The moment where he started relating the quality of a game to it's graphics... I may have switched it off.
Company invents technology 100k times better than existing ones
@crusader8463 said:
I remember seeing this video a few years ago. Was interesting.
Was this video not just uploaded in july of 2011? Did the guy also say when 3d gaming was announced which was not even 2 years ago.. and the video itself said 1 year later after he stated them launching the idea?
I might be wrong but it is a little hard to take them seriously when they try to advertise themselves while using a shitty microphone.
@DoctorWelch said:
I saw almost this exact thing about 2 years ago and I basically call bull shit again. If this technology actually was as amazing and revolutionary as this weirdly accented guy says it is, it would have already become an industry standard by now but clearly no one is using it or talking about it. So if it actually is real than that's awesome, but until game companies actually start using it then I dont really care.
You're crazy! I agree if this tech is real it's awesome, but it does take time to get this type of technology working. Do you think the first 3D games came about with the snap of your fingers?
It took years to develop the technology were using now, and it's still not perfect now. (texture popping for example) This will obviously take some time to be turned into a Software Development Kit for game developers to utilize and integrate into their engine builds. We could build 3D models well before we could use them in a game. We're able to build models with this type of technology through the use of cameras and polygonal conversion, but we still can't use them in a game.
It's all a matter of time to perfect the necessary stuff and of course, console and home computers to advance. The next generation was supposed to last at least another 4-5 years from 2010. so I'm sure by the time this stuff comes around to fruition, Consoles will be beefed up to support them. PC games will lead the way as usual. (I hope id Software gets there hands on this stuff asap).
BTW, id Software is already working on some pretty impressive engine advancements of their own.
@Raven10 said:
@Sjupp said:Speaking they had footage from Crysis 2 released only a couple months ago I doubt you saw it years ago.I saw this some years ago. If this was real or effective it would've made headlines and blown up like crazy. So fuck this.
Great point.
John Carmack's reaction from his twitter.
Re Euclideon, no chance of a game on current gen systems, but maybe several years from now. Production issues will be challenging.
This is just iterative maths applied to what are essentially voxels. Great for static objects, but useless for animated ones because the transformation matrices are too time consuming to do in real time. It also takes a lot of GPU to render those voxels even when not animated.
I don't understand how they could get their "atom" based modeling to not use up more resources than polygon based. Am I missing something or is that their business secret?
But they mention Crysis 2 in the video and that only came out this year?I saw this some years ago. If this was real or effective it would've made headlines and blown up like crazy. So fuck this.
Footage doesn't show any animations, physics or dynamic lighting. Their static lighting seems to be a full-lit system, where everything is as visible as it can be. Also when the camera shows more of their tech demo 'island', you can see the ground geometry is all flat and comprises of small 'tiles' set in different levels. Another problem is the data sizes. Just look at any current game sized around 6-10Gb and just think how much all that detail must take from your hard drive. A simple polygon needs only three points and those points can be shared with adjacent polygons (two needs 5 etc.). A filled elephant like theirs must contain so many 'atoms' / 'points' that storing all of that is going to be very taxing for your system, not to mention when all of that needs to be loaded into memory.
I don't predict polygons going anywhere for the next decade or so. This video is cool in theory, but they need to add more flexibility to their engine.
Well that settles it for me. Cool tech for next generation.From John CArmack's Twitter: "Re Euclideon, no chance of a game on current gen systems, but maybe several years from now. Production issues will be challenging."... if i am willing to listen to anybody saying anything about computer graphics is John Carmack.
I like how he sounds like Peter Molyneux, he's also full of shit like Peter. Just because they switch to "atoms" doesn't make it look 1000k times better...hell, I'd say that tech demo looks worse than most games, but at least it has grains of dirt right?
@MajorToms said:
@DoctorWelch said:
I saw almost this exact thing about 2 years ago and I basically call bull shit again. If this technology actually was as amazing and revolutionary as this weirdly accented guy says it is, it would have already become an industry standard by now but clearly no one is using it or talking about it. So if it actually is real than that's awesome, but until game companies actually start using it then I dont really care.
You're crazy! I agree if this tech is real it's awesome, but it does take time to get this type of technology working. Do you think the first 3D games came about with the snap of your fingers?
It took years to develop the technology were using now, and it's still not perfect now. (texture popping for example) This will obviously take some time to be turned into a Software Development Kit for game developers to utilize and integrate into their engine builds. We could build 3D models well before we could use them in a game. We're able to build models with this type of technology through the use of cameras and polygonal conversion, but we still can't use them in a game.
