Isn't a polygon at its simplest form a triangle which is composed of only 3 verts which can be any length apart from each other, even near overlapping?
And isn't a point an infinitely small thing as to have no dimension itself whatsoever?
So to be drawn a point must in fact become at least the simplest form of a polygon aka. triangle at the very least to be visible no matter the size?
I'd think for a super-cala-fradja-listic-expi-ali-docious programming team that figured their way around the current system would at least refer to it by a proper name or come up with a name for it which would more accurately describe it...
Now that I'm done blowing smoke out of my ass
If its real and it can be lit, animated, have physics applied, yadi yadi yadi, then awesome and I welcome the revoloution and all the knowledge that will come with it, if not...well cold fusion anyone?
Last few posts are a tad pedantic for me...seriously, just deal with it. Also sorry for the wall of text, it kinda bummed me out after I realised I wrote so much.
I agree with the folks in the "cool tech, though not yet" wagon. I know people question "is this needed?" But, a good deal of the tech world has not really been concerned with what is needed, but rather question "is it better?".
It is pretty awesome to think of a 3d world that is composed of essentially simulated molecules. I remember thinking this idea was really keen a shit long time ago, and even posting the concept here (this was prior to my knowledge of voxels). However once we get to a point were things like screen resolution vs asset resolution aren't really an issue (like the aforementioned console problem), so voxel density would match screen resolution. Until then I can't imagine this being much of a standard, but in the future? Why not.
I would imagine if things like collision, physics, and all sorts of physical properties are properly paid attention to, you'd have much more convincing animations in characters and environments. Also like mentioned before, converting polygonal mesh to voxel is not hard. And I would imagine making texture would either be like a voxel paint deal (also mentioned) or alternatively a sort of render to texture, but rather render to voxel. In which you still texture it regularly with uvs, but then have the uv texture colors projected on to their corresponding voxels. But like I mentioned before about voxel density vs resolution, it would only look as good as the created texture if the voxel density matched the vexel density.
I would say that one of the things seems iffy regarding asset creation is that artists would be tasked with a shit ton more work, because now not only will you be responsible for modeling the facade of objects, but also any layer that would happen to exist within it. Let's say you wanna make a house that will be in a war game, and therefore be prone to explosion. Make the house, now make the interior, now make the infrastructure, now make the systems that happen to exist in a home (plumbing and electrical). Same goes for making a human in a kind of game where people are blown to bits, you'd have to learn not only the exterior of the being's body but all it's inner workings. Which I can only imagine being a tad disheartening to non-psychopaths to see someone's intestines spill in a much more realistic manner. Anyway, like I was saying this all leads to a workload that is much more intensive. As grim as it sounds, if this was going to be the new standard in the future, then I would imagine in order to have a job, you'll have to be knowledgeable of such things. So people who think it's hard to get job now as an artist, it may get much harder. Also for people whom are hoping to keep their job it may be harder to keep. Or on the other side of spectrum, this means more people will be employed to just do these substructures of meshes to be calculated into voxels. I do not know, perhaps this will all be incorporated as a gimmick for game and rarely heard from again, or it will indeed eventually become an industry standard. Either way, as an artist within this field I would certainly not disregard it as far as your practice goes.
NG, why would you assume that infinite detail automatically means realistic?
Anyways. I am always excited for new tech to break through, since the cool side effect is that it always brings new ways of working. But on the other hand, claiming that it's gonna be the next big thing without anything to back it up ... is very pedantic from these guys. I feel as if the narrator on these videos never ever worked with polies and ditches them just like that. Try making a quick blockout model in voxels, hehe.
NG, why would you assume that infinite detail automatically means realistic?
Because what else would you use infinite detail for?
I think a distinction should be made between IRL realism and realism in the sense of high detail. A dragon is not realistic (dragons don't exist for all we know), but with this tech you could make it look realistic (scaly skin, round tail instead of a low poly with baked normals). It's a drive for detail and this still requires more artistic prowess.
As for unrealistic games, like those artsy games with blocks and simple shapes, this tech won't give anything but waste resource by rendering a 6 sided cube as million point cluster. It's still a cube.
