the models won't change much. maybe some better tools to get faster workflow, even if i think that most of the tools are pretty good already.(just like photoshop...more tools, but true improvement is hard to find )
the real push will come from better realtime lightning, like realtime ao. this is of course nothing an artist can change. more performance and better code will do this.
for the artist, this means less work, as the engine will do more jobs we have to fake now. like normal mapping, shadow maps, etc..
Force feedback gloves, or mind controlled workflow. (remember they already have the ability to make a mouse move on screen by training people to stimulate certain parts of the brain though a EM? interpretor).
Think of what it must have been like ten years ago. Think of what it is now. Think of what it will be like ten years from now.
That's kinda sad to think about. Besides sculpting there's not much new other than most packages pilfering ideas from others and thus becoming more complete.
I'm gonna go ahead and disagree with everyone and say that the days of high poly, realistic, 3d modeling for games are coming to an end. It's just too expensive and time consuming.
I think web based, flash-style tools will become more and more popular.
In the year 3000 people will rediscover the SNES and worship it in massive temples in the shape of boobs made of glass.
I think artists needs better interfaces to manipulate/interact with digital sculptures (gloves, glasses, 3D monitors, holograms) and (I hope) a seamless technology that can print/transform bytes in real materials.
Anyway. Can someone have a good explain about Voxels? I really don´t know it (pros/cons) and the subject looks very popular nowadays.
Voxel is a combination of the words volume and pixel. It is essentially a 3D pixel. So instead of your 3D model being made of many polygons it is instead made of many boxes that dynamically re-sized and re-position themselves based on how near or far you are from areas of detail.
They have been used in the game industry spuraticaly over the years, as well as other industries such as medical visualization.
IDtech 6 may include support for voxel octrees which is just a fast way of dealing with voxels here is an example:
Uh, voxels have been around in games for awhile. My earliest memory of them is Command & Conquer: Tiberium Sun. If they were going to explode on to the scene, they would have by now. IMO.
I think the future of 3D modeling will probably be the manipulation and management of real world data. As data capture methods increase we will see realistic and viable real time 3D data being used in things like simulations and games. So someone might go out and just scan something like a rock rather than going through the effort of dealing with all the details by hand. As a result artists in some cases may become managers and selectors of real world props and objects rather than creating them from scratch themselves.
I think a lot of art will be pushed more into the fantasy realm just like what happen when photography came out. There wasn't as much need to simply re-produce reality so it encouraged high concept interpretative movements, and we may see the same thing happen again.
Uh, voxels have been around in games for awhile. My earliest memory of them is Command & Conquer: Tiberium Sun. If they were going to explode on to the scene, they would have by now. IMO.
Actually voxels can be used in many ways. The new buzz is about using lots and lots of them to create full blown high detail 3D models and then stream the data in off very large hard drive. This technique wasn't possibly in the past due to a lack of computing power and storage. So voxels time may have yet to come.
Just like at movies/fx to see where games will go. More lighting tech, realtime raytracing, more hdr, glossy reflections, more antialiasing and better filtering, more post effects and compositing in real-time, higher poly/texture limits.
I thought everyone was joking about how everyone went "OMG VOXELS!" when they first came out. I remember everyone made a big deal about it with some sandboxy sci-fi game. (a buncha people posted while I was writing this, making me look like a jerk)
Maybe real displacement mapping on models in 'teh futurez'?
Uh, voxels have been around in games for awhile. My earliest memory of them is Command & Conquer: Tiberium Sun. If they were going to explode on to the scene, they would have by now. IMO.
The reason why they haven't is largely because of the graphic cards arms race. Traditional graphics accelerators have been a staple of the PC gaming and console markets for over a decade. And the rendering pipeline that traditional graphic accelerators all support is targeted at polygonal rendering.
But there has been a lot of talk recently about a major shift in rendering due to the continued expansion of mutli-threaded processing. Multi-core processors open rendering up to less traditional approaches. In time it will likely be possible for rendering to be taken away from the GPU and handled entirely by the CPU. This is especially attractive for laptop systems. (where the option of removing extra chips is very desireable) One of the rendering techniques that could do very well with multi-core rendering is voxels.
As for model editing, voxels are obviously very attractive already. Scalable detail based on what you are looking at is the near-ideal solution for model sculpting. You can zoom into your model and sculpt extremely fine detial with almost no hit to performance. Then zoom out and view the entire model, with the detail automatically scaling, again, no hit to performance. The only real issue is when you attempt to export the voxel model to polygons.
first experienced voxels in comanche, that helicopter game from 92. damn those graphics were so superior to other games then there was outcast a couple years later, haven't seen them often since then. or at least not noticed. i think the problem with voxels is that there is no easy way to animate them so they've been used for terrains pretty much? but i have seen a couple nice tech demos based on voxels recently.