It's all a matter of time to perfect the necessary stuff and of course, console and home computers to advance. The next generation was supposed to last at least another 4-5 years from 2010. so I'm sure by the time this stuff comes around to fruition, Consoles will be beefed up to support them. PC games will lead the way as usual. (I hope id Software gets there hands on this stuff asap).
BTW, id Software is already working on some pretty impressive engine advancements of their own.
I'm not crazy at all in fact. If this technology actually mattered or actually existed they way its demonstrated than we would have heard about it from other sources other than random YouTube videos. Also from what I can tell nothing they guy says in this video, or the video from a couple years ago, proves that anything he says is possible. Just because you are using spheres instead of polygons means nothing except for the roundness of the shape instead of the squareness of the shape, but a certain level it doesnt matter if its squares or spheres, to the human eye it will all look the same. Also, just like last time I saw a video explaining this technology, no where in the video, or anywhere for that matter, does it actually explain why you can display "infinite" spheres. It's the biggest amount of shit I've ever heard. You are still going to need a video card that has to do the work of displaying all those spheres and thinking about them. So trying to say that your just going to magically make your computer display infinite stuff because it is made of spheres instead of squares is the biggest lie ever.
What I'm saying is if they actually have REAL technology that makes a difference and does what they say it does, than it will be cool and awesome and I cant wait. So until they actually explain everything they are claiming instead of just saying the same stupid shit over and over again with no real technological explanation of how things work than I'm calling bull shit.
I want to punch that man in the throat so he never speaks again.
Also, I don't find this impressive to begin with, even less so when there is no mention of what kind of hardware it would take to implement it, performance impact, etc. Empty promises.
@Raven10: @TrippinBungalow: @ChristianCastillo: @Anwar: Apparently they've renamed themselves. Some of that footage is the old stuff I saw when they were still calling themselves Unlimited DetailIt's still watermarked with the old footage. The fact that they seem serious about it has gotten me interested actually..
@RiotBananas said:
Dat voice.
Indeed.
Saw this when they were shopping it around last year to intel, AMD, etc. and they weren't picked up. So obviously there is a crippling flaw in their method. I can guarantee that it's not animation friendly for one, but I'm sure there's a plethora of more technical issues as well that they're trying to hide.
They've done a smart thing though by making their polygon converter. I was thinking before that there was simply no way that they would get very far if they were asking the industry to develop an entirely new toolset to work with their format. So good on them for that. If they work on making it a viable, inexpensive alternative to the norm, it may eventually find it's place. But whether it has a place in games I'm not sure. Pre-render CG artists have always instantly snapped up any tech that will make their jobs easier, but trying to insert a new method like this into games is extremely difficult. Even now most game engines tend to be extremely antiquated and don't react well - or quickly - to change. Unless this method ends up setting the world on fire, there's not much chance it will reach games.
As an environment artist this looks really interesting and I'd love to play around with it, stuff like making grains of dirt is impressive but is seems pretty unnecessary, also I laugh at the way he pronounces maya....
I can't see how it would be easy to animate in. You suddenly have to work out the physics for a colletion of connected molecules rather than a single object with a texture applied.
Not to mention trying to make textures. Try to make a good looking metal texture, computer console, or anything. You have to have every molecule with its own texture. Getting a coherent picture out of that can't be easy. No matter getting text, or a picture made out of it.
It looks cool. But let's see it do water, or any sort of real time physics.
Thats what a lot of people said about Onlive and while its not at good as they claimed, it actually does work rather well and keeps getting better. On the other hand, you could be right and this is all fake. My point is that I try not to judge something ether way until I have is in my hands.I'm pretty sure this is an elaborate joke, guys.
@Pinworm45 said:I know you may be being sarcastic, but that's why the guy was talking about fiction and non-fiction.@Anwar said:Where are we going to find a dragon to scan in?Modeling and texturing is also extremely tedius. Scanning real would objects is likely way faster and produces higher quality, at least in theory.they talk about scanning in real world objects, which is an extremely tedious thing to do and I don't see how that will help game designers.
Anyways, yes. This is very interesting, and I do believe it could work. It looks great, and as far as I'm concerned, I haven't seen anything better than this. I think developers should take a hold of this in the future.
It's a bit of everything. The core technology behind this is likely real or at least there really isn't much reason why it shouldn't be real. The teleshopping voice over however is full of bullshit. However you might see comparable levels of detail a few years down the line.What do you guys think? Is this just smoke and mirrors? Some elaborate ad campaign? Or the next step in visuals for video games?
