So here's my 2 cents as a guy who makes games for a living:
All I see are several high voxel instanciated objects repeated over and over. Sure it's cool that grain of dirt is so detailed, or that tree is so organic but there's a reason the demo's they show always copy/paste the same objects over and over. It's not about the time to make them, it's that the engine works with very limited resources and instances them (they get loaded into memmory and drawn all at once rather then many draw calls).
The detail in our virtual worlds comes from variety as well as detail. Show me this engine with 32 unique tree's and rocks arranged to create a realistic look and I'll be impressed. But not the same 4 objects copied around a cube of land and then copy/pasted and rotated to look diverse.
You can already render 1000's of the same tree in Unity and Unreal through instancing with little effort or strain on the video card. But who the hell wants the same tree 1000 times to make up the forest.
Plus, how do you animate voxels? There is no animation system made that is designed to allow for real-time 3d space updates for voxel animation.
In order to demonstrate that the engine is real, they needed to let someone control it in real time. So they did.
There are still some major hurdles, however, so I'm still skeptical that we'll see it in typical games.
Interesting interview. He is absolutely being deceptive. Carmac wasn't claiming that their demo couldn't run on current hardware. He was claiming that a game based on their technology is years away due to hardware limitations. Their next demo needs 1000x more unique objects, fluid animations, collision detection, ...
Look at games like Skyrim, RAGE or Bioshock infinite.
Games using most ly old tech that look absolutlely amazing.
Why bother putting in years of work to make a game look 2% better than it does currently?
I would rather have people optimize todays games for the hardware we are running now.
It would be great if every game took advantage of 6-8 core CPUs and 2+ gig dual graphics card setups.
The fact that 99% of all games released today are designed to run on hardware 1/10th the power (mostly due to consoles) of todays high end hardware... proves that we should be optimizing to take full advantage of what we have now.
Look at games like Skyrim, RAGE or Bioshock infinite.
Games using most ly old tech that look absolutlely amazing.
Why bother putting in years of work to make a game look 2% better than it does currently?
I would rather have people optimize todays games for the hardware we are running now.
It would be great if every game took advantage of 6-8 core CPUs and 2+ gig dual graphics card setups.
The fact that 99% of all games released today are designed to run on hardware 1/10th the power (mostly due to consoles) of todays high end hardware... proves that we should be optimizing to take full advantage of what we have now.
given that DX10 and 11 both allow much better graphic experiences, it does seem as though consoles and their dx9 limitations (among other hardware limitations) are holding back gaming to a degree.. the problem is that the console market, for single player games at least, is seen as the most lucrative, that may be the opposite of online gaming.. but.. there is still the shift to try and make more online games 'console friendly' which inevitably means that those limitations are going to be influencing game design for a while yet.. as it will most likely be another 4 years before we see an upgrade in the consoles, im not sure when, or even if... we'll see any games out that will be making use of the hardware advantages etc that PC's have over consoles, until they find some way of making consoles upgradeable.. i think this is probably going to be a continuing problem.
In order to demonstrate that the engine is real, they needed to let someone control it in real time. So they did.
There are still some major hurdles, however, so I'm still skeptical that we'll see it in typical games.
They showed it off running on a quad core Sandy Bridge system at around 20 frames per second and a resolution of 1024x768. Graphics tend to be trivially parallelizable, so let's assume that it scales perfectly to as many processor cores as you've got. We're still a long way away from processors being able to deliver 60 frames per second at more typical gaming resolutions.
Now, one could hope for optimizations to make it run better. To get 20 frames per second at 1024x768 is on the order of 16 million pixels per second. Picking out the right "atom" for one pixel would then have to be done in several hundred CPU clock cycles of a single core.
If there are billions of atoms to choose from, then how does one pick the right one? To do it in several hundred clock cycles even ignoring latency issues (the processor could be juggling huge numbers of threads to partially get around this) strikes me as impressive to begin with. How far can they improve on that? It's not necessarily possible to squeeze much more out of it.
But the bigger issue is memory requirements. They had a relative handful of objects repeated a huge number of times. If you do that, then you can have each object in memory just once, and then have everything point to translations of it. That works if you have a handful of objects.