Vcool said it very well. It's all about "realism" relative to the subject matter. But you're right Pior, the examples I used were only for the games that seek absolute realism, which may have been a bad move, because games obviously cover all aspects of style and physics. However, I feel that even with the most cartooney of environments, interaction between a character and their world, and vice versa should be a large concern for developers, and most of the problems I can think of can be resolved in a more advanced animation system that would assign voxels a certain amounts of physical capacity to create a good synergy between character and world. Of course I'm speaking in an ideal hypothetical situation where the technology is finessed to the point where something like a voxel based character is rigged to a skeleton, and then layers of voxels are assigned certain amounts of densities and physics, so when a foot touched the ground the voxels on the out most layer would deform according to the surface and then as the voxels got deeper it would put up more resistance and there for appear the character has weight and volume and that there is indeed material beneath the surface of the "skin" of the character. Because even super cartooney cartoons allude to the the fact that characters have a certain weight. Therefore, I feel the easiest to create such a thing, would be to model the character, then model the layers of said densities, and composite them all into a voxel based mesh. Yeah, it's not modeling intricate systems of a being or an object, but it still would require a series of other meshes for a character. Although, I could certainly be wrong considering this is all hypothetical, haha.
But why would you think it all has to be that complex? The cartoon example is great. Cartoon characters have a sense of weight and balance ... yet I dont think any cartoon animator or inbetweener wastes time drawing accurate skeleton structures, muscle overlays, nervous systems or skin layers.
Exactly, the theory behind voxels are being used all the time, so its not that we're a big bunch of witchburners when it comes to voxels, its just that polygons works for every case.
And in the case of not using procedural stuff and doing everything by hand, slowly, that's how it has to be done, in the cases where we cannot prebake that stuff, since we do not have UNLIMITED hardware, we want to do as much with as little as possible.
or do like red faction did, and put some of the budget towards having such things.
voxels has had its time in gaming, but polygons just always became easier to use and with more accurate results.
voxels did have some interesting points of use though:
Goddamn polygon lovers.
I for one like the highpoly process of making a game asset, but dislikes the lowpoly bit.
A voxel technology would take care of that and i wouldn't have to touch a farking lowpoly model ever again and all that's involved in that process (UVing, Smoothing groups, lod steps, test baking normals etc)
Id rather just make my highpoly model, hit "MakeVoxelModel.exe" and import it into the game...DONE!
And since you can avoid the lowpoly pipeline completely you can spend a few extra days on polishing the highpoly model.
Also, some models are pretty much impossible to make with a low polycount, which limits the designs a bit, or you have to find workarounds. That would also be a thing of the past, you can just go all out on a model and the engine will just handle it.
in a nutshell, game art quality would go up in general = win
that is some aweful music in the background but doesnt really go into the limitations of the system at all and still has that guy narating that sounds real condescending
lol why is HL2 running at like 5 FPS? What's weird is they also show comparison to crysis which actually looks better....than the psychedelic instanced color town demos of theirs :P
saw the second part of that... he talks about putting it onto the wii and ds. He even goes as far as claiming with this tech the wii would be more powerful than a ps3.
I wish that narrator wouldn't talk as if he's explaining it to kindergartners... Seriously he calls the textures "Pictures that you can put onto a model".
Stupid polygonal artists using a stupid outdated system, but it's not their fault, they're just incapable of making their own decisions on switching to the UNLIMITED point cloud dartar and thus wallow in their pitiful existence making hexagonal pots.
Stupid polygonal artists using a stupid outdated system, but it's not their fault, they're just incapable of making their own decisions on switching to the UNLIMITED point cloud dartar and thus wallow in their pitiful existence making hexagonal pots.
Interesting that pictures of Blade Runner have been posted, yet nobody has mentioned the Blade Runner game (made here in Las Vegas by Westwood Studios) used animated voxels for the real-time characters! I hate to say it, but the voxel characters, as they appeared in the game, looked completely awful, even compared to the super lowpoly chars of the day.
I still think voxels are an interesting technology with great potential for future application in real time 3D. That said, I think these guys are the biggest fans of hyperbole EVER! :poly142: They are being way overly dramatic and vocal about what is ostensibly, a minor, niche and nascent technological breakthrough.
Gotta get that cash somehow though and they apparently think it'll happen by turning the volume up to "11".
My message to them is, pull back on the exaggeration and snide tone. Produce some more solid aesthetically pleasing examples (which should not in fact cost a lot more money than the crap sandwich they've already posted). Give us some downloadable tech demos to prove you aren't completely full of it, and please...dear god, get somebody else to narrate the videos. Listening to that guy makes me want to take an icepick to my eardrums...or his larynx.