The situation is basically this, our hardware is fast enough to render a polygon per pixel, so it could in theory with the right data fill the screen with "unlimited" detail (you can't cram more into a pixel then a pixel). The problems preventing that are however that you first need a scene that is that detailed and secondly you need a way to break the scene down to fewer polygons to make it possible to render (i.e. you want to render tiny tree with a million polygons in the distance as a single polygon or pixel, not as a million). The first problem can be easily solved by throwing a lot of artist at the problem or using copy&paste. For the second problem there exist plenty of space partitioning algorithms to store the scene in such a way that you can quickly traverse it and limit the detail as you go.
Building the space partition data structure however takes time, essentially you are trading the "draw each polygon to the screen" for "look at each polygon to figure out if it would be on the screen to avoid drawing it", both require that you look at each polygon. The advantage of the later on is that you can do it off-line and store it on the DVD and thus you don't have to do the costly computation while running the game. Where that whole pretty illusion however breaks down is when it comes to animation: Animation changes the scene, sometimes dramatically and when your scene is just flying around because it has been hit by an explosion, you have to update or rebuild your space partitioning data structure and that is quite expensive. So you are essentially back to square one: A scene with lots of polygons requires lots of computing power. It's only fast when it is static.
There is of course also the issue of memory, the repetition you saw in that video wasn't due to a lack of artist, it was because storing high detail requires high amounts of storage. Thus reusing the same tree a million times is easy, storing a million different trees however will be to much for your computer to handle.
John Carmack talked about similar stuff a while back when discussion possibilities for idTech6.
Another much more basic issue is also the: What would be the benefit? Polygons are rarely a big issue with modern graphics, as they are small enough to not really bother anybody and even the few sharp edges that are still presented could be smoothed out with geometry shader in the coming years. The much bigger problem is really lighting, animation and motion. Getting a static scene look photorealistic isn't that difficulty, getting it to behave realisticly is a whole different problem and millimeter sized dirt helps you little when your physics engine can't even do a barrel that looks like it has proper weight to it.
So in essence, stuff like that is doable, but still quite a bit away from actually being practical. The reason why it might still happen one day: Games consist of a lot of static geometry, most things are not destructible and even those that are, have limits to their destructibility, i.e. a car door may fall of, but it won't deform in dynamic ways, which can be handled reasonable well in space partitioning. There is also nothing stopping you from rendering different parts of a scene with different technologies and then combining the results.
1. that guy's voice is the worst.
2. this showed absolutely nothing that had anything to do with interactivity.
3. the last thing the game industry needs is more emphasis on photorealism. if nothing else, it makes games more expensive to make, and it's not like scanning real-world objects would actually get you very far in terms of creating an interesting game experience.
4. i'm not a computer graphics expert, but i thought this video really exaggerated how dependent graphical fidelity is on polygon count. something something normal maps.
i think it's more important to consider if it's relevant before wondering if it's real.
@Anwar said:Pretty much spot on though I will say it is pretty subjective to state that polygons aren't an issue when there are tons of curved surfaces in games that are not real curves as they are limited by the current polygonal system and processing power. Having to never see false curved surfaces again will be a great thing to experience for all artists and developers. Also not having to worry about one more limitation is always a good thing. These kinds of topics usually have a lot more benefits to developers than consumers anyway because why do consumers care about polygons? This video isn't really to sell it to them.What do you guys think? Is this just smoke and mirrors? Some elaborate ad campaign? Or the next step in visuals for video games?It's a bit of everything. The core technology behind this is likely real or at least there really isn't much reason why it shouldn't be real. The teleshopping voice over however is full of bullshit. However you might see comparable levels of detail a few years down the line.
The situation is basically this, our hardware is fast enough to render a polygon per pixel, so it could in theory with the right data fill the screen with "unlimited" detail (you can't cram more into a pixel then a pixel). The problems preventing that are however that you first need a scene that is that detailed and secondly you need a way to break the scene down to fewer polygons to make it possible to render (i.e. you want to render tiny tree with a million polygons in the distance as a single polygon or pixel, not as a million). The first problem can be easily solved by throwing a lot of artist at the problem or using copy&paste. For the second problem there exist plenty of space partitioning algorithms to store the scene in such a way that you can quickly traverse it and limit the detail as you go.
Building the space partition data structure however takes time, essentially you are trading the "draw each polygon to the screen" for "look at each polygon to figure out if it would be on the screen to avoid drawing it", both require that you look at each polygon. The advantage of the later on is that you can do it off-line and store it on the DVD and thus you don't have to do the costly computation while running the game. Where that whole pretty illusion however breaks down is when it comes to animation: Animation changes the scene, sometimes dramatically and when your scene is just flying around because it has been hit by an explosion, you have to update or rebuild your space partitioning data structure and that is quite expensive. So you are essentially back to square one: A scene with lots of polygons requires lots of computing power. It's only fast when it is static.
