The future of graphics in games - Page 5
Forum Index > General Forum |
MrCon
France29748 Posts
| ||
Danjoh
Sweden405 Posts
On August 02 2011 02:24 DeltaSigmaL wrote: Many people seem to doubt the demonstration based on the supposed memory requirements the system would need. Is there a demonstration using polygons that can acheive detail like this? Even if this does take up absurd amounts of memory, it's still cool that you can render those scenes so quickly. A presentation could probably be done in microstation/autocad quite easely with point cloud that they're using. Since most of the things they showed were exactly thesame thing, they could've just attached a whole bunch of references. Never made a presentation movie in MicroStation myself, but I saw one made over a road... I think it was 2 minutes long, highly detailed and very fluent in I think it was 20fps also. But they said that after they had the whole thing referenced in, and had layed out the camera path. It took the computer 18 hours or something like that to create it. | ||
Deleted User 101379
4849 Posts
On August 02 2011 02:34 MrCon wrote: Isn't this "atom" thing was used in outcast ? They used no polygon and all the rendering was done by the CPU. It was Voxel Space Technology, basically the same thing. The first game using it was released in 1992 (so several years older than SC). It was later made obsolete because Polygon graphics were a lot more Memory and CPU efficient than Voxels and graphic cards only supported polygons. | ||
Bibdy
United States3481 Posts
On August 02 2011 02:34 MrCon wrote: Isn't this "atom" thing was used in outcast ? They used no polygon and all the rendering was done by the CPU. Yep. Voxels. Used in Delta Force, Command and Conquer: Tiberian Sun and many others back in the 90s. I think I'm with this guy and the quote he dug up: On August 02 2011 02:00 arbitrageur wrote: A comment on the video: "The word you're looking for is Voxels, and its a technology that's as old as polygons. You've got nothing new here. You're passing off basic fundamental well-known 3d rendering techniques as something new and fascinating by making up your own words for decades old technology and fluffing it up with... fluffy meaningless phrases like "UNLIMITED POWAR!!!". Oh well, have fun fleecing your investors then disappearing with the money as all modern snake oil salesman do. NinjaSeg 2 months ago 20 " It just looks like voxels. Smaller voxels, thanks to more powerful processors, but voxels nevertheless. The limitations of which have already been explored and thus the industry more readily adopted polygons as the primary rendering technique. Hell, that's all GPUs do. Render polygons by doing a metric fuckton of vector and matrix math in real-time, so essentially the whole thing HAS to be rendered by the CPU, because the GPU isn't built for it. I don't find it a coincidence that the video is rendered at a piss-poor level of detail in order to hide it's shortcomings. Come up with 'IMPOSSIBLE DETAIL' and only render in 480p? Really? The whole thing seems like a means to grab investor money and then bail. | ||
rasnj
United States1959 Posts
On August 02 2011 02:32 RageOverdose wrote: Like they said, they aren't artists. A lot of what you see today is just artistic, from the high contrast to the exaggerated blurs. They don't seem to have very interesting lighting or specular mapping so it looks boring and bland, but the detail is a lot higher otherwise. Also, he's throwing around unlimited and unlimited is, of course, technically impossible because even with compression and decompression and streaming, you still need to have space for it all. In all likelihood, just like all graphics technology, you'll be able to utilize this technology in bits and pieces where it seems fit and not just render a whole world with it and every little granulation of it you can think of. I would expect maybe world objects, like trees or leaves, to be utilizing point-cloud atoms while the character models still use normal polygons (if it's an FPS, probably the gun would also use atoms). The dirt would probably still be a textured surface of polygons done in a more regular fashion, with maybe certain objects (still objects) utilizing atoms. Maybe someday it can be utilized for everything. Maybe. Still, I want to see a better tech demo than this. I'm skeptical, but the cries of "impossible" are as exaggerated as this promo video. They seem to have no real claims beyond their UNLIMITED power which is obviously bogus. The only "technical" details do not explain how this is different from any voxel rendering system (perhaps with a sparse representation of graphics). The video has very little actual content. What it has is either clearly incorrect, meaningless, or not impressive. The video itself could obviously easily be created with existing polygonal technology and may even be prerendered. We have no evidence that this video is or is not genuine, but even if genuine it does not seem impressive from a classical standpoint. Just like with people claiming to have constructed perpetual motion machines, techniques for turning water into wine, and creating gold from common metals these claims are hard to refute because we are given very few details. From what little is given it seems unlikely to work anything like advertised. Until a proper technical explanation is provided there is no reason to assume this is anything spectacular. | ||
Hobosnipe
Canada13 Posts
| ||
spinesheath
Germany8679 Posts