But what happens if you want many thousands of distinct objects, as modern games do? Suddenly you need a hundred times as much memory as before. And that's not easy to get.
Now, it would be possible to work around this by having games only use a handful of objects. Or perhaps rather, have many different objects, but only a handful available at a time. You could have one set of objects on one area, then a zoning screen, then a different set of objects in a different area.
Early computer games had to do this to a considerable degree to make the game take little enough space. For example, Super Mario Bros. was only 32 KB for the entire game. To squeeze things down, you make a few blocks and tile them everywhere. Some things that look a priori like different objects are really just the same thing with a different color scheme. Some objects are symmetric and tiled, so instead of storing an entire rock, you only store 1/4 of it and rotate or reflect that 1/4 to get the whole rock. And then you use that one rock in thousands of places, and every time the game needs a rock in the terrain, it's the same one.
Now, you can do that and make a lot of things work. That's exactly what they did for their demo. But would people really accept that in modern games? Really?
The next problem is animation. If you have one object, then you can probably animate it. If you have thousands of copies of one object, then you can animate them all in the same way at the same time. That would use the previous trick of having one copy in memory and then a zillion translations of it.
But that falls apart if you want to animate them differently. For terrain, you don't need to do that. But what about mobs? If you have ten mobs on the screen at once, even if they're ten copies of the same mob, they're probably not all moving in exactly the same way at exactly the same time. One attacks before another. One is running while another stands still. They'd have to be separate objects in memory, and then the memory requirements explode.
And what about players? Have a hundred players on the screen, with some way off in the distance, and you have a hundred separate large objects in memory. That can run you out of memory very quickly at the levels of detail they're using.
Now, you can get around this to some degree by saying, we're not going to draw players off in the distance at all. But they criticize other game engines for popping from one model to another. That would be popping from one model to invisible, and not even to another model. That's not good.
There's also the issue of how to move things onto the video card. Video cards come with their own dedicated video memory, and you can't easily add more the way you can with system memory. You can easily equip a system with 16 GB of system memory. Indeed, that only costs around $100 for the system memory. You can't do that with a video card, though.
The most memory you can get in a video card is 6 GB, and that costs over $3000 for the card. The most you can get in a consumer graphics card is 3 GB (2 GB for each GPU in a Radeon HD 6990 doesn't count as 4 GB), and that costs $600 for the card. You can stick low end cards with huge amounts of DDR3 memory, but then you'd have to buy a different card for this game engine from what you use for "normal" game engines. That's not going to go anywhere for about the same reasons that GPU PhysX never amounted to much.
-----
But that's not to say that their system can't be useful at all. If it works great for background terrain that repeats a lot, but completely falls apart when you need a bunch of mobs animated independently, then you could use something along the lines of their system for the background, and then more traditional rasterization for everything that moves, and perhaps for objects where polygons look good. (E.g., if an object is supposed to be a cube, then rendering as an actual cube using rasterization is already the best possible, so there's no need to do anything fancy.)
Actually doing that in a game engine might end up being cost++; but it would at least be technically feasible. This is in the sense that a permanent human colony on Mars is technically feasible in ways that humans traveling at the speed of light simply isn't. Technically feasible doesn't necessarily mean easy or practical.
Is that where some games will head in the next several years? It's plausible. But I'd bet against it.
Please stop...
You talk alot & never have anything real to say. U seem like a well read person & mostly up-to-date on your knowledge, but you also have a very miopic sense of software & what hardware does. Not just in this post, but namely all of your posts. You are too caught up in the latst benchtests, than understanding the meaning of that bench.
This "type" of technology has been around a long time.. yet this really isn't voxels.
This is something completly new & different. A software solution that renders the SCREEN GRAPHICS, (whats shown to the end-user), while all the actual geography is handled in memeory (software). Using their propreitary alogrithm.
Video cards are predicated on triangles, not arbitrary points.
"No they are not charity. That is where the whales come in. (I play for free. Whales pays.) Devs get a business. That is how it works."