I think we're finally getting modeling under control but what really needs a new revolutionary update would be animation. Unlike the near constant modeling tool updates and brand new packages that make huge strides forward, animation hasn't really progressed at the same level. Probably because real-time lags behind what the tools can already do.
I agree with Vig. Modeling has progressed far, far beyond what it used to be. For less than $1000, you can buy a sculpting software package. And for another $1000 (or less) you can assemble a computer to run it on. (with acceptable performance) All of that will give you what you need to sculpt incredibly detailed models. Some of the things that are being done with 3D modeling right now actually rival traditional sculpting mediums. Aside from refined usability in interfaces, we are fast approaching diminishing returns.
But there is still a lot more opportunity for expanding on animation techniques, both in real-time and pre-rendered. More effective auto-rigging, complex physics interaction, deformable meshes, the list goes on and on. There are already a lot of companies looking into these areas, and I wouldn't be surprised to see a much stronger emphasis on this kind of technology in the coming years.
I'm guessing the graphic card manufacturers don't like the idea
Definitely not. Although if I'm not mistaken, weren't the major GPU manufactuers purchased by larger CPU manufacturers. I know ATI was gobbled up by AMD. (I have to go to AMD's website to upgrade my graphics drivers now) It could be that someone saw the writing on the wall.
It's still going to be a while anyway. Software engineers still haven't gotten used to the idea of multi-thread development. It requires a wholly different programming methodology. Right now C is the best language for multi-platform multi-thread development. So a lot of C coders are being employed to explore these alternatives. You can be certain that Intel and IBM are going to be keen on this sort of advanced software rendering.
first experienced voxels in comanche, that helicopter game from 92. damn those graphics were so superior to other games then there was outcast a couple years later, haven't seen them often since then. or at least not noticed. i think the problem with voxels is that there is no easy way to animate them so they've been used for terrains pretty much? but i have seen a couple nice tech demos based on voxels recently.
i imagine voxel animation is more like stop motion, every animation frame is stored as a voxel object, like they did in that blade runner game.
No more Zbrushing folds and wrinkles of clothing. You just make a shirt, tesselate it to shit, and let physics do all the folding and wrinkling.
2) Normal Maps use diminished.
No more will we be sculpting wrinkles in Zbrush. We will instead be modelling every single wrinkle, so they will fold and crease naturally with the animations. Normal Maps will instead be used for micro details like pores, and clothing textures.
No more will we be sculpting wrinkles in Zbrush. We will instead be modelling every single wrinkle, so they will fold and crease naturally with the animations. Normal Maps will instead be used for micro details like pores, and clothing textures.
sounds liek a really inefficient way to do it no? How is it done in film now, dont they just blend between displacement maps for this stuff?
I kinda see this work on some stuff but far from everything... I need to see it in action on a wide variety of objects i guess, not just some blobby sculpts
automatic/procedural/dynamic low poly generation on the fly using material volumes to accuratly simulate material degradation, this is also how you would sculpt, want a stone statue thats been weathered, just apply a granite material volume and get your hard rain brush out
so what's holding back real time tesselation and displacement? is it GPU power?
I'm guessing it would be the same workflow, you build a high poly and a low poly in game mesh but instead of baking out a normal map you'd bake out a displacement map and the mesh gets subdivided at render (possibly the number of subdivisions could be based on performance and/or distance from the player)?
Granted I have no idea how all that programming stuff in the background works so tell me if I'm off my rocker.
Justin, the problem with displacement is just the number of polys that have to be rendered. Subdividing a mesh at render time is still going to result in a lot of polys. The more detail you want, the more polys will have to be genarated. And the number of polygons in a scene is still a major bottleneck for how quickly a scene can be rendered.
The whole point of normal maps is that "detail" is added through the way in which lighting is rendered on a low-poly model. It is a cheat, and is much more performance friendly than displacement rendering. With displacement, hundreds of thousands of polys would be required for the lowest detail settings of a sculpted mesh. And you couldn't be certain of how well those polys handled things like creases.
so you don't think it's a possibility in a future where people play "classics" like Crysis on their mobile phone?
We won't be calling them phones in the future. It's stupid that we call them phones, now. Think about all the things 'cel phones' are used for today: camera, text, web-browsing, gps, games, the list goes on. I'm sure their are some people for whom talking on a cel is the thing the do the least with it.
