Maybe Ageia isnt the bottleneck

[kraemer007]

So lets supposed I want to have a fully physics based world. This means I would have to be able to tessellate and deform all the geometry in the game world. Is this even possible with current or next generation 3d hardware? Think about it, this is why grenades dont work right in games. They would literally have to have the ability to "damage" and deform ANYTHING in the game.

The same goes for fluids. Sure, the Ageia can do the computations for fluid dynamics, but can the videocard really handle all the polygons needed for a big splashy blob of flowing water whose outer surface is tesselated a zillion times a second?

Before anybody goes lynching Ageia I think they are going to expose a lot of weaknesses in the way we generate 3d objects. I can almost smell the resurgence of the voxel!!!

 

[Jacob Freeman]

Since we have seen working demos of fluid dynamics, of course current video cards can do it. I have seen fluid demos on Havok phyiscs as well.

 

[Langford]

I get the feeling it will be many years before physics in games become realistic. All I keep hearing about is how it is being implemented for graphical effects, but not much in the way of anything thatt effects gameplay. Life-like physics in games still appears to be little more than an illusion, even if the illusion has improved. it's still an important stage in development though.

What it seems like, is that in order for a game to have actual realistic physical gameplay, it would need to be written with a bunch of specific properties. First off, it would need to be able to alter the rendered objects in the game, to the extent that it would have to deform or create new unique-objects as a reaction to stimulie. If these objects are to remain hollow wire-frames, it will need some textures to account for the interior when the parent object is busted up. These new objects would need to inherit the correct properties from the original object, as well as new properties of it's own. Pretty complicated coding would need to go into the game engine.

For the eventual future, what would probably be the ultimate solution is to escape from all things pre-rendered. No textures by way of bitmaps. Games would need to have some sort of "virtual matter" in which they would be colored-in automatically by way of the properties the objects were made of. Sound effects would probably also be created in this manner. Instead of having a wire-frame of a car with a bitmat on it for the actual object, a mold would have to be made for each component of the car in order to define where virtual particles would exist, and then each component's matter would have to be decided from the list of available matter types. An engine block would start as a specific shape of a matter that is given metal-like properties, and the tires would have rubber properties. As in real life, entire objects would have to be assembled from parts. Physical reactions would have to be based on the properties of those objects. The computer power required for anything like this to be done in real-time is probably many many many years away. At first it will probably be unsatisfyingly simple constructs, with a cell-shaded kind of look to it. Eventually as systems supported greater numbers of objects and supported smaller particles, the envirnments would surpass those of today, just as rendered games today surpass older sprite based games from before.

 

[kraemer007]

Games would need to have some sort of "virtual matter" in which they would be colored-in automatically by way of the properties the objects were made of.

Yes, my point exactly. These already exist. They are called voxels.

Oh, and just in case anybody thinks its impossible: At a past Quakecon Carmack told everybody he had a working Voxel renderer for Quake2.

 

[Langford]

Yes, my point exactly. These already exist. They are called voxels.

Oh, and just in case anybody thinks its impossible: At a past Quakecon Carmack told everybody he had a working Voxel renderer for Quake2.

I think voxels is pretty much on the money, becasue it more of less cannot be destroyed within the runtime enviornment. An eventual use for them is to just have a bunch of basic shapes that stick together by way of the game-matter physics, instead of defining a whole limb with one voxel. Kinda like a sculpture in oatmeal. Then when you shoot the opponent, chunks of him would break off, and other parts of him would be structurally weakend. Not only would buildings break into bricks, the bricks would break into gravel, and the gravel into dust, becasue it was always just a sculpture in dust to begin with. The trouble right now for games is processor-power vs presentation. Someone needs to step up and make an simple ugly game with good physics, just to have the groundwork in place. If they are making a Quake engine with that kind of stuff, I want to try it out, becasue it would be cool.

 

[Sly]

I think voxels is pretty much on the money, becasue it more of less cannot be destroyed within the runtime enviornment. An eventual use for them is to just have a bunch of basic shapes that stick together by way of the game-matter physics, instead of defining a whole limb with one voxel. Kinda like a sculpture in oatmeal. Then when you shoot the opponent, chunks of him would break off, and other parts of him would be structurally weakend. Not only would buildings break into bricks, the bricks would break into gravel, and the gravel into dust, becasue it was always just a sculpture in dust to begin with. The trouble right now for games is processor-power vs presentation. Someone needs to step up and make an simple ugly game with good physics, just to have the groundwork in place. If they are making a Quake engine with that kind of stuff, I want to try it out, becasue it would be cool.

They would need hardware support first. One of the developers interviewed remarked that the lack of hardware support hampers their game development. As it is, any kind of voxel support would have to be software driven since they aren't compatible with the current polygon based cards.

When i played comanche, the terrain was smoother than any other polygon based flight sim at the time. Think about it, if you can get voxel development to continue, you'll get an entire world with no sharp edges.

