Impressive New Graphics Tech

[Del Valle]

Potential to revolutionize gaming graphics if it isn't all smoke and mirrors.

[mrmouth]

As someone in the youtube comments sections stated: they have shown nothing animated, just static images. But those static images look amazing. The rocks specifically (that is towards th end if you dont want to watch all the way through).

[Top-Tier-Tech]

As someone in the youtube comments sections stated: they have shown nothing animated, just static images.

I'm not sure any current gen hardware could run anything like that animated and not stutter like heck. This is certainly the future of gaming we just have to wait a couple more years for the GPU's that can do something with it.

[therussian]

Another rather pressing concern that someone commented on is dynamic lighting. How the hell are you going to accurately do lighting for tens of thousands of polygons per item and not die of old age while doing it?

Still, it looks ing sweet. Too bad I'll never own a computer good enough to run it.


Also kind of interesting that they put so much energy into little things like dirt but make the "river banks" straight lines

[Crazyeyesreaper]

its fake

its voxel rendering its not possible in games not today not tomorrow not even 5-6 years from now

Procedurally generated worlds don't work for games today. Good luck trying to mimic a Crysis or BC2 map with procedurally created data and fractals. lol

For a real example of what we are talking about read this:

http://artis.imag.fr/Publications/20...E09/CNLE09.pdf

and watch this:

http://www.youtube.com/watch?v=HScYu...layer_embedded

In this video the only example that is using actual data instead of procedurally creating it or instancing using a fractal is the last example, the human body.

Highlights of that example:

Running on a GTX280.
512x512 resolution.
10-20 fps
dataset > 32 GB (and it's just a human body)


The idea of Voxel rendering is to have higher detail + more control over the details than with polygon+tesselation+displacement, at least if you are going to put the graphics world upside down. In order to achieve that every voxel has to be unique and represent the world as the artist wanted it to be. If you are talking about creating the point/voxel data out of a polygonal mess+procedural modifiers, forget about it it would be extremely slow (not posible with current hardware) and the resulting data would be huge nonetheless. At some point, on the HDD or main memory or cache the models have to be deconpressed and they just wouldn't fit in current hardware. You can't even stream it.

The best compression ratio for sparse voxel octrees data structures have been achieved (claimed) by Jon Olick from Id and that means an average of 1.5 bits per voxel iirc, color included, quite impressive if you think about it, but... Now look at this video of a rabbit created out of voxels:

http://www.youtube.com/user/Akvaknarre

As you can see it's not extremely detailed, but it's made out of a 512x512x360 voxel grid. That's 94.5 million voxels, so with Olick's compression that's 140 million bits ~= 17 MB, just for the geometric data of a small animal that doesn't even look very well. You would most probably want 1024x1024x700 grid (136 MB) and to make it really worth it over poly+tesselation+displacement you'd probably need 2048x (1080 MB). For a rabbit. same rabbit will look 1000 times better with a 100 KB mesh and a 2 MB displacement texture.

For a good worldmap, I'd say the maximum relative size of a voxel would need to be like 1 mm in the real world (imagine a world made out of 1mm sized small boxes) so just imagine what the voxel number would be for a 10mx10mx3m room, 3x10^12 = 3000000000000 voxels. That's 520 GB worth of data, compressed.

so yea its all a giant ing hoax and has been other companies have far better showings then the guys above and they have real tech demos and usage and even they admit its limited and not possible these guys above have been turned down by every major company so far, including Nvidia, AMD, Intel etc, because its just not feasible, we will have Ray tracing in games *we do to an extent already before we go to Voxels and ray tracing wont be implemented game wise untill the Next ID tech engine is released after Rage, which would be 4-6 years from now. So Basically ID The only company remotely interested in this dosent need any other companies help at this point Id also has the best compression algorithm out of anyone and even they dont expect it to be worth experimenting with for another 5 years when Id Tech 6 will be near the end of its development stage.