We won't be calling them phones in the future. It's stupid that we call them phones, now. Think about all the things 'cel phones' are used for today: camera, text, web-browsing, gps, games, the list goes on. I'm sure their are some people for whom talking on a cel is the thing the do the least with it.
Seriously, though, how long are we going to go on calling them phones? It's like calling your house your 'bed shelter'. Of course you sleep there, but that's not it's only use.
The near future could be easily subdiv models + normal maps + displacements + GI real time renderers + realistic shaders, so with modelling, i imagine better tools to "draw" polygons to make subdiv models.
The modelling tools are improving so fast, and a good example is now with modo 401. Tools can be better, but at the end, what makes the difference is the human. Can you evolve? hahaha
To quote a lecturer at GDC : 'Carmack is a false god'. The quality comes from the artists, not the fancy tech. So yeah I don't think we should hope too much for crazy realtime folds and such (if it was true, IRL sculptors would just throw a plastered towel on their sculpts and call it done). Also voxels are I think the next thing for highpoly creation, but not for the actual ingame asset. I don't think we'll see anything more than polygons, patches, particules and geometry/pixel shaders in a 3D game anytime soon. It covers a LOT already!
It reminds me of all these cool conversations you hear between technical guys/reps sometimes, like yeah don't worry the engine/app can do this this and that. In the end we still clean un stuff in PS, or do unexpected color adjustements to textures to make them work. So yeah my hope for the future is better sculpting and painting tools, and better retopo tools. These alone need tons of improvement still. On the game side itself it's cool as is, well maybe ID will manage to virtualize geometry even further but it wouldn't impact asset creation too much I think.
Also I want more ultra lowpoly games with no 3D shading, just crazy good texture reuse and handpainted shading. WITH antialiasing.
Whether the use of voxels will become more widespread or not, I'm surprised so many game-artists scoff at the very idea. To quote Adam: "If they were going to explode on to the scene, they would have by now"
They've always been fully dependant on cpu, not gpu. I remember one game --Outcast-- coming out when 3d cards had become standard equipment, and it tanked. Not because it was a poor game, but because it ran incredibly slow on most people's hardware, and their shiny new Voodoo-card didn't do anything to change that. Bit of an over-simplification, ofcourse.
The video that was posted earlier in this thread doesn't really tell us much about how fast/viable the technology is, but as far as I can tell we've come to a point where anything can be run (albeit with some delay, in my understanding) on the gpu thanks to things like Cuda, so it should at least be possible.
sounds liek a really inefficient way to do it no? How is it done in film now, dont they just blend between displacement maps for this stuff?
They might to some degree. But having wrinkles 'cross-dissolve' in looks a LOT different than wrinkles actually forming due to a mass of skin cramping and stretching to form them.
I personally believe the progression of video game art is greatly dependant on immersion more than anything.
Painted textures DO look amazing, and as a artists we can truly appreciate the craft behind it. But PS360 generation art is more about more believable lighting and surface properties. Thus the artists now have to work with Normal Maps, and Spec maps for more detailed lighting, and material properties.
I think the 'future' of 3D art will generally be things that increase quality in 'motion, and rendering', thus the modellers will have to follow suit.
The reasons I suggested per-poly collision as a way of clothing in the future, is because Rubbery Normal Mapped clothing looks like complete trash in motion. Having a fold in place stay in stationary place completely breaks the believability of cloth and skin. It might not be the greatest piece of artistry to let the physics do it, but it certainly LOOKS much more natural.
God knows how many times weve had to watch Solid Snake, or Marcus Fenix 'angry eyes' brow wrinkles make them look from VERY VERY angry to, VERY angry, to Sad Angry, to Angrily laughing.
Skin wrinkles will likely have to have a much better solution (hence my 'modelled wrinkles' quasi-solution) if we want the animators to TRULY be able to act with more engaging ranges of emotion and higher quality of lip sync.
Heck, if we DO go with a rendering solution for REAL SSS, then we might soon have to paint 5-6 textures for flesh.
Who knows, maybe I'm out to lunch, but I really think there's other greater factors that break immersion in games than low poly counts, and smaller texture resolutions.
Replies
Can't believe they are releasing yet another freebie.
zomg how the hell do i get more info on zbrush 4???????????
the real push will come from better realtime lightning, like realtime ao. this is of course nothing an artist can change. more performance and better code will do this.
for the artist, this means less work, as the engine will do more jobs we have to fake now. like normal mapping, shadow maps, etc..
