won't take you to court, you ought to keep all of this in mind before accusing a company of fraud. Are you really prepared in court to prove Unlimited Detail technology is voxel technology and that Euclideon are liars? Highly doubt it.

 

I lol'd. Lots of companies have been able to sucker idiots into investing millions of dollars into failed products. If they want to take me to court for calling their "technology" voxel rendering, they're more than welcome to.

 

"Polygon to voxel conversion is still young but is working as expected." - Bruce Dell (CEO, Euclideon)

"Unlimited Detail is a sorting algorithm that retrieves only the 3D atoms (I won't say voxels any more. It seems that word doesn't have the prestige in the games industry that it enjoys in medicine and the sciences) ... " - Bruce Dell (CEO, Euclideon)

----

Things which are true tend not to be libel (slander). He states plainly that he has elected to use an alternate term because he is concerned with prestige.

If they led you to believe that their technology is not using voxels, this shows pretty clearly that they were, in fact, lying.

While I appreciate the healthy skepticism by raising doubt about what is known (by yourself, and by others) - a TRUE skeptic would not assume either way. It's certainly possible from your perspective that I know what I'm talking about, or that I don't.

As it turns out, it's extremely easy to show that their technology does indeed use voxels, even without the quotations. You merely listen to the explanation of the game engine, and then look up the word "voxel". It's literally that simple.

I would also appreciate it if you would back off as far as judging the legal status of my criticism, unless you have a more serious concern. Claiming that someone has broken the law, especially when they have not, is actually much more in line with what libel (slander) is, than anything else written here.

 

So much rage in this thread, I may subscribe.

Anyway, 3d pixels are voxels. That's pretty much the defninition of a voxel. Done. You can say they're something new and revolutionary, just like razer said the razerblade would be, but that doesn't mean it's true.

Also, quite frankly I'm not concerned about being sent to court. Firstly, it's an Australian company and I live in the UK, therefore I'm not bound by their laws. Secondly, what hulawafu77 is describing would only ever happen in America, I don't mean to make stereotypes, but the majority of ridiculous lawsuits I hear about come from America. It might just be internet demographics, but there's bound to be some truth underneath.

Lastly, besides the very first line of this post, none of it was intentional trolling.

 

I don't think anyone is concerned about being sent to court, but libel is a moral wrong as well. However, it's totally misplaced to be pointing fingers and throwing around that term loosely.

 

There's a difference between casting doubts and deliberately spreading false information.

As for whether it's wrong or not, that really depends on the circumstances. There are some completely diabolical companies in the world where, technically if you even told the truth about them it could still be classified as libel since it would be negatively affecting their image.

 

Writing down something that negatively affects the image of something isn't sufficient to merit libel. An obvious example is with reviews. Let's say a game is terrible, and you write that it's terrible. It could negatively affect the image of the development studio, among other entities. However, this would not be libel. As a general rule, if you say something that is demonstrably true about a person (or a company), it wouldn't be libel.