EDIT: I also just noticed in his Description Video 2 at 3:15-4:15 he makes an entirely false claim. He quite obviously mistakes tessellation and displacement for normal mapping...which are not the same thing at all. His statement that "advanced displacement mapping does save the flower pot from looking like a hexagon!" is either him being just flat out wrong, or boldly deceptive as I'd imagine somebody developing technology of this type would, in fact, know better.
It makes sense, but to use it you need to be able to access all the millions and billions of pieces of data (digital atoms) at any given moment. Just like Google has indexed search information you would have to index these digital atoms.
I can see his point about doing away with the polygons though. To get a pixel off a polygon you need to reference all the points that make up that polygon. To get a pixel from a point cloud system you need to access just one point. There is no larger element "polygon" to break down further.
You would have to run game servers (even when your playing the game on your machine) just like Google does so that searches happen faster. That is unless you could store all the point cloud data on your computer and have it indexed efficiently.
I don't really see the point in being a fan boy for one side of the other as that always just seems to stems from ignorance... or the Nintendo vs Sega mentality of a 80s baby :P
Don't pick sides just be happy that people are always searching for new and better ways to do things. It does away with graphics cards by just using software rendering to access these models that are really just (guessing here) spherical information with a shader controlling their cumulative surfaces.
"Tim Sweeney, chief executive officer of Epic Games, said during his keynote at High Performance Graphics 2009 conference earlier this month that it is dramatically more expensive to develop software that relies on general purpose computing on graphics processing units (GPGPU) than to create a program that utilizes central processing units. He also re-iterated his earlier claims that the days of GPUs are counted." - Keynote speech High Performance Graphics 2009 conference
...
"Tim Sweeney, chief executive officer of Epic Games, said during his keynote at High Performance Graphics 2009 conference earlier this month that it is dramatically more expensive to develop software that relies on general purpose computing on graphics processing units (GPGPU) than to create a program that utilizes central processing units. He also re-iterated his earlier claims that the days of GPUs are counted." - Keynote speech High Performance Graphics 2009 conference
Yay for human ingenuity!
Ingenuity is people who do stuff, not people who have oral diarr
I know, I was talking more in general about the "this will change your LIFE!" kind of people, I know they want to sell their idea and get funding to do the real stuff, but it is a turn off.
edit: I did some googling, and it does seem that epic is partnering with intel, my tasteless jokes had some basis.
I mean: "$1 million intel make something unreal contest"
Carmack is also interested in voxel technology and sparse voxel octree's. He's mentioned about possibly storing geometry data as a 3D volume array of points or something and rendering "into that" whatever that means. Something about apparently extending the benefits of MegaTexture and 100% uniqueness to geometry as well as texture detail. Makes the real bottleneck storage capacity and the speed with which it is accessed, if your rendering speed is fixed and determined by the number of pixels you're writing to the display.
I'm looking forward to this... always had a niggling thought at the back of my mind that this is the way it should be done, especially for vegetation.
The next step could be just throwing algorithms at it. Imagine being able to create a scene just from a basic set of instructions about types of plants and terrain type? or DNA?
giga-voxels are also interesting and do pretty much the same thing but need huge amounts of ram (a demo I saw used 32gb ram)
and yes. Outcast was a great game. A great game that proved voxels were shit
If anyone would write a software-renderer and have it run crazy fast it would be tim sweeney, I give you that.
but still, gpu's aren't still around just due to random chance.
we're working in an industry where we constantly sacrifice productivity and time for performance, tim sweeney wants that to change around.
like in a perfect world we could:
skip out on uv-mapping.
render the mesh out directly from the sourcemesh.
dont ever worry about video memory around, just stream stuff from storage.
but still, developers tends to want to take shortcuts that are more complicated, to have a better looking game than the next studio.
I sent them an email asking a few technical questions, and just got this today:
"Thank you for your interest in Unlimited Detail,
We are at present building the commercial SDK and Demonstrations.
The planned form of the SDK is as follows:
IMPORT OF DATA:
Data can be imported from a variety of sources. We expect to directly support some popular scanning and modelling applications and file formats for both point cloud and polygon style data. The SDK will provide a simple way to import data from any source.