[Dewy]

its fake

That's a paradox if I've ever seen one.

its voxel rendering its not possible in games not today not tomorrow not even 5-6 years from now

Voxel rendering has been used in games for well over a decade. So yes it's entirely possible. And who said it's voxels? They claim it's point cloud data, and until they release the SDK we have no strong evidence to claim its not point cloud data.

Procedurally generated worlds don't work for games today. Good luck trying to mimic a Crysis or BC2 map with procedurally created data and fractals. lol

The island wasn't procedurally generated.

In this video the only example that is using actual data instead of procedurally creating it or instancing using a fractal is the last example, the human body.
Highlights of that example:

Running on a GTX280.
512x512 resolution.
10-20 fps
dataset > 32 GB (and it's just a human body)

That unlimited detail island demo ran at 20 fps in software. If what they say is true that's already a lot better than this voxel tech.

The idea of Voxel rendering is to have higher detail + more control over the details than with polygon+tesselation+displacement, at least if you are going to put the graphics world upside down. In order to achieve that every voxel has to be unique and represent the world as the artist wanted it to be.

Why does every voxel need to be unique? That is just stupid from a technical and artistic perspective.

For a good worldmap, I'd say the maximum relative size of a voxel would need to be like 1 mm in the real world (imagine a world made out of 1mm sized small boxes) so just imagine what the voxel number would be for a 10mx10mx3m room, 3x10^12 = 3000000000000 voxels. That's 520 GB worth of data, compressed.

Your maths failed for the same reasons Notch's did.

so yea its all a giant ing hoax and has been other companies have far better showings then the guys above and they have real tech demos and usage and even they admit its limited and not possible these guys above have been turned down by every major company so far, including Nvidia, AMD, Intel etc, because its just not feasible, we will have Ray tracing in games *we do to an extent already before we go to Voxels and ray tracing wont be implemented game wise untill the Next ID tech engine is released after Rage, which would be 4-6 years from now.

Wow dude use sentences.

so yea its all a giant ing hoax and has been other companies have far better showings then the guys above

So 20-30 fps in hardware is better than 20 fps in software?

and they have real tech demos

Because investors wouldn't want to have a go at a tech demo before putting money into it? I'm pretty sure the Australian government wouldn't give $2 million for commercialisation (not research) without seeing a tech demo in person. If it was for research then maybe they wouldn't have seen a tech demo. And by the way the Australian Government thought Metal Storm was a hoax at first. 16,000 bullets per second does sound to good to be true right? Yeah and so does Unlimited Detail. But metal storm turned out to be real, so hopefully Unlimited Detail does.

So Basically ID The only company remotely interested in this

Euclideon Pty Ltd is a company.

Id also has the best compression algorithm out of anyone

Can you verify that claim?

and even they dont expect it to be worth experimenting with for another 5 years when Id Tech 6 will be near the end of its development stage.

They are experimenting with it.

[Doono]

I remeber reading about this tech about a year ago, as impressive as it looks it had one major flaw, you cannot animate objects using it, rendering it useless for the creating of things such as characters or rendering trees/grass blowing in the wind.

I suppose it may be possible to use some form of hybrid system where the static environment is rendered using the unlimited detail system and animated objects are rendered in the traditonal way however I have concerns that the lower detail of the animated objects may look out of place.

[Dewy]

Anyone else notice Bruce Dell's voice changed from the video last year to this latest video? Or is that just me?

I remeber reading about this tech about a year ago, as impressive as it looks it had one major flaw, you cannot animate objects using it, rendering it useless for the creating of things such as characters or rendering trees/grass blowing in the wind.

I suppose it may be possible to use some form of hybrid system where the static environment is rendered using the unlimited detail system and animated objects are rendered in the traditonal way however I have concerns that the lower detail of the animated objects may look out of place.

It is possible to animate in both point cloud data and voxels.

[a tw player]

The future of computer graphics for me would be creating the game worlds where everything has "weight" to it, thus allowing realistic physics to occur spontaneously rather than have things animated to create an illusion of it. Right now it can be done in computer simulations, but I wonder how long it will take for the average gaming computer to be able to do it without lag...