+1 Voxels as well
That's kinda sad to think about. Besides sculpting there's not much new other than most packages pilfering ideas from others and thus becoming more complete.
I think web based, flash-style tools will become more and more popular.
In the year 3000 people will rediscover the SNES and worship it in massive temples in the shape of boobs made of glass.
Or, there will just be a "Make Art" button.
Voxel is a combination of the words volume and pixel. It is essentially a 3D pixel. So instead of your 3D model being made of many polygons it is instead made of many boxes that dynamically re-sized and re-position themselves based on how near or far you are from areas of detail.
They have been used in the game industry spuraticaly over the years, as well as other industries such as medical visualization.
IDtech 6 may include support for voxel octrees which is just a fast way of dealing with voxels here is an example:
[ame]
And here is the wikipedia page if you want to find out more:
http://en.wikipedia.org/wiki/Voxel
What this means for 3D modeling is theory infinite detail. Since the topology is streamed in and not interconnected in the same way that polygons are.
I think a lot of art will be pushed more into the fantasy realm just like what happen when photography came out. There wasn't as much need to simply re-produce reality so it encouraged high concept interpretative movements, and we may see the same thing happen again.
Actually voxels can be used in many ways. The new buzz is about using lots and lots of them to create full blown high detail 3D models and then stream the data in off very large hard drive. This technique wasn't possibly in the past due to a lack of computing power and storage. So voxels time may have yet to come.
Maybe real displacement mapping on models in 'teh futurez'?
The reason why they haven't is largely because of the graphic cards arms race. Traditional graphics accelerators have been a staple of the PC gaming and console markets for over a decade. And the rendering pipeline that traditional graphic accelerators all support is targeted at polygonal rendering.
But there has been a lot of talk recently about a major shift in rendering due to the continued expansion of mutli-threaded processing. Multi-core processors open rendering up to less traditional approaches. In time it will likely be possible for rendering to be taken away from the GPU and handled entirely by the CPU. This is especially attractive for laptop systems. (where the option of removing extra chips is very desireable) One of the rendering techniques that could do very well with multi-core rendering is voxels.
As for model editing, voxels are obviously very attractive already. Scalable detail based on what you are looking at is the near-ideal solution for model sculpting. You can zoom into your model and sculpt extremely fine detial with almost no hit to performance. Then zoom out and view the entire model, with the detail automatically scaling, again, no hit to performance. The only real issue is when you attempt to export the voxel model to polygons.
But there is still a lot more opportunity for expanding on animation techniques, both in real-time and pre-rendered. More effective auto-rigging, complex physics interaction, deformable meshes, the list goes on and on. There are already a lot of companies looking into these areas, and I wouldn't be surprised to see a much stronger emphasis on this kind of technology in the coming years.
Definitely not. Although if I'm not mistaken, weren't the major GPU manufactuers purchased by larger CPU manufacturers. I know ATI was gobbled up by AMD. (I have to go to AMD's website to upgrade my graphics drivers now) It could be that someone saw the writing on the wall.
It's still going to be a while anyway. Software engineers still haven't gotten used to the idea of multi-thread development. It requires a wholly different programming methodology. Right now C is the best language for multi-platform multi-thread development. So a lot of C coders are being employed to explore these alternatives. You can be certain that Intel and IBM are going to be keen on this sort of advanced software rendering.
Also, other advanced input methods:
its mad crazy. Something like that needs zbrush support stat
VOXELS lol weren't they all the rage years ago too?
I'd like to see real-time displacement taking the popularity of normal mapping, and good lighting systems throughout the board
i imagine voxel animation is more like stop motion, every animation frame is stored as a voxel object, like they did in that blade runner game.
The only reason I don't have one is the lack of support from apps.
Others possible future things:
1) Per-Poly Collision
No more Zbrushing folds and wrinkles of clothing. You just make a shirt, tesselate it to shit, and let physics do all the folding and wrinkling.
2) Normal Maps use diminished.
No more will we be sculpting wrinkles in Zbrush. We will instead be modelling every single wrinkle, so they will fold and crease naturally with the animations. Normal Maps will instead be used for micro details like pores, and clothing textures.
3) RealTime Mental Ray Quality Rendering.
Something like Lpics:
http://graphics.pixar.com/library/Lpics/paper.pdf
4) Props and environment objects will all be made using geometry maps:
http://www3.ntu.edu.sg/home/asprakash/games2004/news/An_Overview_of_Geometry_Maps.pdf
I don't think I can process that
jox, thanks for those links, very interesting stuff and its nice to see complex ideas broken down so simply.
sounds liek a really inefficient way to do it no? How is it done in film now, dont they just blend between displacement maps for this stuff?