ENGINE:
The Unlimited Detail Engine displays 3D models or groups of models which may be positioned, rotated, scaled and viewed from any direction and with any projection. The technique allows geometry of unlimited complexity to be displayed at interactive frame rates without the need for 3D acceleration hardware.
ANIMATION:
Unlimited Detail supports Animation. (Please see attached video clip of an articulated bird. And remember this is a simple first attempt for R&D purposes so it doesn’t look “polished”) http://www.mediafire.com/file/ndumvromniq/bird_articulated.wmv
In addition to animation, we will be supporting dynamic and destructible environments.
MEMORY STORAGE
The Unlimited Detail rendering technology does not require significant system memory to run. The geometry may be compressed, instanced and streamed to allow the best mix of runtime and disk based storage as well as support unique geometry.
SHADOWS:
Unlimited Detail will have both real-time shadows and pre-calculated lighting as well as a variety of advanced lighting options.
DEMONSTRATIONS AND DOCUMENTATION:
Demonstrations will be available for a great variety of industries, mining, architecture, science, medical, games, and movies/entertainment. In all cases these demonstrations will use industry specific data and will be downloadable from the website.
PRICE :
Pricing will be chosen in such a way that this technology can be accessed by all, including those in developing nations were graphics cards are not commonplace and software based 3D graphics are vital to industry purposes.
MAILING LIST AND WEBSITE:
Nearer to completion we will be releasing short newsletters to inform interested parties of our progress, we will also have a new website in a few months as there have been comments about the inadequacy of the current website, whilst we do agree with the comments please understand we are primarily creators of technology and the recent media wave was unexpected. Please expect a more commercial incarnation of our public image soon enough.
Downloadable Demos will be available before the SDK release date
As interesting as this sounds, I'm still thinking more about real time ray tracing. I know that tech is at least a few years off, but then you'd have the eventual potential for Mental Ray quality images in real time. And yes, I also agree that it sounds like this makes use of really heavy instancing.
Besides...I like flat things like pictures on a plane. Especially pretty stylized pictures.
Arshlevon: That's what I found conspiciuosly absent, as well. Mind you, this is just something they mailed to everyone who showed interest, so I don't think they're replying to specific questions, just the most asked questions.
this probably got mentioned somewhere in the thread, but shame on me for not reading every comment, anyway:
not efficient!!
You'd have to spend so much time in making things that you couldve otherwise done faster with textures/shaders, this will make game development cost waaay more time and especially money (time = money and more money).
That and everything pointcloud that i've seen so far just doesn't look as good as the "flat ugly angular polygO4Nz!"
this probably got mentioned somewhere in the thread, but shame on me for not reading every comment, anyway:
not efficient!!
You'd have to spend so much time in making things that you couldve otherwise done faster with textures/shaders, this will make game development cost waaay more time and especially money (time = money and more money).
That and everything pointcloud that i've seen so far just doesn't look as good as the "flat ugly angular polygO4Nz!"
i dont see how it would necesarly take so much longer, since most games now are deriving your models from high poly originals seems like you could just convert the high poly to the point cloud and not need to bake any maps, make any low polys or anything, just make your high poly perhaps texture it with poly paint, and poof export to point cloud,
(disclaimer i just imagined that whole workflow. none of that is based on things) but in theory it could be real simple to get point cloud objects, even simpler than making really nice polygon normal maped objects
i'm still totaly skeptical but would be interested in seeing if this has any real meat an taters behind it.
What I think is quite strange as well is that there is no mention of lighting whatsoever, atm just shadows... what about realtime global illumination or hdrr or other fancy stuff?
i dont see how it would necesarly take so much longer, since most games now are deriving your models from high poly originals seems like you could just convert the high poly to the point cloud and not need to bake any maps, make any low polys or anything, just make your high poly perhaps texture it with poly paint, and poof export to point cloud,
(disclaimer i just imagined that whole workflow. none of that is based on things) but in theory it could be real simple to get point cloud objects, even simpler than making really nice polygon normal maped objects
i'm still totaly skeptical but would be interested in seeing if this has any real meat an taters behind it.
I was going to say the same thing. In fact, dartar boy has already mentioned parsing data from standard polygon models. To be a viable option at all, and fit in with existing workflows, it would HAVE to be able to reconstruct a polygonal model into point cloud dartar.