There is of course also the issue of memory, the repetition you saw in that video wasn't due to a lack of artist, it was because storing high detail requires high amounts of storage. Thus reusing the same tree a million times is easy, storing a million different trees however will be to much for your computer to handle.
John Carmack talked about similar stuff a while back when discussion possibilities for idTech6. So in essence, stuff like that is doable, but still quite a bit away from actually being practical. Another much more basic issue is also the: What would be the benefit? Polygons are rarely a big issue with modern graphics, as they are small enough to not really bother anybody and even the few sharp edges that are still presented could be smoothed out with geometry shader in the coming years. The much bigger problem is really lighting, animation and motion. Getting a static scene look photorealistic isn't that difficulty, getting it to behave realisticly is a whole different problem and millimeter sized dirt helps you little when your physics engine can't even do a barrel that looks like it has proper weight to it.
Also it is usually more the programmers fault for failing to make the physics engine more accurate and at least beleivable, not so much an inherent problem with hardware. Doing more and more physics-based things simultaneously meets more hardware limitations than programmer faults, but singular events that are incorrect are almost always the fault of the developer who made the engine in the first place.
Yes, curved surfaces or the lack there off is a problem, but it's a problem independent of "infinite detail" and can be solved with tesselation and geometry shader provided by modern GPUs:Pretty much spot on though I will say it is pretty subjective to state that polygons aren't an issue when there are tons of curved surfaces in games that are not real curves as they are limited by the current polygonal system and processing power.
The phantom meter on this company is off the charts. Sure, it might be real. But my monies is on this being a elaborate scam to rip of the Australian government out of their Aussie dollars. Every year they'll need to refile the government grants and they'll release a work in progress video to make it seem like they're doing their shit. But they're really in the outbacks having a barby and double fisting Fosters.
Oh wow, it's good for the games industry. I just wonder how the movies industry or the televisions industry will approach. If somehow we could utilize this for the musics and photographies industries...
@mosdl said:
He says unlimited, which is just hype. Every system is limited by memory...
In terms of traditional approaches, yes, that's true. Iterative approaches go beyond memory limitations because the machine is constantly deleting and recreating information dynamically.
But isn't that what he was railing against with textures popping in? Granted, I don't know a lot about computers, but those two things sound fairly similar. If you mean "deleting and creating game data on the fly", then I honestly can't imagine any game doing that, since how the hell would that even work? Again, not computer savvy.@mosdl said:
He says unlimited, which is just hype. Every system is limited by memory...
In terms of traditional approaches, yes, that's true. Iterative approaches go beyond memory limitations because the machine is constantly deleting and recreating information dynamically.
@Video_Game_King said:
@SeriouslyNow said:But isn't that what he was railing against with textures popping in? Granted, I don't know a lot about computers, but those two things sound fairly similar. If you mean "deleting and creating game data on the fly", then I honestly can't imagine any game doing that, since how the hell would that even work? Again, not computer savvy.@mosdl said:
He says unlimited, which is just hype. Every system is limited by memory...
In terms of traditional approaches, yes, that's true. Iterative approaches go beyond memory limitations because the machine is constantly deleting and recreating information dynamically.
There are already games which use iterative encoding. Spore is the most recognisable but there are others too like kkrieger by demoscene gurus .theprodukkt. a few years ago. kkrieger is a 96Kilobyte file (something like 95% smaller than the average MP3) which iteratively generates a multi level 3D fps with modern graphics (bump mapping, specularity and many other methods of per pixel lighting), sound and physics effects. When It was released in 06 it rivalled some full retail games in terms of engine capabilities and detail and even now it's still impressive in many ways beyond its iterative aspect.
This seems similar to how flame fractals are calculated or even more so mandelbulb fractals. I'm not sure that this approach is really useful in games. So much stuff has to be figured out after the models are loaded in. If you have to recalculate model density that often, stuff like lighting would have to be totally recalculated much more frequently. This seems interesting for still rendering where that kind of detail is always appreciated.
I dunno, it'd be cool if they could make it work.
@Pinworm45 said:Clay maquette. They used this method on LOTR for the giant elephant things.@Anwar said:Where are we going to find a dragon to scan in?Modeling and texturing is also extremely tedius. Scanning real would objects is likely way faster and produces higher quality, at least in theory.they talk about scanning in real world objects, which is an extremely tedious thing to do and I don't see how that will help game designers.
Though arguably that reduces the need to use the scanner if a modeller could produce it in the same time in Zbrush/Max
Please Log In to post.
Log in to comment