You could have the best engine in the world available but unfortunately the latest trend of most MMOs is to dumb down the graphics. The excuse is always that they want more players to run it on their machines which I never really understood since a game can be designed to turn a lot of the graphics options off so that you can play. The other excuse card played is lag. The hard truth is that you need a good machine and a decent graphics card or cards if you want to see the pretty eye candy. I'd like to see MMOs give options to dumb down the graphics with game settings rather than sell dumbed down graphic games. MMOs seem to be going backwards these days insted of forward. Even SWTOR has dumbed down graphics and it is a major IP but hopefully story and content make up for it. Skyrim has nice graphics but not a online MMO. If something like a online Fallout MMO could be made with the kind of graphics in Skyrim then it would be awesome. I'd also like to see hybrids such as MechWarrior which could be both a online MMO and a tactical shooter. Right now I have to wait till someone breaks out of the box which likely will be 5 to 10 years down the line if ever.
The video looks pretty damn incredible... but how is this different from voxel tech?
It's different because it isn't really Voxels as we know them from the past. This video explains the technology a lot better than the OP's. It's basically a very advanced search algorithm that only renders the points needed for any given instance. So instead of rendering billions of points of data it only renders the points you can actually see at any given moment. It's kinda like a cross between Google, Voxels and the tech On-Live uses to stream live gaming.
Edit: So for example if your screen resolution is set at 1366x768 it is only processing that many points at any given moment. One point per pixel.
Bren
All I can focus on is the game with the female elf that looks like Zelda on xbox. Anyone know what game that is?
Edit: I found it. It called Kameo.
-- "Any free people have the right to choose how it wants to be govern thats the essence of democracy. It's sad when America has chosen for the stability and consistency of a dictatorship and doing it democratically" -utnow
Voxels are volumetric objects, even empty space is a voxel volume technically and must be accounted for.
What if you don't use voxels though? A fractal compression algorithm can approximate any object or texture and can produce shapes that are complex to the eye but easy to store and resolve to different scales of detail. They've been used for terrain and trees, but is there anyone out there that incorporates different layers of fractal rendering? You can simulate any texture, any level of detail you want, it would just becomes a question of how efficiently you do the calculations to zoom in and around multiple features simultaneously.
(I bring this up because I notice that the demo included a Sierpinski pyramid and a few other objects common in fractal toolkits)
Originally posted by maplestone Originally posted by Deivos
Voxels are volumetric objects, even empty space is a voxel volume technically and must be accounted for.
What if you don't use voxels though? A fractal compression algorithm can approximate any object or texture and can produce shapes that are complex to the eye but easy to store and resolve to different scales of detail. They've been used for terrain and trees, but is there anyone out there that incorporates different layers of fractal rendering? You can simulate any texture, any level of detail you want, it would just becomes a question of how efficiently you do the calculations to zoom in and around multiple features simultaneously. (I bring this up because I notice that the demo included a Sierpinski pyramid and a few other objects common in fractal toolkits)
There's a difference between using fractals to generate trees and using fractals to make an avatar or a building.
I can not remember winning or losing a single debate on the internet.
Hey Desert - I know you only have almost 2000 posts, but let me give you some advice - when posting a link, first click the little globe with the "chain link" symbol on it, then paste your link in the text box
It would be great if every game took advantage of 6-8 core CPUs and 2+ gig dual graphics card setups.
One of the features of DirectX 11 is to make it easier for games to scale well to many cores. Implementing that feature is optional, of course, just like a DirectX 11 game can use tessellation or order-independent transparency but does not have to. So far, Civilization V is the only game that I'm aware of that does implement it, and the company says that the game rendering engine will scale well to 12 cores.
Civilization V is also quite an outlier in benchmarks. Apparently in order to to make the game scale to so many cores properly, something has to be done in video drivers to accommodate it. When the game launched, neither AMD nor Nvidia had done so, and they both scrambled to get it working properly. Nvidia had it more or less working before AMD, but benchmarks on that game even on fixed hardware can vary wildly by which drivers you're using.
Fuck more graphic enchantments? I want games to be long not end in 2 days... gameplay before graphics idc if graphic wores want to fully use their beast pc (Whch I have one) I don't feel like buying any game now is worth it for their length excluding most rpgs and some online games.