I am of course still skeptical about the whole thing, especially considering DS graphics are more visually appealing than the demos they've posted. The problem with voxels, or point cloud dartar, is they just look so bitty and kind of ugly. Angular or not, I love the look of polygonal graphics, it's the textures that make them, and with good textures polygon visuals look far better than any voxel esque graphics I've seen.
i dont see how it would necesarly take so much longer, since most games now are deriving your models from high poly originals seems like you could just convert the high poly to the point cloud and not need to bake any maps, make any low polys or anything, just make your high poly perhaps texture it with poly paint, and poof export to point cloud,
(disclaimer i just imagined that whole workflow. none of that is based on things) but in theory it could be real simple to get point cloud objects, even simpler than making really nice polygon normal maped objects
i'm still totaly skeptical but would be interested in seeing if this has any real meat an taters behind it.
Hoping to see voxel painting make it's way into 3dcoat soon myself. Storing all the detail and color info into the high rez makes a lot of sense to me.
Replies
And isn't a point an infinitely small thing as to have no dimension itself whatsoever?
So to be drawn a point must in fact become at least the simplest form of a polygon aka. triangle at the very least to be visible no matter the size?
I'd think for a super-cala-fradja-listic-expi-ali-docious programming team that figured their way around the current system would at least refer to it by a proper name or come up with a name for it which would more accurately describe it...
Now that I'm done blowing smoke out of my ass
If its real and it can be lit, animated, have physics applied, yadi yadi yadi, then awesome and I welcome the revoloution and all the knowledge that will come with it, if not...well cold fusion anyone?
I agree with the folks in the "cool tech, though not yet" wagon. I know people question "is this needed?" But, a good deal of the tech world has not really been concerned with what is needed, but rather question "is it better?".
It is pretty awesome to think of a 3d world that is composed of essentially simulated molecules. I remember thinking this idea was really keen a shit long time ago, and even posting the concept here (this was prior to my knowledge of voxels). However once we get to a point were things like screen resolution vs asset resolution aren't really an issue (like the aforementioned console problem), so voxel density would match screen resolution. Until then I can't imagine this being much of a standard, but in the future? Why not.
I would imagine if things like collision, physics, and all sorts of physical properties are properly paid attention to, you'd have much more convincing animations in characters and environments. Also like mentioned before, converting polygonal mesh to voxel is not hard. And I would imagine making texture would either be like a voxel paint deal (also mentioned) or alternatively a sort of render to texture, but rather render to voxel. In which you still texture it regularly with uvs, but then have the uv texture colors projected on to their corresponding voxels. But like I mentioned before about voxel density vs resolution, it would only look as good as the created texture if the voxel density matched the vexel density.
I would say that one of the things seems iffy regarding asset creation is that artists would be tasked with a shit ton more work, because now not only will you be responsible for modeling the facade of objects, but also any layer that would happen to exist within it. Let's say you wanna make a house that will be in a war game, and therefore be prone to explosion. Make the house, now make the interior, now make the infrastructure, now make the systems that happen to exist in a home (plumbing and electrical). Same goes for making a human in a kind of game where people are blown to bits, you'd have to learn not only the exterior of the being's body but all it's inner workings. Which I can only imagine being a tad disheartening to non-psychopaths to see someone's intestines spill in a much more realistic manner. Anyway, like I was saying this all leads to a workload that is much more intensive. As grim as it sounds, if this was going to be the new standard in the future, then I would imagine in order to have a job, you'll have to be knowledgeable of such things. So people who think it's hard to get job now as an artist, it may get much harder. Also for people whom are hoping to keep their job it may be harder to keep. Or on the other side of spectrum, this means more people will be employed to just do these substructures of meshes to be calculated into voxels. I do not know, perhaps this will all be incorporated as a gimmick for game and rarely heard from again, or it will indeed eventually become an industry standard. Either way, as an artist within this field I would certainly not disregard it as far as your practice goes.
Anyways. I am always excited for new tech to break through, since the cool side effect is that it always brings new ways of working. But on the other hand, claiming that it's gonna be the next big thing without anything to back it up ... is very pedantic from these guys. I feel as if the narrator on these videos never ever worked with polies and ditches them just like that. Try making a quick blockout model in voxels, hehe.
Because what else would you use infinite detail for?