I kinda see this work on some stuff but far from everything... I need to see it in action on a wide variety of objects i guess, not just some blobby sculpts
I'm guessing it would be the same workflow, you build a high poly and a low poly in game mesh but instead of baking out a normal map you'd bake out a displacement map and the mesh gets subdivided at render (possibly the number of subdivisions could be based on performance and/or distance from the player)?
Granted I have no idea how all that programming stuff in the background works so tell me if I'm off my rocker.
The whole point of normal maps is that "detail" is added through the way in which lighting is rendered on a low-poly model. It is a cheat, and is much more performance friendly than displacement rendering. With displacement, hundreds of thousands of polys would be required for the lowest detail settings of a sculpted mesh. And you couldn't be certain of how well those polys handled things like creases.
We won't be calling them phones in the future. It's stupid that we call them phones, now. Think about all the things 'cel phones' are used for today: camera, text, web-browsing, gps, games, the list goes on. I'm sure their are some people for whom talking on a cel is the thing the do the least with it.
Can I call mine a tricorder?
I think that's copyrighted.
Seriously, though, how long are we going to go on calling them phones? It's like calling your house your 'bed shelter'. Of course you sleep there, but that's not it's only use.
I want a new word for cel phones.
The modelling tools are improving so fast, and a good example is now with modo 401. Tools can be better, but at the end, what makes the difference is the human. Can you evolve? hahaha
It reminds me of all these cool conversations you hear between technical guys/reps sometimes, like yeah don't worry the engine/app can do this this and that. In the end we still clean un stuff in PS, or do unexpected color adjustements to textures to make them work. So yeah my hope for the future is better sculpting and painting tools, and better retopo tools. These alone need tons of improvement still. On the game side itself it's cool as is, well maybe ID will manage to virtualize geometry even further but it wouldn't impact asset creation too much I think.
Also I want more ultra lowpoly games with no 3D shading, just crazy good texture reuse and handpainted shading. WITH antialiasing.
They've always been fully dependant on cpu, not gpu. I remember one game --Outcast-- coming out when 3d cards had become standard equipment, and it tanked. Not because it was a poor game, but because it ran incredibly slow on most people's hardware, and their shiny new Voodoo-card didn't do anything to change that. Bit of an over-simplification, ofcourse.
The video that was posted earlier in this thread doesn't really tell us much about how fast/viable the technology is, but as far as I can tell we've come to a point where anything can be run (albeit with some delay, in my understanding) on the gpu thanks to things like Cuda, so it should at least be possible.
Here is a example of someone using eye tracking on a desktop. There are actually a lot of experiments out there like this, pretty cool stuff.
[ame]http://video.google.com/videoplay?docid=-5430018385346925426&ei=NF3vSd60HoHg-wH0yqDHBQ&q=eye+tracking+solution&hl=en&client=firefox-a[/ame]
Amen.
B
They might to some degree. But having wrinkles 'cross-dissolve' in looks a LOT different than wrinkles actually forming due to a mass of skin cramping and stretching to form them.
I personally believe the progression of video game art is greatly dependant on immersion more than anything.
Painted textures DO look amazing, and as a artists we can truly appreciate the craft behind it. But PS360 generation art is more about more believable lighting and surface properties. Thus the artists now have to work with Normal Maps, and Spec maps for more detailed lighting, and material properties.
I think the 'future' of 3D art will generally be things that increase quality in 'motion, and rendering', thus the modellers will have to follow suit.
The reasons I suggested per-poly collision as a way of clothing in the future, is because Rubbery Normal Mapped clothing looks like complete trash in motion. Having a fold in place stay in stationary place completely breaks the believability of cloth and skin. It might not be the greatest piece of artistry to let the physics do it, but it certainly LOOKS much more natural.
God knows how many times weve had to watch Solid Snake, or Marcus Fenix 'angry eyes' brow wrinkles make them look from VERY VERY angry to, VERY angry, to Sad Angry, to Angrily laughing.
Skin wrinkles will likely have to have a much better solution (hence my 'modelled wrinkles' quasi-solution) if we want the animators to TRULY be able to act with more engaging ranges of emotion and higher quality of lip sync.
Heck, if we DO go with a rendering solution for REAL SSS, then we might soon have to paint 5-6 textures for flesh.
Who knows, maybe I'm out to lunch, but I really think there's other greater factors that break immersion in games than low poly counts, and smaller texture resolutions.