The real issue I have with this is: Data transfer rate. EVERY big render they showed was full of clones. It makes loading and storing that data trivial. Load data for one object, render it 10000 times. But what happens when you want to have graphics where everything is unique? Well, at least WAY less repetitive than what they showed? There is no way you can stream data THIS quickly with current (affordable) technology. You can't pre-load it either because the sheer amount of data will be too much. | ||
Disquiet
Australia628 Posts
| ||
sVnteen
Germany2238 Posts
sick | ||
maartendq
Belgium3115 Posts
On August 02 2011 02:24 GizmoPT wrote: i think polygons work just great tbh lol I've said it a lot before and I'll repeat it again: UE3 is one of the most powerful, beautiful and long lasting engines out there. If I didn't know it was UE3, I'd say it was some kind of top-notch hollywood CGI.. The first time I played Bulletstorm I had to collect my jaw from the floor.. | ||
Desti
Germany138 Posts
| ||
leecH
Germany385 Posts
On August 02 2011 00:44 Liveon wrote: What I don't understand though, is how they can 'import' rocks/cacti from reality. Do they just scan and figure out where the atoms are to recreate it in their simulation? It looked really impressive though. official site there are 3d scanners for a long long time. for example police/military use them to scan certain areas they dont want to go in before making sure its safe. for example a crashed building. also its used to scan cars/objects. there are companies who scan in cars for a PRE-MODEL. you only get the dimension. you have to rebuild it by hand to get a clean 3d-model. a scanned 3D model is nothing you can directly import in a game. NO WAY. its a piece of polygon bullshit. its only for measuring. figuring atoms? no no no.. this is science fiction. were not in geneva letting fucking atoms and shit collide just for the fun of it. this video is full of bullshit and has nothing to do with a breakthrough. and by the looks of the video and by the homepage this is less than professional and purely a joke. literally. a joke. sorry for breaking the news.. | ||
LegendaryZ
United States1583 Posts
On August 02 2011 02:14 Keitzer wrote: so if the only thing they can do with this is render epic-looking cinematics, then sweet, now we have better cinematics.... why do you assume nothing? are you the terminator back from the future to make sure it doesn't happen? Cinematics are pre-rendered anyway (which is why they always look so much better than actual in-game graphics) so this technology would be of no consequence on that end. You can already render absurdly good cinematics using current techniques and it's really only limited by the actual budget you have available. The only way this "new" (questionable) technology would matter is if it was somehow applicable to actual gameplay where we're currently bottle-necked by the capabilities of our computer components (CPU, graphics cards, RAM, etc.). Either that, or if this somehow drastically reduced the cost of creating content, which really doesn't seem to be the case so far as I can tell. Obviously if this represents a leap forward in freeing up resources on any front, it's a welcome development, but I don't think you can fault the skeptics given the limited amount of actual information in that video. | ||
THE_oldy
Australia97 Posts
A computer wont be required to consistently keep the location of every atom in memory, or some other brute force way of doing things. I don't know how they would go about it, but that's just the point; They're developing a new method. If they're method was obvious someone would have done it already. So to all skeptics of this, i doubt its going to be anything like a method that might jump into your head when your saw the video, like using terrabites of raw data of something. | ||
superjoppe
Sweden3682 Posts
| ||
semantics
10040 Posts
On August 02 2011 00:54 WniO wrote: the problem with this is they cant render it in real time, or animate for that matter. ^^^ animate how does object interaction works for them seems to have been glazed over. Anyways we increase polygon count dynamically now though tessellation which is bound to be much more memory efficient, in-fact most rendering is done though streaming increasing detail as you get closer to something because it saves on memory and cost to run the game. | ||
leecH
Germany385 Posts
On August 02 2011 03:02 THE_oldy wrote: I think they're doing something different than standard polygon tech. Its not just going to be something as simple as just lots and lots of polygons, only smaller. A computer wont be required to consistently keep the location of every atom in memory, or some other brute force way of doing things. I don't know how they would go about it, but that's just the point; They're developing a new method. If they're method was obvious someone would have done it already. So to all skeptics of this, i doubt its going to be anything like a method that might jump into your head when your saw the video, like using terrabites of raw data of something. you dont know what a polygon is. you have no clue what your talking about. i play the buzzkiller in this thread. im a dick. youre welcome. | ||
Cold-Blood
United States200 Posts
Giving the reason why they have been silent for so long. | ||
Slaytilost
Netherlands968 Posts
On August 02 2011 02:08 exog wrote: It appears to be rendered in real time by the way the camera moves. They also answered that you dont need a supercomputer to run it... There, thrashed all the arguments in the thread. Will be interesting to see how it goes. Because the camera moves it is in real time? Right, movies like Toy Story and Independance Day are just stills, because they sure as hell arent real time! Santacause is real, there, just trashed all the arguments made ever. | ||
Southlight
United States11761 Posts
| ||
| ||