I still don't quite understand how come nobody came up with a solution to voxel animation, but I believe we're all hoping one day we'll be able to forget about polygons and UVs and focus our efforts on doing art, and nothing more.
This demonstration gives me a certain amount of hope, but I'm gonna keep the benefit of the doubt since really, this sounds way too promising to be real. Though amazing things have been invented many times, so who knows.
Oh yeah, regarding unigine engine - it looks amazing, really. But unlike the aussie engine, it's just more of the same. A more advanced version of an existing tech. If we could really break through with a fresh new technology that's more natural in our fingers, I think it's much better.
I believe we're all hoping one day we'll be able to forget about polygons and UVs and focus our efforts on doing art, and nothing more.
That's like saying you look forward to the day when we can forget all about paint and focus on just making art. Or pencil/pen. Nothing will ever "replace" Polygons- just like nothing can replace paint. It is a technique for creation, and people will always practice it so long as they can.
And besides, half this art-form is based on the limitations.
Not sure if anyone gets what I mean- it's hard to explain, I suppose.
But I love polygons and UV mapping. I mean, most of us must, if we want to do it for a living, or do already.
I'm not sure it's such a good idea to try and make an entire renderer solely voxels. They're good for representing some types of data but not for the majority of data. Take a look around whatever office, living room, whatever you're currently in and you'll notice how non-uniform detail is. Most inorganic surfaces have areas of extremely fine and precise detail then large areas of low frequency detail.
Notch does seem to bring up solid points as well.
Its easy to say haters gonna hate and we should be dreamers but we should also be critical thinkers
Even though it's a nice idea. I am not really fond of the blocky textures that are imposed on the models in sacrifice of supposed high detail. Sometimes the flatness of polygons is better sorry to say. Zbrush and a pretty dense polygonal model still can deliver great results.
Pretty interesting to see all this, I remember last year big discussions amongst the programers where I worked in relation to the claims.
I am in Brisbane myself and recently met the brother of the Euclideon CEO, he has talked to me a bit about his brother's venture so I will see what he has to say since a month or so back he was pretty excited about a new vid being released.
I am unsure as to what they plan to do, they seem to be coming at everything certainly from outside the industry and I think only recently hired an artist to assist them with the demos.
They are a very small group too so are trying to get noticed as much as possible, not the best way to go about it as it has been pointed out.
Always good to keep an open mind and tech will always evolve, I think there is definately a lot to come from related voxel tech and as an artist I love the idea of new tools to work with in the future.
i think voxels can be a useful tool for landscape type stuff but ultimately characters should just be polygons for the time being..
I mean, i think all voxel games are great but so far making AA+ type titles doesn't seem to be entirely possible on a pure voxel engine.
With the way the product is presented it doesnt deserve to be taken seriously. There are many things that the developer could be trying to accomplish with these videos.
Maybe he explains the tech in such a noobish and aggressive way to attract viral attention, but by doing that he is making a fool of himself to his actual market and failing to properly market the tech.
Maybe it is an actual vaporware scam and he plans on getting investors and then disappearing, or selling the "idea" to someone who sees magic without boundaries (which is exactly what he is marketing).
In any case the fact is, it is not some new technology. Any graphics programmer has had the idea to create such a renderer, any many have. The big issues with the technique are unbreakable barriers. The biggest being disk and video memory limits. As Notches post explains the amount of data required to create even a relatively small environment by todays standards would require an unthinkable amount of storage, even at very generous 1 byte per point.
Is it possible to make a game with something like this? of course, check out Voxatron
Is it possible to make CoD, Bf3, Crysis or just cause 2? absolutely not.
This definitely is going to be good. Hope they enjoy getting bankrupt and defunct with how many game companies out there that will tear em apart. Rebellion like this is for chumps.
I'd like to give a word of caution to skeptics, in a couple of years this could be the next huge thing and you'll be eating a hat.