Even if the haters are right about the memory problems, it's only a matter of time before this or a similar tech takes over. It's the future of gaming imo.
You talk alot & never have anything real to say. U seem like a well read person & mostly up-to-date on your knowledge, but you also have a very miopic sense of software & what hardware does. Not just in this post, but namely all of your posts. You are too caught up in the latst benchtests, than understanding the meaning of that bench.
This "type" of technology has been around a long time.. yet this really isn't voxels.
This is something completly new & different. A software solution that renders the SCREEN GRAPHICS, (whats shown to the end-user), while all the actual geography is handled in memeory (software). Using their propreitary alogrithm.
Video cards are predicated on triangles, not arbitrary points.
First, lay off the ad hominem.
They say it's a search engine to pick the right "atom". Well that's what sparse voxel octree is: a search algorithm to pick the right voxel for each pixel on the screen.
There are some serious disadvantages to voxels, and Euclideon hasn't demonstrated that they can overcome any of them. They say that what they're doing isn't voxels, but their description makes it sound as if it is.
If they're storing atoms at arbitrary positions in space, then rotating that by arbitrary angles is completely trivial to do. It's simple matrix multiplication, is very, very fast to do by computer, and undergraduate students in even an introductory linear algebra course may have seen it. Wanting to throw something together in three weeks can explain why they put so few distinct objects into the demo, but it's not a meaningful impediment to showing rotations.
Voxels, on the other hand, really can't rotate objects very well except at right angles. It's a limitation of the technology. If they wanted to say, see, this isn't voxels, they could have shown the same objects, but each object rotated in dozens of different directions. Indeed, that's the natural thing to do if you want to demonstrate that what you have isn't voxels. And they didn't do it.
You could do rasterization in software, too. It typically isn't done that way because it's slow. Intel has been known to offload some traditional graphical functions onto the processor, because its integrated graphics are so pitiful. Saying that it's done in software and in memory could describe basically anything.
You're right that rasterization, which is what current video cards are optimized for, is based on polygons. But it didn't used to be that way. In the early days of 3D, it was far from clear that rasterization was the way to go, and other approaches were tried in hardware. Rasterization won out because for real games, it worked by far the best.
-----
There are pros and cons of different approaches to graphics. The cons can be intrinsic to the approach, and not something easy to work around. For example, rasterization can't do complex reflections. Well, it kind of can, but it's an enormous performance hit to the degree that it completely cripples everything. If you want to reflect something off of one flat surface (say, a large body of water), you can make it work. But if you want a bunch of reflective surfaces, or to reflect something off of a curved surface, you're out of luck. You can throw on a texture map that looks like it plausibly could be the reflection of something, but isn't anything like what the "real" reflection would show.
Rasterization has been used for quite a few years now, and no one has found a way around this. It's just an drawback to the graphical approach. Reflections are completely trivial for ray tracing, however.
-----
"Even if the haters are right about the memory problems, it's only a matter of time before this or a similar tech takes over. It's the future of gaming imo."
What makes you so certain that it will be something along the lines of what Euclideon is doing? Why not ray tracing? Why not quadratic surfaces?
Euclideon likes to talk about "unlimited detail". Well yes, you can get unlimited detail by any graphical method, if only you have unlimited processing power, unlimited memory capacity, and unlimited memory bandwidth. Put limits on any of those and you no longer get unlimited detail by any method. The question is what you can make work the best with the hardware limits that you have.
I wont dismiss this idea.check pci the other was dismissed what we use today?pcie.so the idea might need a bit of work.but check how much effort was put in to what we use today.saying there is no chance is like saying pci would never work a genious just had to lay the base for mass adoption to kick in.arm could very well adopt this. Hope they dont cause not even the mighty ati ggrapic god will be able to counter arm(even intel would face dire conseqence. If we check in the past arm was also dismissed by intel and ms.guess what arm is the god of mobility and intel is sitting in the dust left by arm on the mobility front.so just dont dismiss this idea yet.it might not be the full scope but it look like this team is in the process of getting a very mighty help.