 

[kraemer007]

They would need hardware support first.

Suprise, suprise. There IS hardware Voxel acceleration. How long until nvidia buys these guys???
Volume graphics Inc and info on Voxels About volume graphics rendering

 

[nhusby]

my understanding of voxels is limited, but it appears to me that it is merely object oriented programming in the most literal way. every object is made of smaller objects and inherits most of its properties, while over-riding others.
Voxels in the game world would technically start at the cell level (as chemistry and nuclear physics are not necisary for the game level of simulation [yet.]) , but that would be rediculus in terms of modeling, programming, and processing. It would make more sence to start at a unit level starting at a grain of sand, and maybe as large as a brick, dependant upon how much processing power is available, and how breakable an item is. if necisary a larger unit such as a brick could be converted to smaller units like sand when destruction is necisary.

I amagine if an engine was developed from scratch such models would be possible with current hardware. each voxel would be a polygonal 3-D object with a single color bitmap. Each voxel would be tied together by the physics engine (based on bond strength of voxels) and each voxel would then have to be drawn by the GPU.

sorry I got in a bit of a hurry to go about halfway through this post... laterz

 

[Langford]

my understanding of voxels is limited, but it appears to me that it is merely object oriented programming in the most literal way. every object is made of smaller objects and inherits most of its properties, while over-riding others.
Voxels in the game world would technically start at the cell level (as chemistry and nuclear physics are not necisary for the game level of simulation [yet.]) , but that would be rediculus in terms of modeling, programming, and processing. It would make more sence to start at a unit level starting at a grain of sand, and maybe as large as a brick, dependant upon how much processing power is available, and how breakable an item is. if necisary a larger unit such as a brick could be converted to smaller units like sand when destruction is necisary.

I amagine if an engine was developed from scratch such models would be possible with current hardware. each voxel would be a polygonal 3-D object with a single color bitmap. Each voxel would be tied together by the physics engine (based on bond strength of voxels) and each voxel would then have to be drawn by the GPU.

sorry I got in a bit of a hurry to go about halfway through this post... laterz

Yeah, like you said, and to be good it would take a lot of processor, plus in multiplayer you have to transfer all changes over the network. It would be so cool though. I guess if each brick just broke into 2 or 3 chunks, it would still probably be enough for a game to be really impressive.

 

[Sly]

I'm actually wondering how much processing power it would take to run a voxel based game in software. There's already another game that came out that used voxels, but i didn't get a copy so i can't compare.

As for modeling, maybe we'll use NURBs?

@nhusby
That's also the only way i can think of that you can get voxel to work in a D3D card but 1 pixel per polygon is impractical. Have you seen the horse model in 3DMark03 (Or was it 01). When i first saw that, i thought it was practical hardware accelerated voxels

 

[Terra]

Yeah, like you said, and to be good it would take a lot of processor, plus in multiplayer you have to transfer all changes over the network. It would be so cool though. I guess if each brick just broke into 2 or 3 chunks, it would still probably be enough for a game to be really impressive.

Hell, even just every brick breaking into 2 parts would be impressive....

Terra - Wait 5 years...and we are all gonna look dumb...I love progress

 

[kraemer007]

And lets not forget Carmack said it was much easier to do realistic ray traced lighting with Voxels than it is with polygon rendering...

 

[Spewn]

Eh, destructable environment isn't exactly new, up and coming, bleeding-edge stuff. You ask if it can be done with the next gen cards? Since you're asking a question like that, I'll give an appropriate answer; yes. Considering that, for the time, the destruction in Red Faction was pretty good and included most parts of the environment, it's not hard to make such a leap. Red Faction is quite old, remember. Polygon counts have gone way up, and will continue to do so.

Here's how I see it working; You want a "killer app"? Integration. When you shoot a zombie's arm with a shotgun, you only need two things for it to look physically real. First, you need the wound/damage to the arm to look real. With the right textures and high polygon counts, this would not be hard at all to do. Next, you need the spray/"removed" flesh to look/act realistically. Guess where Aegia's PhysX card comes in? This applies to anything, from walls and bricks to trees, dirt and, well, anything you want.

There you have it, all without completely revamping graphics hardware technology.

 

[vanilla_guerilla]

i dont know. in games i am usually moving too fast or ducking/hiding too much to admire what my grenades do. isnt this a little ass-backwards? all this toenail biting about finding something for the aegia card to do. shoudnt there have been a need, then the aegia developed to fill that need? the way its going, the aegia was developed, and now we are hunting for something for it to do. "i just got a ppu and its not doing anything. must be a pc problem"

 

[Sly]

Eh, destructable environment isn't exactly new, up and coming, bleeding-edge stuff. You ask if it can be done with the next gen cards? Since you're asking a question like that, I'll give an appropriate answer; yes. Considering that, for the time, the destruction in Red Faction was pretty good and included most parts of the environment, it's not hard to make such a leap. Red Faction is quite old, remember. Polygon counts have gone way up, and will continue to do so.