Once the fundamental thing above is done, then the visuals of the things can be improved and each game would actually feel real rather than just look real. Does anyone else get annoyed when, for example, in Shogun 2 you see soldiers moveing into each other and limbs disappearing into the other's body? I just want everything to have weight to it rather than simply be an illusion.

Ie. in this system you would see realistic damage done by collision, footprints in dirt or snow, things actually getting wet rather than the mirage of Shogun 2's wet ground etc etc...Am I the only one who wants this dream, and is it possible in the next 10 years?

[Doono]

I stand corrected, I'm sure I read it was not possible to animate using it in the past though.

[Crazyeyesreaper]

voxels have been used i ngames in terms of low lvl sprites etc, or used in production phase to create polygon geometry

using point cload data in game using voxel like system isnt doable in real time on todays hardware or software, not with physics, animation, HDR lighting, etc, and again it still dosent change the fact that storage needed for this kind of tech to be used i ngames is astronomical, but im guessing we all know more then Carmack and Id software since there the only actual game developer pursuing this tech for there own game engines, and they still dont expect it to happen for another 5+ years.

We will have ray tracing in games before we have Point Cloud data generated voxel systems using atom sized objects to create meshes. The only reason there getting 20fps in software is because there using Instancing, every damn tree leave etc is the same as the one in the exact same tree next to it,

overall its taken them over a year, to produce 1 tech demo that shows nothing new, other companies ahve done what they have already and there further along, not to mention theyve managed fluid dynamics etc but it dosent run like these guys state it does.
large #s of voxels act similar to point cloud data, the sad fact is the amount of space and resources needed to use them is astronomical

a game like Oblivion if using point cloud data, or voxel system would be unable to run on todays hardware or tomorrows.

My key point is tho if these guys are on the right track,

Why has AMD, Intel, Nvidia, and every other major tech player not grabbed them? why are they still looking for investments from these companies, lol ill be old and grey before this tech is actually used.

using point cloud data, is a great concept, just theres been companies working on this for 5 6 7 years now and they all reach the same point as Euclideon or w.e and they tend to dissappear or run out of funding.

for reference

http://www.youtube.com/watch?v=lpfaFrazOn4

that 1 room is 2.7gb of data now imagine that being used for an entire city or game world yea, again it wont happen.

CPU Ram and HDD/ SSD speeds arent up to par yet. for true streaming of Point Cloud Data simply looking at the lowest common denominator means for this to be used in games and push polygons out of the way Consoles need to be able to implement it, and look better then what polgyons can use currently Polygons + Normal is far more efficient theres a reason Voxels were abandoned years ago, not to mention these guys make excuses for not having shadows and lighting ready to go, when a single guy by himself has been working for 10 years on a voxel engine, that also allows the use of Polygons, he also has physics and fluid dynamics, but yea lets say these guys know what there doing when they have even less to show then the others researching point cloud data / voxels

so untill Euclideon can offer something better then this guy
http://www.atomontage.com/?id=home

he talks about Voxels and polygons and the fact that voxels or point cloud data wont replace polygons it can be used to augment and offer volumetric data, but he goes a bit more in depth, where as Euclideon just basically brags hey were awesome pardon the no shadows and other were missing but were still awesome. lol


Euclideon is pretty much a joke.

[Dewy]

Edit: I love how the only technical thing I said in this post was telling you what geometric instancing is and how it saves memory. Please someone come up with a good technical argument against voxels.

voxels have been used i ngames in terms of low lvl sprites etc, or used in production phase to create polygon geometry

Arh no Voxels are usually used in terrain. Off the top of my head Novalogic games and Outcast. C4 engine has voxel terrain. While talking about Ken Silverman below I remembered two other games, Blood and Shadow Warrior.

using point cload data in game using voxel like system isnt doable in real time on todays hardware or software, not with physics, animation, HDR lighting, etc, and again it still dosent change the fact that storage needed for this kind of tech to be used i ngames is astronomical, but im guessing we all know more then Carmack and Id software since there the only actual game developer pursuing this tech for there own game engines, and they still dont expect it to happen for another 5+ years.