By the way the base of this idea isnt new check java off screen rendering accelerate get a wooping 600 fps(when ms intentional)bug isnt around.ms changed thing not long ago so java can now go to 148fps instead of 37 fps. If ms corrected the issue with java oracle sun we all would get 600 fps.cheer at allrunescape player.no more msbug induced lag of java.but corp the way ms are rhey say bha why correct that nobody use it. Yes a lot of people use it and ms knows it.so this graphic thing doesnt promise unlimited detail on everything just on everything viewable.so nothing prevent them fromusing all the trick industry has lol.
Awesome stuff. I've been looking forward to this for a while. I'm not sure why people believe that this is fake? It may be a long while out but there is no reason to disbelieve it. People are just afraid of change.
*he thinks LoD stands for Level of Distance, but it's Level of Detail. Maybe he's not the engineer, but who makes that kind of mistake.
*he plays it on a tiny screen, which means you can't see the 'noise' voxels and raycasting engines have
*he won't show animation, his excuse is that he will only show it when it's finished, uhm...you can't even show an object moving in your scene? strange...
*he cleverly avoided the memory issue, which is there, he says "memory is not an issue"...really...you have incredibly detailed models and memory is not an issue, how does that work exactly?
*why is this an only outdoor scene with a lot of diffuse lighting, is it because it would otherwise show the noise his voxel lighting has? Is there no animation because there is no lighting at all? Is his lighting just mapped or is it actually there? how can we tell without animation?
*he says he's not using raytracing, and he can't be using polygon lighting, so what kind of lighting is he using?
He's full of it. Voxel engines are great, and in the future engines will be a kind of voxel / polygon hybrid system, but he is trying to sell snake oil.
I'm not sure why people believe that this is fake?
Because of the high number of weasel words in the voiceover and the fact that people are wildly speculating rather than playing with it hands-on.
Not just that, it's the amount of data in a cloud that's so hard for me to believe. People are looking for silver bullets and this is doing a sales job.
IMO, they are looking at getting investor dollars.
Kyleran: "Now there's the real trick, learning to accept and enjoy a game for what
it offers rather than pass on what might be a great playing experience
because it lacks a few features you prefer."
John Henry Newman: "A man would do nothing if he waited until he could do it so well that no one could find fault."
FreddyNoNose: "A good game needs no defense; a bad game has no defense." "Easily digested content is just as easily forgotten."
LacedOpium: "So the question that begs to be asked is, if you are not interested in
the game mechanics that define the MMORPG genre, then why are you
playing an MMORPG?"
I'm not sure why people believe that this is fake?
Because of the high number of weasel words in the voiceover and the fact that people are wildly speculating rather than playing with it hands-on.
Not just that, it's the amount of data in a cloud that's so hard for me to believe. People are looking for silver bullets and this is doing a sales job.
IMO, they are looking at getting investor dollars.
You mean everybody is responding to this just like they did to the offset engine or like they did to the first videos of half-life 2? Oh my! You're right; I should go make sure everybody knows the truth! Voxels are a lie!
Excuse me for being so sarcastic but I think I'll do the sensible thing and not respond to hysteria. It remains to be seen whether or not it is a lie but the technology has been around for longer than most of us here have been computer geeks.
No offense intended to you of course Waynejr2. I realize you were just responding to my comment.
Originally posted by waynejr2 IMO, they are looking at getting investor dollars.
I wouldn't be so quick to write it off as that. Getting investment in Australia for a small business is *extremely* hard if you don't have a prototype to show off in-front of investors. This is purely because you can only receive investment from Professional and Sophisticated Investors who do their research before signing any cheques.
To quality as a Professional Investor in Australia, you need $10m in assets.
Sophisticated? $2.5m in assets and a net income of $250k per year for the last two years.
Anyone who does not meet those requirements cannot legally invest in a small business. They either don't know about this, are quite serious in what they're doing or can use the force to mind-trick rich folk into throwing their money into something with a lot of controversy on the net.
Comments
So here's my 2 cents as a guy who makes games for a living:
All I see are several high voxel instanciated objects repeated over and over. Sure it's cool that grain of dirt is so detailed, or that tree is so organic but there's a reason the demo's they show always copy/paste the same objects over and over. It's not about the time to make them, it's that the engine works with very limited resources and instances them (they get loaded into memmory and drawn all at once rather then many draw calls).