Really, revolutions happened before, there's no need to flat out deny the possibility of such a thing existing. I say this, because among ALL of the different engines you guys posted here (which seemed cool enough) - NONE of them even came close to the quality of textures, lighting and detail that this engine presents. Voxel tech is great in concept, but until today nobody ever made anything that really looks good, detailed and realistic. They all suffer from the same phenomenon - bad lighting, choppy framerates, and sculpting on them (say in 3Dcoat) just doesn't feel good, unlike ZBrush.
What I'm saying is, I believe they're working on something, I believe it holds some credibility because the pictures look quite good unlike other voxel engines, but I won't be holding my breath until some actual facts come out. Until then, though, I won't be eulogizing them.
Oh yeah, John Carmack is possibly the best programmer in the industry.
I'd like to give a word of caution to skeptics, in a couple of years this could be the next huge thing and you'll be eating a hat.
Really, revolutions happened before, there's no need to flat out deny the possibility of such a thing existing.
Practical laws aren't just broken because people wish for magic to happen.
No one is saying that the videos are fake (they're very much real), we're saying that they're not mentioning every downside that comes with the tech and tries to sell it as something that could practically replace polygonal tech currently.
I say this, because among ALL of the different engines you guys posted here (which seemed cool enough) - NONE of them even came close to the quality of textures, lighting and detail that this engine presents. Voxel tech is great in concept, but until today nobody ever made anything that really looks good, detailed and realistic. They all suffer from the same phenomenon - bad lighting, choppy framerates, and sculpting on them (say in 3Dcoat) just doesn't feel good, unlike ZBrush.
Which is funny, because zbrush also uses a voxel-like way of handling things.
What I'm saying is, I believe they're working on something, I believe it holds some credibility because the pictures look quite good unlike other voxel engines, but I won't be holding my breath until some actual facts come out. Until then, though, I won't be eulogizing them.
Oh yeah, John Carmack is possibly the best programmer in the industry.
wtf, a kotaku article with words? (they're hard to get at with all the ads)
Nice to see him speaking honestly, without all the sale pitch crap. Makes it sound a lot more down to earth.
"We have animation," claims Bruce, confidently. "We're certainly going to do a lot more work in that area. I have faith that you'll find our animation quite satisfactory, but we have no intention of releasing anything in that department until it looks absolutely 100% because if we release it now, I assure you that no-one will take it as that's where we're up to and we're still working on it', they'll just scream it's not perfect yet! They can't make it perfect! This can't compare to polygons!'"
I'd like to give a word of caution to skeptics, in a couple of years this could be the next huge thing and you'll be eating a hat.
Really, revolutions happened before, there's no need to flat out deny the possibility of such a thing existing. I say this, because among ALL of the different engines you guys posted here (which seemed cool enough) - NONE of them even came close to the quality of textures, lighting and detail that this engine presents. Voxel tech is great in concept, but until today nobody ever made anything that really looks good, detailed and realistic. They all suffer from the same phenomenon - bad lighting, choppy framerates, and sculpting on them (say in 3Dcoat) just doesn't feel good, unlike ZBrush.
What I'm saying is, I believe they're working on something, I believe it holds some credibility because the pictures look quite good unlike other voxel engines, but I won't be holding my breath until some actual facts come out. Until then, though, I won't be eulogizing them.
Oh yeah, John Carmack is possibly the best programmer in the industry.
Doubt it. Voxels will eventually happen, that I agree on; but not in the first couple of years. And when it does happen; I'm pretty sure it will be a hybrid approach of static voxel "meshes" and polygonal dynamic meshes.
And I'm willing to bet you that it will be one of the greats like Mack Carjohn to bring it to the masses. Because people like him have the leverage to get NVidia/AMD/Intel to follow suit and make them add specialized data structures right into the hardware.
But I'm pretty sure we'll see a sort of raytracing development first. The jump from polygons to voxels is just too big right now imho.