I think a distinction should be made between IRL realism and realism in the sense of high detail. A dragon is not realistic (dragons don't exist for all we know), but with this tech you could make it look realistic (scaly skin, round tail instead of a low poly with baked normals). It's a drive for detail and this still requires more artistic prowess.
As for unrealistic games, like those artsy games with blocks and simple shapes, this tech won't give anything but waste resource by rendering a 6 sided cube as million point cluster. It's still a cube.
And in the case of not using procedural stuff and doing everything by hand, slowly, that's how it has to be done, in the cases where we cannot prebake that stuff, since we do not have UNLIMITED hardware, we want to do as much with as little as possible.
or do like red faction did, and put some of the budget towards having such things.
voxels has had its time in gaming, but polygons just always became easier to use and with more accurate results.
voxels did have some interesting points of use though:
Character creation of today:
I for one like the highpoly process of making a game asset, but dislikes the lowpoly bit.
A voxel technology would take care of that and i wouldn't have to touch a farking lowpoly model ever again and all that's involved in that process (UVing, Smoothing groups, lod steps, test baking normals etc)
Id rather just make my highpoly model, hit "MakeVoxelModel.exe" and import it into the game...DONE!
And since you can avoid the lowpoly pipeline completely you can spend a few extra days on polishing the highpoly model.
Also, some models are pretty much impossible to make with a low polycount, which limits the designs a bit, or you have to find workarounds. That would also be a thing of the past, you can just go all out on a model and the engine will just handle it.
in a nutshell, game art quality would go up in general = win
There's a part 2 as well (sorry if it has been posted before).
This really brings into light more of their situation and may answer some of the questions that have popped up here about it.
Somebody strangle hiiiiiiim!!
"For some strange reason these wankers keep punching me in the throat after every pitch meeting..."
I still think voxels are an interesting technology with great potential for future application in real time 3D. That said, I think these guys are the biggest fans of hyperbole EVER! :poly142: They are being way overly dramatic and vocal about what is ostensibly, a minor, niche and nascent technological breakthrough.
Gotta get that cash somehow though and they apparently think it'll happen by turning the volume up to "11".
My message to them is, pull back on the exaggeration and snide tone. Produce some more solid aesthetically pleasing examples (which should not in fact cost a lot more money than the crap sandwich they've already posted). Give us some downloadable tech demos to prove you aren't completely full of it, and please...dear god, get somebody else to narrate the videos. Listening to that guy makes me want to take an icepick to my eardrums...or his larynx.
EDIT: I also just noticed in his Description Video 2 at 3:15-4:15 he makes an entirely false claim. He quite obviously mistakes tessellation and displacement for normal mapping...which are not the same thing at all. His statement that "advanced displacement mapping does save the flower pot from looking like a hexagon!" is either him being just flat out wrong, or boldly deceptive as I'd imagine somebody developing technology of this type would, in fact, know better.
I can see his point about doing away with the polygons though. To get a pixel off a polygon you need to reference all the points that make up that polygon. To get a pixel from a point cloud system you need to access just one point. There is no larger element "polygon" to break down further.
You would have to run game servers (even when your playing the game on your machine) just like Google does so that searches happen faster. That is unless you could store all the point cloud data on your computer and have it indexed efficiently.
I don't really see the point in being a fan boy for one side of the other as that always just seems to stems from ignorance... or the Nintendo vs Sega mentality of a 80s baby :P
Don't pick sides just be happy that people are always searching for new and better ways to do things. It does away with graphics cards by just using software rendering to access these models that are really just (guessing here) spherical information with a shader controlling their cumulative surfaces.
"Tim Sweeney, chief executive officer of Epic Games, said during his keynote at High Performance Graphics 2009 conference earlier this month that it is dramatically more expensive to develop software that relies on general purpose computing on graphics processing units (GPGPU) than to create a program that utilizes central processing units. He also re-iterated his earlier claims that the days of GPUs are counted." - Keynote speech High Performance Graphics 2009 conference
Yay for human ingenuity!
Ingenuity is people who do stuff, not people who have oral diarr
edit: I did some googling, and it does seem that epic is partnering with intel, my tasteless jokes had some basis.
I mean: "$1 million intel make something unreal contest"
pretty subtle.