The detail in our virtual worlds comes from variety as well as detail. Show me this engine with 32 unique tree's and rocks arranged to create a realistic look and I'll be impressed. But not the same 4 objects copied around a cube of land and then copy/pasted and rotated to look diverse.
You can already render 1000's of the same tree in Unity and Unreal through instancing with little effort or strain on the video card. But who the hell wants the same tree 1000 times to make up the forest.
Plus, how do you animate voxels? There is no animation system made that is designed to allow for real-time 3d space updates for voxel animation.
Arioc Murkwood
Environment Artist
Sad but true.
Interesting interview. He is absolutely being deceptive. Carmac wasn't claiming that their demo couldn't run on current hardware. He was claiming that a game based on their technology is years away due to hardware limitations. Their next demo needs 1000x more unique objects, fluid animations, collision detection, ...
Honestly I think todays tech is great.
Games already almost look freaking real.
Look at games like Skyrim, RAGE or Bioshock infinite.
Games using most ly old tech that look absolutlely amazing.
Why bother putting in years of work to make a game look 2% better than it does currently?
I would rather have people optimize todays games for the hardware we are running now.
It would be great if every game took advantage of 6-8 core CPUs and 2+ gig dual graphics card setups.
The fact that 99% of all games released today are designed to run on hardware 1/10th the power (mostly due to consoles) of todays high end hardware... proves that we should be optimizing to take full advantage of what we have now.
Playing: Nothing
Looking forward to: Nothing
given that DX10 and 11 both allow much better graphic experiences, it does seem as though consoles and their dx9 limitations (among other hardware limitations) are holding back gaming to a degree.. the problem is that the console market, for single player games at least, is seen as the most lucrative, that may be the opposite of online gaming.. but.. there is still the shift to try and make more online games 'console friendly' which inevitably means that those limitations are going to be influencing game design for a while yet.. as it will most likely be another 4 years before we see an upgrade in the consoles, im not sure when, or even if... we'll see any games out that will be making use of the hardware advantages etc that PC's have over consoles, until they find some way of making consoles upgradeable.. i think this is probably going to be a continuing problem.
Please stop...
You talk alot & never have anything real to say. U seem like a well read person & mostly up-to-date on your knowledge, but you also have a very miopic sense of software & what hardware does. Not just in this post, but namely all of your posts. You are too caught up in the latst benchtests, than understanding the meaning of that bench.
This "type" of technology has been around a long time.. yet this really isn't voxels.
This is something completly new & different. A software solution that renders the SCREEN GRAPHICS, (whats shown to the end-user), while all the actual geography is handled in memeory (software). Using their propreitary alogrithm.
Video cards are predicated on triangles, not arbitrary points.
"No they are not charity. That is where the whales come in. (I play for free. Whales pays.) Devs get a business. That is how it works."
-Nariusseldon
You could have the best engine in the world available but unfortunately the latest trend of most MMOs is to dumb down the graphics. The excuse is always that they want more players to run it on their machines which I never really understood since a game can be designed to turn a lot of the graphics options off so that you can play. The other excuse card played is lag. The hard truth is that you need a good machine and a decent graphics card or cards if you want to see the pretty eye candy. I'd like to see MMOs give options to dumb down the graphics with game settings rather than sell dumbed down graphic games. MMOs seem to be going backwards these days insted of forward. Even SWTOR has dumbed down graphics and it is a major IP but hopefully story and content make up for it. Skyrim has nice graphics but not a online MMO. If something like a online Fallout MMO could be made with the kind of graphics in Skyrim then it would be awesome. I'd also like to see hybrids such as MechWarrior which could be both a online MMO and a tactical shooter. Right now I have to wait till someone breaks out of the box which likely will be 5 to 10 years down the line if ever.
All I can focus on is the game with the female elf that looks like Zelda on xbox. Anyone know what game that is?
Edit: I found it. It called Kameo.