Still wonder when those HVDs will be released with it's 6 TB capacity to the general public. It definately will be something to up the graphic capacity maybe for polygonal based games to blast this voxel idea out of the water.
When he explained the tech in Kotaku, it seemed to fit perfectly with practical laws.
ZBrush is using these Pixols - but as far as I know, when you're actually sculpting - it's using polygons only. Once you go to pixol mode, it just "bakes" everything down as pixols and that's it. It's kinda different from what 3DCoat does, which is letting you actually sculpt in 3D space, using voxels. Correct me if I'm wrong, because I could be.
And now I see that I was pretty slow in catching on the Carmack meme. Well, what can you do.
I understand almost nothing about anything said in the video or in the articles. In fact the first time I heard the word voxel was reading this thread. I think. That being said I've got a question...
How much hard drive space would this take up in comparison to our current methods? I understand why you can have all that shit on the screen and what not, that's cool and such, but what types of improvements are we looking at in terms of file size? Or are there not any - would it be worse?
I say this, because among ALL of the different engines you guys posted here (which seemed cool enough) - NONE of them even came close to the quality of textures, lighting and detail that this engine presents.
Now let's set some stuff straight.
Lighting is baked into the objects. Seems like they used Autodesk's new photo modeling service that creates sort of point cloud kinda results but everything is baked into the color textures. Google Photofly.
You could use models of the same (or better) detail in most of today's polygon based engines and they'd even be able to shade them properly. The thing is, noone thought the public would like to see a game world of the same damn tree and rock instanced ten thousand times because it's boring.
I'd say Battlefield 3 can eat this techdemo any day for breakfast.
A 1024x1024 2d texture is 4mb uncompressed. now take that to the 3rd demension. A 1024x1024x1024 volume is about 4gb uncompressed for just color, then you also need surface normal and some other info. for the kind of detail seen in that video you need at least 1 1024x1024x1024 volume per 4 meters or so. You can try to compress that all you want, but its still wayyyyyyyyy too much data for an actual game. If they use points instead of voxels that might ease up on the size a little, but really its still same problem, plus they mention indexed searches for rendering, which makes things even worse.
And Carmack's been talking about this tech years ago anyway. He's done serious research and development, went even further than this guy - instead of instancing everything he looked into data compression and fast streaming from storage to build a unique world.
They've probably had their reasons to stick with polygons for now.
When he explained the tech in Kotaku, it seemed to fit perfectly with practical laws.
He said the exact same things he has said already.
He still avoided talking about storage requirements and dodged the animation-issue with "oh it isn't finished", because of reasons explained by others: it's not fast enough to be feasible in any way.
He goes leaving the unexplained parts still unexplained.
as I understand it each unique point takes a byte of data, at least, without any texture or lighting information. He's talking about "64 "atoms" (read: voxels) per cubic millimeter. That means their grid is 1/4096th of a meter on each edge, so each cubic meter is 4096x4096x4096 = 68.7 billion voxels per cubic meter." That's 63 gigabytes of information in just a cubic meter, but that's only for unique detail, you can repeat model for next to no cost, as demoed by the videos.
If they show animation, 100+ unique models instead of large maps made up of the same 10 models, lighting details, physics/collision solutions, and some actual "game play" demos instead of fly throughs, I'd be a lot more interested.
A 1024x1024 2d texture is 4mb uncompressed. now take that to the 3rd demension.
Voxels aren't stored in 3D textures nowadays, the data structure is called an octree. Basically take a cube and cut it up into 8 smaller cubes again and again. Easier to compress, empty spaces are ignored completely.
That's why everything in this demo is instanced in neat vertical/horizontal offsets, too. He's basically re-using the same content for a certain level of the subdivided cubes as it seems.
It's still a lot of data for current games though...
If they show animation, 100+ unique models instead of large maps made up of the same 10 models, lighting details, physics/collision solutions, and some actual "game play" demos instead of fly throughs, I'd be a lot more interested.