Why is John Carmack the be and all end all for voxels? Ken Silverman has never said voxels won't happen for another 5+ years, actually quite the contrary he has been working with voxels since he included them in the Build Engine, a little bit after Duke Nukem 3D released. Ken said the greatest disadvantage to voxels is the lack of 3D hardware support. Nothing about the hard drive space required to make a game.

Oh and here's a quote from John Carmack

If I had to pick who I think is just the most talented, it would probably be Ken Silverman

We will have ray tracing in games before we have Point Cloud data generated voxel systems using atom sized objects to create meshes. The only reason there getting 20fps in software is because there using Instancing, every damn tree leave etc is the same as the one in the exact same tree next to it,

You do know what geometric instancing is right? Geometric instancing allows you to draw the same model multiple times without having mulitple copies of the vertex and index data, which saves memory. I don't know if it's the same for point cloud data. You still have to draw the model multiple times you just use the same vertex data.

overall its taken them over a year, to produce 1 tech demo that shows nothing new, other companies ahve done what they have already and there further along, not to mention theyve managed fluid dynamics etc but it dosent run like these guys state it does.
large #s of voxels act similar to point cloud data, the sad fact is the amount of space and resources needed to use them is astronomical

Voxelstein isn't massive.

a game like Oblivion if using point cloud data, or voxel system would be unable to run on todays hardware or tomorrows.

Why does the game have to be like Oblivion? As Ken said, make a game with voxels that a polygon game can't do. That wasn't verbatim.

My key point is tho if these guys are on the right track,

Why has AMD, Intel, Nvidia, and every other major tech player not grabbed them? why are they still looking for investments from these companies, lol ill be old and grey before this tech is actually used.

If you've invested billions into polygons why would you pour that down the drain?

using point cloud data, is a great concept, just theres been companies working on this for 5 6 7 years now and they all reach the same point as Euclideon or w.e and they tend to dissappear or run out of funding.

Bruce started this as a hobby 15 years ago. So they've already doubled your threshold for failing.

for reference

http://www.youtube.com/watch?v=lpfaFrazOn4

that 1 room is 2.7gb of data now imagine that being used for an entire city or game world yea, again it wont happen.

Voxelstein3D is 65 mb (or there abouts) including everything used to run it (such as sound, software) not just voxels.

for true streaming of Point Cloud Data simply looking at the lowest common denominator means for this to be used in games and push polygons out of the way Consoles need to be able to implement it, and look better then what polgyons can use currently Polygons + Normal is far more efficient theres a reason Voxels were abandoned years ago, not to mention these guys make excuses for not having shadows and lighting ready to go, when a single guy by himself has been working for 10 years on a voxel engine, that also allows the use of Polygons, he also has physics and fluid dynamics, but yea lets say these guys know what there doing when they have even less to show then the others researching point cloud data / voxels

Again use sentences. It's easier to read and for me to break it up and quote it.

for true streaming of Point Cloud Data simply looking at the lowest common denominator means for this to be used in games and push polygons out of the way Consoles need to be able to implement it, and look better then what polgyons can use currently

Why do voxels need to look better than polygons? The gameplay reasons for using voxels IMO far outweighs the graphical reasons. Then again I love Doom while quite a lot of people these days wouldn't play it due to it's graphics.

when a single guy by himself has been working for 10 years on a voxel engine, that also allows the use of Polygons, he also has physics and fluid dynamics, but yea lets say these guys know what there doing when they have even less to show then the others researching point cloud data / voxels

Eric Lengyel in less time has craped all over the atomontage engine. Yeah sure the C4 engine doesn't have (IIRC) voxel destruction, but then again the C4 engine can have a 10km^2 map with multiplayer physics--bet the atomontage engine can't do that. Oh and I think Eric is making a background loader for seamless transition between worlds, so you can have giant maps.

so untill Euclideon can offer something better then this guy
http://www.atomontage.com/?id=home

Define better.

he talks about Voxels and polygons and the fact that voxels or point cloud data wont replace polygons it can be used to augment and offer volumetric data, but he goes a bit more in depth, where as Euclideon just basically brags hey were awesome pardon the no shadows and other were missing but were still awesome. lol

There's shadows in Unlimited Detail. The atomontage (what a stupid name ) engine guys are also boasting about how they're going to change the world with their engine.