--
"Any free people have the right to choose how it wants to be govern thats the essence of democracy. It's sad when America has chosen for the stability and consistency of a dictatorship and doing it democratically" -utnow
What if you don't use voxels though? A fractal compression algorithm can approximate any object or texture and can produce shapes that are complex to the eye but easy to store and resolve to different scales of detail. They've been used for terrain and trees, but is there anyone out there that incorporates different layers of fractal rendering? You can simulate any texture, any level of detail you want, it would just becomes a question of how efficiently you do the calculations to zoom in and around multiple features simultaneously.
(I bring this up because I notice that the demo included a Sierpinski pyramid and a few other objects common in fractal toolkits)
(I bring this up because I notice that the demo included a Sierpinski pyramid and a few other objects common in fractal toolkits)
There's a difference between using fractals to generate trees and using fractals to make an avatar or a building.
I can not remember winning or losing a single debate on the internet.
Hey Desert - I know you only have almost 2000 posts, but let me give you some advice - when posting a link, first click the little globe with the "chain link" symbol on it, then paste your link in the text box
One of the features of DirectX 11 is to make it easier for games to scale well to many cores. Implementing that feature is optional, of course, just like a DirectX 11 game can use tessellation or order-independent transparency but does not have to. So far, Civilization V is the only game that I'm aware of that does implement it, and the company says that the game rendering engine will scale well to 12 cores.
Civilization V is also quite an outlier in benchmarks. Apparently in order to to make the game scale to so many cores properly, something has to be done in video drivers to accommodate it. When the game launched, neither AMD nor Nvidia had done so, and they both scrambled to get it working properly. Nvidia had it more or less working before AMD, but benchmarks on that game even on fixed hardware can vary wildly by which drivers you're using.
Fuck more graphic enchantments? I want games to be long not end in 2 days... gameplay before graphics idc if graphic wores want to fully use their beast pc (Whch I have one) I don't feel like buying any game now is worth it for their length excluding most rpgs and some online games.
Imagine Euclideon's engine and this one meshed!
Didn't Notch already beat this down enough times? Why bring it here?
Per Notch:
http://notch.tumblr.com/post/8386977075/its-a-scam
http://notch.tumblr.com/post/8423008802/but-notch-its-not-a-scam
He who keeps his cool best wins.
Even if the haters are right about the memory problems, it's only a matter of time before this or a similar tech takes over. It's the future of gaming imo.
First, lay off the ad hominem.
They say it's a search engine to pick the right "atom". Well that's what sparse voxel octree is: a search algorithm to pick the right voxel for each pixel on the screen.
There are some serious disadvantages to voxels, and Euclideon hasn't demonstrated that they can overcome any of them. They say that what they're doing isn't voxels, but their description makes it sound as if it is.
If they're storing atoms at arbitrary positions in space, then rotating that by arbitrary angles is completely trivial to do. It's simple matrix multiplication, is very, very fast to do by computer, and undergraduate students in even an introductory linear algebra course may have seen it. Wanting to throw something together in three weeks can explain why they put so few distinct objects into the demo, but it's not a meaningful impediment to showing rotations.
Voxels, on the other hand, really can't rotate objects very well except at right angles. It's a limitation of the technology. If they wanted to say, see, this isn't voxels, they could have shown the same objects, but each object rotated in dozens of different directions. Indeed, that's the natural thing to do if you want to demonstrate that what you have isn't voxels. And they didn't do it.
You could do rasterization in software, too. It typically isn't done that way because it's slow. Intel has been known to offload some traditional graphical functions onto the processor, because its integrated graphics are so pitiful. Saying that it's done in software and in memory could describe basically anything.
You're right that rasterization, which is what current video cards are optimized for, is based on polygons. But it didn't used to be that way. In the early days of 3D, it was far from clear that rasterization was the way to go, and other approaches were tried in hardware. Rasterization won out because for real games, it worked by far the best.
-----
There are pros and cons of different approaches to graphics. The cons can be intrinsic to the approach, and not something easy to work around. For example, rasterization can't do complex reflections. Well, it kind of can, but it's an enormous performance hit to the degree that it completely cripples everything. If you want to reflect something off of one flat surface (say, a large body of water), you can make it work. But if you want a bunch of reflective surfaces, or to reflect something off of a curved surface, you're out of luck. You can throw on a texture map that looks like it plausibly could be the reflection of something, but isn't anything like what the "real" reflection would show.