I guess they just cant. I think they are heavily cheating with that instancing (to reduce needed space and to improve the performance). I mean, who would do a tech demo with 1000 times the same model. Like Notch said, theyre just trying to find investors and the way they sell reminds me of those teleshopping channels on tv.
Well yes, cutting out the empty data would be part of the compression, but thats still not close to bringing it to a reasonable disk/video memory usage (especially when in video memory it actually has to be stored as a full 3d texture per chunk).
IT's not really cutting out empty data as it's not even there. You store voxels by storing the number of divisions and which 1/8th of the divided cube they're in, so you don't even store empty data. It's called an octree data structure, basically every voxel is a leaf and you access them by going through the branches of the tree from the root. It's supposed to be very fast to raytrace into in 3D space.
That's why HVDs need to be released for this to be considered even capable. If Consoles get this into pla games will be raped on what's possible
"HVD (Holographic Versatile Disc) is the next generation in optical disk technology. HVD is still in a research phase that would phenomenally increase the disk storage capacities over the currently existing HD DVD and Blu-ray optical disk systems. According to published statistics, when produced in full scale, HVDs will have a storage capacity of 3.9 terabytes (39,000 GB) and a data transfer rate of 1 GB/s, which is at least six times more than the speed of DVD players. This would, without a doubt, become a giant step in revolutionizing the disk storage industry."
Hmm, Carmack gives them a benefit of doubt and doesn't dismiss the whole thing as a scam.
So it's on now, the word of Notch against the word of Carmack.
I know I will go with opinion of the guy who has a proven track record, doesn't happen to act like a douche and is known to be an accomplished programmer for several years.
Or maybe, I will just wait and see if it goes anywhere? :poly142:
Shouldn't games be faster as well Just as beautiful. This Voxel idea will just make things slower. Lately games have been at a 20/30 FPS norm. However it would be nice to see them reach 50/60 FPS like the old days. Getting used to polygons just is helping that more, Games been slower than usual.
Even Carmack says that it's not likely to be used in quite a few years.
He knows as he's done a lot more research into this several years ago, when he talked about sparse voxel octrees back in early 2008. It's just that he hasn't uploaded a video with outrageous claims to youtube to promote it, he simply gave a talk and some interviews about all the details and that's not as sensational for the gaming sites and forums...
wtf, a kotaku article with words? (they're hard to get at with all the ads)
Nice to see him speaking honestly, without all the sale pitch crap. Makes it sound a lot more down to earth.
Yea if he actually pitched it like he talked to Kotaku then I think a lot more people would be a little bit more receptive but still I think his pitch needs a lot more polish.
"Basically we're in the middle of a trilogy and this is like our Empire Strikes Back," he explains
So this is Episode 2 of Star Wars and the people you're trying to sell this too are the evil empire? Oh man...
Episode 1
___________
Euclideon: You can use this to make games! /desk flip
Industry: How? Then lets see you make a game?
Euclideon: I'll be back! /cape twirl
Episode 2
___________
Euclideon:WE'RE BACK! You thought we where dead, you laughed! Well take a look at THIS! /roll video of a duck farting
Industry: Sigh... but can you make a game out of it?
Euclideon: We import objects from the real world into UNLIMITED DETAIL! /echo
Industry: Right ok so you got some better art.
Euclideon: Yea pretty much... We'll be back! /cape twirl
Episode 3
___________
Industry: (talking to a closed door) Hello? We had some questions? Where are you guys?
Euclideon: Shh go away...
Industry: Does your tech scale from a low end smart phone up to a super computer? We still need to know how to make a game with it.
Euclideon: Shh go away...
Euclideon: (hangs a sign that says): "Coming soon... or maybe not. It depends if we run out of money before we manage to invent a parallel industry."
Industry: Hey what's that guy doing over there... it looks like working SVO technology...
Euclideon: (bursts out of the door with a box) TAA--DAA! We have this! hello? anyone?
Even Carmack says that it's not likely to be used in quite a few years.