The next step could be just throwing algorithms at it. Imagine being able to create a scene just from a basic set of instructions about types of plants and terrain type? or DNA?
giga-voxels are also interesting and do pretty much the same thing but need huge amounts of ram (a demo I saw used 32gb ram)
and yes. Outcast was a great game. A great game that proved voxels were shit
but still, gpu's aren't still around just due to random chance.
we're working in an industry where we constantly sacrifice productivity and time for performance, tim sweeney wants that to change around.
like in a perfect world we could:
skip out on uv-mapping.
render the mesh out directly from the sourcemesh.
dont ever worry about video memory around, just stream stuff from storage.
but still, developers tends to want to take shortcuts that are more complicated, to have a better looking game than the next studio.
"Thank you for your interest in Unlimited Detail,
We are at present building the commercial SDK and Demonstrations.
The planned form of the SDK is as follows:
IMPORT OF DATA:
Data can be imported from a variety of sources. We expect to directly support some popular scanning and modelling applications and file formats for both point cloud and polygon style data. The SDK will provide a simple way to import data from any source.
ENGINE:
The Unlimited Detail Engine displays 3D models or groups of models which may be positioned, rotated, scaled and viewed from any direction and with any projection. The technique allows geometry of unlimited complexity to be displayed at interactive frame rates without the need for 3D acceleration hardware.
ANIMATION:
Unlimited Detail supports Animation. (Please see attached video clip of an articulated bird. And remember this is a simple first attempt for R&D purposes so it doesn’t look “polished”)
http://www.mediafire.com/file/ndumvromniq/bird_articulated.wmv
In addition to animation, we will be supporting dynamic and destructible environments.
MEMORY STORAGE
The Unlimited Detail rendering technology does not require significant system memory to run. The geometry may be compressed, instanced and streamed to allow the best mix of runtime and disk based storage as well as support unique geometry.
SHADOWS:
Unlimited Detail will have both real-time shadows and pre-calculated lighting as well as a variety of advanced lighting options.
DEMONSTRATIONS AND DOCUMENTATION:
Demonstrations will be available for a great variety of industries, mining, architecture, science, medical, games, and movies/entertainment. In all cases these demonstrations will use industry specific data and will be downloadable from the website.
PRICE :
Pricing will be chosen in such a way that this technology can be accessed by all, including those in developing nations were graphics cards are not commonplace and software based 3D graphics are vital to industry purposes.
MAILING LIST AND WEBSITE:
Nearer to completion we will be releasing short newsletters to inform interested parties of our progress, we will also have a new website in a few months as there have been comments about the inadequacy of the current website, whilst we do agree with the comments please understand we are primarily creators of technology and the recent media wave was unexpected. Please expect a more commercial incarnation of our public image soon enough.
Downloadable Demos will be available before the SDK release date
Thank you for your interest.
Kindest Regards
Bruce Robert Dell"
Besides...I like flat things like pictures on a plane. Especially pretty stylized pictures.
not efficient!!
You'd have to spend so much time in making things that you couldve otherwise done faster with textures/shaders, this will make game development cost waaay more time and especially money (time = money and more money).
That and everything pointcloud that i've seen so far just doesn't look as good as the "flat ugly angular polygO4Nz!"
i dont see how it would necesarly take so much longer, since most games now are deriving your models from high poly originals seems like you could just convert the high poly to the point cloud and not need to bake any maps, make any low polys or anything, just make your high poly perhaps texture it with poly paint, and poof export to point cloud,
(disclaimer i just imagined that whole workflow. none of that is based on things) but in theory it could be real simple to get point cloud objects, even simpler than making really nice polygon normal maped objects
i'm still totaly skeptical but would be interested in seeing if this has any real meat an taters behind it.
I was going to say the same thing. In fact, dartar boy has already mentioned parsing data from standard polygon models. To be a viable option at all, and fit in with existing workflows, it would HAVE to be able to reconstruct a polygonal model into point cloud dartar.
I am of course still skeptical about the whole thing, especially considering DS graphics are more visually appealing than the demos they've posted. The problem with voxels, or point cloud dartar, is they just look so bitty and kind of ugly. Angular or not, I love the look of polygonal graphics, it's the textures that make them, and with good textures polygon visuals look far better than any voxel esque graphics I've seen.
Hoping to see voxel painting make it's way into 3dcoat soon myself. Storing all the detail and color info into the high rez makes a lot of sense to me.