Rasterization has been used for quite a few years now, and no one has found a way around this. It's just an drawback to the graphical approach. Reflections are completely trivial for ray tracing, however.
-----
"Even if the haters are right about the memory problems, it's only a matter of time before this or a similar tech takes over. It's the future of gaming imo."
What makes you so certain that it will be something along the lines of what Euclideon is doing? Why not ray tracing? Why not quadratic surfaces?
Euclideon likes to talk about "unlimited detail". Well yes, you can get unlimited detail by any graphical method, if only you have unlimited processing power, unlimited memory capacity, and unlimited memory bandwidth. Put limits on any of those and you no longer get unlimited detail by any method. The question is what you can make work the best with the hardware limits that you have.
You can substitute any word you want in place of "fractals" there and your statement is still pretty true, isn't it?
Awesome stuff. I've been looking forward to this for a while. I'm not sure why people believe that this is fake? It may be a long while out but there is no reason to disbelieve it. People are just afraid of change.
Because of the high number of weasel words in the voiceover and the fact that people are wildly speculating rather than playing with it hands-on.
There are way too many red flags going up for me.
*he thinks LoD stands for Level of Distance, but it's Level of Detail. Maybe he's not the engineer, but who makes that kind of mistake.
*he plays it on a tiny screen, which means you can't see the 'noise' voxels and raycasting engines have
*he won't show animation, his excuse is that he will only show it when it's finished, uhm...you can't even show an object moving in your scene? strange...
*he cleverly avoided the memory issue, which is there, he says "memory is not an issue"...really...you have incredibly detailed models and memory is not an issue, how does that work exactly?
*why is this an only outdoor scene with a lot of diffuse lighting, is it because it would otherwise show the noise his voxel lighting has? Is there no animation because there is no lighting at all? Is his lighting just mapped or is it actually there? how can we tell without animation?
*he says he's not using raytracing, and he can't be using polygon lighting, so what kind of lighting is he using?
He's full of it. Voxel engines are great, and in the future engines will be a kind of voxel / polygon hybrid system, but he is trying to sell snake oil.
Not just that, it's the amount of data in a cloud that's so hard for me to believe. People are looking for silver bullets and this is doing a sales job.
IMO, they are looking at getting investor dollars.
Epic Music: https://www.youtube.com/watch?v=vAigCvelkhQ&list=PLo9FRw1AkDuQLEz7Gvvaz3ideB2NpFtT1
https://archive.org/details/softwarelibrary_msdos?&sort=-downloads&page=1
Kyleran: "Now there's the real trick, learning to accept and enjoy a game for what it offers rather than pass on what might be a great playing experience because it lacks a few features you prefer."
John Henry Newman: "A man would do nothing if he waited until he could do it so well that no one could find fault."
FreddyNoNose: "A good game needs no defense; a bad game has no defense." "Easily digested content is just as easily forgotten."
LacedOpium: "So the question that begs to be asked is, if you are not interested in the game mechanics that define the MMORPG genre, then why are you playing an MMORPG?"
You mean everybody is responding to this just like they did to the offset engine or like they did to the first videos of half-life 2? Oh my! You're right; I should go make sure everybody knows the truth! Voxels are a lie!
Excuse me for being so sarcastic but I think I'll do the sensible thing and not respond to hysteria. It remains to be seen whether or not it is a lie but the technology has been around for longer than most of us here have been computer geeks.
No offense intended to you of course Waynejr2. I realize you were just responding to my comment.
I wouldn't be so quick to write it off as that. Getting investment in Australia for a small business is *extremely* hard if you don't have a prototype to show off in-front of investors. This is purely because you can only receive investment from Professional and Sophisticated Investors who do their research before signing any cheques.
To quality as a Professional Investor in Australia, you need $10m in assets.
Sophisticated? $2.5m in assets and a net income of $250k per year for the last two years.
Anyone who does not meet those requirements cannot legally invest in a small business. They either don't know about this, are quite serious in what they're doing or can use the force to mind-trick rich folk into throwing their money into something with a lot of controversy on the net.