He knows as he's done a lot more research into this several years ago, when he talked about sparse voxel octrees back in early 2008. It's just that he hasn't uploaded a video with outrageous claims to youtube to promote it, he simply gave a talk and some interviews about all the details and that's not as sensational for the gaming sites and forums...
You're completely right. Even if they released it tomorrow, it would take years for companies to change their pipelines, create proper tools and train their staff. But my point is; Carmack is not dismissing the whole tech based on the way its marketed. I'm not a fan their marketing style either, but it appears that they might have some interesting tech in there after all. Whether it will live up to the hype or not is another story.
I'm simply ticked off that people treat Notch as an authority; which he is not.
I feel like people dismiss Notch simply because. If the information you're reading seems plausible, informative or educated it shouldn't matter who's saying it. At least the videos and info he presented is a good platform for you to do your own research.
His point was this, which I think anyone can understand and agree with who knows a thing or two about this industries history: Voxel technology is nothing new and, if you watch the video and ignore the fella talking in it, this technology has been done before and by others for some time. It's what they do with it next that will make or break the technology.
It's not just the pipeline stuff, it's the insane amount of data that studios would have to create, edit, manage, and then somehow ship to their customers. It is not yeat feasible to have a piece of entertainment software that requires several hundreds of gigabytes of space, it can't be downloaded, it can't be stored on optical disks... And of course the source datasets would be even bigger, you can't compress it while editing.
Sure, there's continuous advancement, so we might get there in like 5 years or so.
In the mean time however, polygons won't stay still either and with tessellation and displacement mapping it may be possible to catch up in detail while keeping the flexibility... Voxels have always been just not yet right for like 20 years. The Commanche games used them and Outcast, and Blade Runner for characters and one of the C&C games for units... but they've always remained a curiosity and poly based tech has always managed to outdo them shortly. So that's why most people are a bit scpetical.
Hey guys, this thread reminds me of a (hypothetical) story.
All of my programmer friends totally believe in this new "MAKE ART" software, does that seem very likely to you? I tried to explain all of the reasons it wouldn't work, but they said that i wasn't an authority because i was just an artist.
You're completely right. Even if they released it tomorrow, it would take years for companies to change their pipelines, create proper tools and train their staff. But my point is; Carmack is not dismissing the whole tech based on the way its marketed. I'm not a fan their marketing style either, but it appears that they might have some interesting tech in there after all. Whether it will live up to the hype or not is another story.
I'm simply ticked off that people treat Notch as an authority; which he is not.
Even Carmack has admitted that it's most likely a better idea to have polygonal characters in a voxel world due to animating voxels being so complex an issue.
But the point has been; it's not the interesting tech they keep trying to say it is if it isn't flexible enough, which ticked Notch (and many others) off, they've been polishing the only part of their package that actually works in that scenario.
They've had since 2003 to actually implement a working animation system.
He pushes back with "and people just don't like our art" "imagine what this would look like in the hands of an artist"
It's not the art, its fine, it gets the point across, it's even impressive at a few points.
What people want to see or play with, is Voxelstien 3D or Mario Voxel (both of these have probably already been posted?)
Even the Mario example gives more info and detail about the project and its just one person? Voxelstien 3D you can play...
I'm not sure if anyone is going to embrace this technology in the future and if they do it probably won't be Unlimited Detail, given their slow pace of development and horrible PR... It would probably need to be a hardware developer that embraces it and I don't see MS or Sony doing that.
Probably not from microsoft they'll use the DX11 tech they've created.
Sony might but I doubt it, they have their own complex set of development tools built off of polygons also. Doesn't the PS3 have some kind of Voxel hair rendering technique? If they aren't done unleashing the PS3's power it's probably still to early to tell what they have lined up next, its probably not anything as radical as throwing out polygons.
Nintendo MIGHT because it suits their style a bit better and its the odd quirky thing to pick.