Euclideon is pretty much a joke.

Yeah and so are a lot of things but you don't see me going around bashing them.

[the_mango55]

Vaporware

[Bolkonsky]

Oh not the voxel argument again. Looks a lot better than the last time I saw these things though. One of the main problems was that they just aren't conventional at all, thus making doing things with them infinitely more difficult. There's worthless at the moment, however, it is people like this that are truly advancing technology, directly or indirectly.

Also yes, you can animate them, but it's not as easy as a few simple keyframes in 3ds Max. Again, not conventional at all. You'll never see these used in a production environment.

@Dewy, yeah, it threw me off, I wasn't sure if it was the same guys.

[english tyrant]

From Notches blog. Frankly I agree with everything he says

Perhaps you’ve seen the videos about some groundbreaking “unlimited detail” rendering technology? If not, check it out here, then get back to this post: http://www.youtube.com/watch?v=00gAbgBu8R4

Well, it is a scam.

They made a voxel renderer, probably based on sparse voxel octrees. That’s cool and all, but.. To quote the video, the island in the video is one km^2. Let’s assume a modest island height of just eight meters, and we end up with 0.008 km^3. At 64 atoms per cubic millimeter (four per millimeter), that is a total of 512 000 000 000 000 000 atoms. If each voxel is made up of one byte of data, that is a total of 512 petabytes of information, or about 170 000 three-terrabyte harddrives full of information. In reality, you will need way more than just one byte of data per voxel to do colors and lighting, and the island is probably way taller than just eight meters, so that estimate is very optimistic.

So obviously, it’s not made up of that many unique voxels.

In the video, you can make up loads of repeated structured, all roughly the same size. Sparse voxel octrees work great for this, as you don’t need to have unique data in each leaf node, but can reference the same data repeatedly (at fixed intervals) with great speed and memory efficiency. This explains how they can have that much data, but it also shows one of the biggest weaknesses of their engine.

Another weakness is that voxels are horrible for doing animation, because there is no current fast algorithms for deforming a voxel cloud based on a skeletal mesh, and if you do keyframe animation, you end up with a LOT of data. It’s possible to rotate, scale and translate individual chunks of voxel data to do simple animation (imagine one chunk for the upper arm, one for the lower, one for the torso, and so on), but it’s not going to look as nice as polygon based animated characters do.

It’s a very pretty and very impressive piece of technology, but they’re carefully avoiding to mention any of the drawbacks, and they’re pretending like what they’re doing is something new and impressive. In reality, it’s been done several times before.

There’s the very impressive looking Atomontage Engine: http://www.youtube.com/watch?v=Gshc8GMTa1Y

Ken Silverman (the guy who wrote the Build engine, used in Duke Nukem 3D) has been working on a voxel engine called Voxlap, which is the basis for Voxelstein 3d: http://www.youtube.com/watch?v=oB1eMC9Jdsw

And there’s more: http://www.youtube.com/watch?v=xUe4ofdz5oI http://www.youtube.com/watch?v=lEHIUC4LNFE http://www.youtube.com/watch?v=Zl9CiGJiZuc

They’re hyping this as something new and revolutionary because they want funding. It’s a scam. Don’t get excited.

Or, more correctly, get excited about voxels, but not about the snake oil salesmen.

[the_mango55]

How are they going to do collision detection

[a tw player]

How are they going to do collision detection

Well I posted about my dream to make everything have "weight" in my post earlier, perhaps as they're using an atom mesh right the way through, the models will have some sort of depth to them, giving them this weight I crave?