Replies
=P
This demonstration gives me a certain amount of hope, but I'm gonna keep the benefit of the doubt since really, this sounds way too promising to be real. Though amazing things have been invented many times, so who knows.
Oh yeah, regarding unigine engine - it looks amazing, really. But unlike the aussie engine, it's just more of the same. A more advanced version of an existing tech. If we could really break through with a fresh new technology that's more natural in our fingers, I think it's much better.
http://notch.tumblr.com/post/8386977075/its-a-scam
That's like saying you look forward to the day when we can forget all about paint and focus on just making art. Or pencil/pen. Nothing will ever "replace" Polygons- just like nothing can replace paint. It is a technique for creation, and people will always practice it so long as they can.
And besides, half this art-form is based on the limitations.
Not sure if anyone gets what I mean- it's hard to explain, I suppose.
But I love polygons and UV mapping. I mean, most of us must, if we want to do it for a living, or do already.
google bought youtube??
Notch does seem to bring up solid points as well.
Its easy to say haters gonna hate and we should be dreamers but we should also be critical thinkers
It starts:
http://kotaku.com/5827034/minecraft-creator-calls-bullshit-on-unlimited-detail-graphics-hype
I am in Brisbane myself and recently met the brother of the Euclideon CEO, he has talked to me a bit about his brother's venture so I will see what he has to say since a month or so back he was pretty excited about a new vid being released.
I am unsure as to what they plan to do, they seem to be coming at everything certainly from outside the industry and I think only recently hired an artist to assist them with the demos.
They are a very small group too so are trying to get noticed as much as possible, not the best way to go about it as it has been pointed out.
Always good to keep an open mind and tech will always evolve, I think there is definately a lot to come from related voxel tech and as an artist I love the idea of new tools to work with in the future.
In the meantime >insert popcorn gif<
I mean, i think all voxel games are great but so far making AA+ type titles doesn't seem to be entirely possible on a pure voxel engine.
Maybe he explains the tech in such a noobish and aggressive way to attract viral attention, but by doing that he is making a fool of himself to his actual market and failing to properly market the tech.
Maybe it is an actual vaporware scam and he plans on getting investors and then disappearing, or selling the "idea" to someone who sees magic without boundaries (which is exactly what he is marketing).
In any case the fact is, it is not some new technology. Any graphics programmer has had the idea to create such a renderer, any many have. The big issues with the technique are unbreakable barriers. The biggest being disk and video memory limits. As Notches post explains the amount of data required to create even a relatively small environment by todays standards would require an unthinkable amount of storage, even at very generous 1 byte per point.
Is it possible to make a game with something like this? of course, check out Voxatron
Is it possible to make CoD, Bf3, Crysis or just cause 2? absolutely not.
Typical Game company reaction right now
Nope its a hoax.
Minecraft is voxels, then rendered as polygons. I believe that answers all 3 of your questions at once.
Yes
No
It can be
Really, revolutions happened before, there's no need to flat out deny the possibility of such a thing existing. I say this, because among ALL of the different engines you guys posted here (which seemed cool enough) - NONE of them even came close to the quality of textures, lighting and detail that this engine presents. Voxel tech is great in concept, but until today nobody ever made anything that really looks good, detailed and realistic. They all suffer from the same phenomenon - bad lighting, choppy framerates, and sculpting on them (say in 3Dcoat) just doesn't feel good, unlike ZBrush.
What I'm saying is, I believe they're working on something, I believe it holds some credibility because the pictures look quite good unlike other voxel engines, but I won't be holding my breath until some actual facts come out. Until then, though, I won't be eulogizing them.
Oh yeah, John Carmack is possibly the best programmer in the industry.
Practical laws aren't just broken because people wish for magic to happen.
No one is saying that the videos are fake (they're very much real), we're saying that they're not mentioning every downside that comes with the tech and tries to sell it as something that could practically replace polygonal tech currently.
Which is funny, because zbrush also uses a voxel-like way of handling things.
Who is this Car Johnmack?
wtf, a kotaku article with words? (they're hard to get at with all the ads)
Nice to see him speaking honestly, without all the sale pitch crap. Makes it sound a lot more down to earth.
Doubt it. Voxels will eventually happen, that I agree on; but not in the first couple of years. And when it does happen; I'm pretty sure it will be a hybrid approach of static voxel "meshes" and polygonal dynamic meshes.
And I'm willing to bet you that it will be one of the greats like Mack Carjohn to bring it to the masses. Because people like him have the leverage to get NVidia/AMD/Intel to follow suit and make them add specialized data structures right into the hardware.
But I'm pretty sure we'll see a sort of raytracing development first. The jump from polygons to voxels is just too big right now imho.
ZBrush is using these Pixols - but as far as I know, when you're actually sculpting - it's using polygons only. Once you go to pixol mode, it just "bakes" everything down as pixols and that's it. It's kinda different from what 3DCoat does, which is letting you actually sculpt in 3D space, using voxels. Correct me if I'm wrong, because I could be.
And now I see that I was pretty slow in catching on the Carmack meme. Well, what can you do.
How much hard drive space would this take up in comparison to our current methods? I understand why you can have all that shit on the screen and what not, that's cool and such, but what types of improvements are we looking at in terms of file size? Or are there not any - would it be worse?
Sorry if that's a dumb thing to ask.
Now let's set some stuff straight.
Lighting is baked into the objects. Seems like they used Autodesk's new photo modeling service that creates sort of point cloud kinda results but everything is baked into the color textures. Google Photofly.
You could use models of the same (or better) detail in most of today's polygon based engines and they'd even be able to shade them properly. The thing is, noone thought the public would like to see a game world of the same damn tree and rock instanced ten thousand times because it's boring.
I'd say Battlefield 3 can eat this techdemo any day for breakfast.
They've probably had their reasons to stick with polygons for now.
He said the exact same things he has said already.
He still avoided talking about storage requirements and dodged the animation-issue with "oh it isn't finished", because of reasons explained by others: it's not fast enough to be feasible in any way.
He goes leaving the unexplained parts still unexplained.
If they show animation, 100+ unique models instead of large maps made up of the same 10 models, lighting details, physics/collision solutions, and some actual "game play" demos instead of fly throughs, I'd be a lot more interested.
Voxels aren't stored in 3D textures nowadays, the data structure is called an octree. Basically take a cube and cut it up into 8 smaller cubes again and again. Easier to compress, empty spaces are ignored completely.
That's why everything in this demo is instanced in neat vertical/horizontal offsets, too. He's basically re-using the same content for a certain level of the subdivided cubes as it seems.
It's still a lot of data for current games though...
I guess they just cant. I think they are heavily cheating with that instancing (to reduce needed space and to improve the performance). I mean, who would do a tech demo with 1000 times the same model. Like Notch said, theyre just trying to find investors and the way they sell reminds me of those teleshopping channels on tv.
"HVD (Holographic Versatile Disc) is the next generation in optical disk technology. HVD is still in a research phase that would phenomenally increase the disk storage capacities over the currently existing HD DVD and Blu-ray optical disk systems. According to published statistics, when produced in full scale, HVDs will have a storage capacity of 3.9 terabytes (39,000 GB) and a data transfer rate of 1 GB/s, which is at least six times more than the speed of DVD players. This would, without a doubt, become a giant step in revolutionizing the disk storage industry."
So it's on now, the word of Notch against the word of Carmack.
I know I will go with opinion of the guy who has a proven track record, doesn't happen to act like a douche and is known to be an accomplished programmer for several years.
Or maybe, I will just wait and see if it goes anywhere? :poly142:
He knows as he's done a lot more research into this several years ago, when he talked about sparse voxel octrees back in early 2008. It's just that he hasn't uploaded a video with outrageous claims to youtube to promote it, he simply gave a talk and some interviews about all the details and that's not as sensational for the gaming sites and forums...
Episode 1
___________
Euclideon: You can use this to make games! /desk flip
Industry: How? Then lets see you make a game?
Euclideon: I'll be back! /cape twirl
Episode 2
___________
Euclideon:WE'RE BACK! You thought we where dead, you laughed! Well take a look at THIS! /roll video of a duck farting
Industry: Sigh... but can you make a game out of it?
Euclideon: We import objects from the real world into UNLIMITED DETAIL! /echo
Industry: Right ok so you got some better art.
Euclideon: Yea pretty much... We'll be back! /cape twirl
Episode 3
___________
Industry: (talking to a closed door) Hello? We had some questions? Where are you guys?
Euclideon: Shh go away...
Industry: Does your tech scale from a low end smart phone up to a super computer? We still need to know how to make a game with it.
Euclideon: Shh go away...
Euclideon: (hangs a sign that says): "Coming soon... or maybe not. It depends if we run out of money before we manage to invent a parallel industry."
Industry: Hey what's that guy doing over there... it looks like working SVO technology...
Euclideon: (bursts out of the door with a box) TAA--DAA! We have this! hello? anyone?
You're completely right. Even if they released it tomorrow, it would take years for companies to change their pipelines, create proper tools and train their staff. But my point is; Carmack is not dismissing the whole tech based on the way its marketed. I'm not a fan their marketing style either, but it appears that they might have some interesting tech in there after all. Whether it will live up to the hype or not is another story.
I'm simply ticked off that people treat Notch as an authority; which he is not.
His point was this, which I think anyone can understand and agree with who knows a thing or two about this industries history: Voxel technology is nothing new and, if you watch the video and ignore the fella talking in it, this technology has been done before and by others for some time. It's what they do with it next that will make or break the technology.
Sure, there's continuous advancement, so we might get there in like 5 years or so.
In the mean time however, polygons won't stay still either and with tessellation and displacement mapping it may be possible to catch up in detail while keeping the flexibility... Voxels have always been just not yet right for like 20 years. The Commanche games used them and Outcast, and Blade Runner for characters and one of the C&C games for units... but they've always remained a curiosity and poly based tech has always managed to outdo them shortly. So that's why most people are a bit scpetical.
All of my programmer friends totally believe in this new "MAKE ART" software, does that seem very likely to you? I tried to explain all of the reasons it wouldn't work, but they said that i wasn't an authority because i was just an artist.
Even Carmack has admitted that it's most likely a better idea to have polygonal characters in a voxel world due to animating voxels being so complex an issue.
But the point has been; it's not the interesting tech they keep trying to say it is if it isn't flexible enough, which ticked Notch (and many others) off, they've been polishing the only part of their package that actually works in that scenario.
They've had since 2003 to actually implement a working animation system.
It's not the art, its fine, it gets the point across, it's even impressive at a few points.
What people want to see or play with, is Voxelstien 3D or Mario Voxel (both of these have probably already been posted?)
Both of these show off a lot more and are way father along and probably under development for a lot less.
http://www.youtube.com/watch?v=RfT0qGGrIzc
http://www.youtube.com/watch?v=6ZUdsRdokO4
Even the Mario example gives more info and detail about the project and its just one person? Voxelstien 3D you can play...
I'm not sure if anyone is going to embrace this technology in the future and if they do it probably won't be Unlimited Detail, given their slow pace of development and horrible PR... It would probably need to be a hardware developer that embraces it and I don't see MS or Sony doing that.
Probably not from microsoft they'll use the DX11 tech they've created.
Sony might but I doubt it, they have their own complex set of development tools built off of polygons also. Doesn't the PS3 have some kind of Voxel hair rendering technique? If they aren't done unleashing the PS3's power it's probably still to early to tell what they have lined up next, its probably not anything as radical as throwing out polygons.
Nintendo MIGHT because it suits their style a bit better and its the odd quirky thing to pick.