Here's how I see it working; You want a "killer app"? Integration. When you shoot a zombie's arm with a shotgun, you only need two things for it to look physically real. First, you need the wound/damage to the arm to look real. With the right textures and high polygon counts, this would not be hard at all to do. Next, you need the spray/"removed" flesh to look/act realistically. Guess where Aegia's PhysX card comes in? This applies to anything, from walls and bricks to trees, dirt and, well, anything you want.

There you have it, all without completely revamping graphics hardware technology.

Dude. In those examples, physics has ZERO influence on model deformation. The actual deformation is done by the CPU and updates the mesh on the graphics card. PPU's are incapable of directly modifying how an object looks unless it was the one holding it together in the first place.

 

[AACDIRECT]

BTW Voxels have been used in gaming just not as much as Pixels

http://en.wikipedia.org/wiki/Voxel

http://en.wikipedia.org/wiki/NovaLogic

 

[Spewn]

Dude. In those examples, physics has ZERO influence on model deformation. The actual deformation is done by the CPU and updates the mesh on the graphics card. PPU's are incapable of directly modifying how an object looks unless it was the one holding it together in the first place.

The model deformation is the easy part, or are you forgetting that? It's a simple boolean subtraction in any case I can think of. This is another case where I'll bring up indistinguishability. A simple shape with noise randomly generated could be used to apply the wound, and its position would be based on the position of the "center" of the shotgun pellet cloud, for example. Would it be exactly what happens when you shoot a zombie with a shotgun? To be honest, I don't know, since I've never shot part of a zombie's arm off with a shotgun. Maybe someone with more experience in this area can chime in.

Understand how unimportant the accuracy of these effects really is? This isn't a murder mystery, you won't be going over the level with a fine-toothed comb. Even if it was, it would all be scripted and pre-modeled anyway. Using voxels would add a level of complexity which is, at this point, completely unnecessary.

 

[Langford]

It's not just about zombie chunks. What people want is buildings that collaps into debris and roads that develop real chug-holes. Something that is unique every time and not pre-scripted. They want damage that is proportional to impact speeds and directions, and trees that react realistically instead of just falling or not falling. Some people even want the ability to stack odd objects to make primative catapults just for fun.

 

[Sly]

The model deformation is the easy part, or are you forgetting that? It's a simple boolean subtraction in any case I can think of. This is another case where I'll bring up indistinguishability. A simple shape with noise randomly generated could be used to apply the wound, and its position would be based on the position of the "center" of the shotgun pellet cloud, for example. Would it be exactly what happens when you shoot a zombie with a shotgun?

And how exactly do they deform the model. Claiming that the PPU can deform an object is nuts. You're overlapping the capabilities of the PPU with the responsibilities of the CPU.

 

[LuminaryJanitor]

Claiming that the PPU can deform an object is nuts.

One of these videos would suggest otherwise.

 

[Sly]

One of these videos would suggest otherwise.

And once again, those objects are modified by the CPU. The PPU may have some influence, but the actual modification is done 100% by the CPU.

 

[cyks]

And once again, those objects are modified by the CPU. The PPU may have some influence, but the actual modification is done 100% by the CPU.

Yea. The PPU is really just doing some paperwork for the CPU.

 

[nhusby]

the PPU may not actually redraw the detached arm, but to say that it has nothing to do with it is entirely wrong. depending how picky the programmer is, and how realistic / detailed, a bullet hitting the arm is actually a matter of fluid dynamics. the size of projectile, speed, and material (both of projectile, and target) have very different results.

for instance, you shoot a small bird with a bb gun, compared to a desert eagle. the bb gun merely kills it (maybe.) the eagle leaves nothing but mush and feathers, because the shockwave through the fluid (bird) literally rips it apart.

a realistic gunshot would require more physics calculations than the CPU could do in real time. That is why there is a PPU.

 

[LuminaryJanitor]

And once again, those objects are modified by the CPU. The PPU may have some influence, but the actual modification is done 100% by the CPU.

Why is it that a PPU updating meshes and sending them back is beyond the realm of possibility? Bandwidth might be a limitation (on a PCI bus at least), but it certainly doesn't seem impossible.

You're overlapping the capabilities of the PPU with the responsibilities of the CPU.

Rigid-body physics were the responsibility of the CPU too. So was texture rendering back in the day. Why can't this work be offloaded like anything else?

 

[cyks]

Why is it that a PPU updating meshes and sending them back is beyond the realm of possibility?

I forget. Isn't the GPU a problem?

 

[LuminaryJanitor]

I forget. Isn't the GPU a problem?

Only if you're doing the physics calculations on the GPU. Apparently they're incapable of sending the data back to the processor. PPUs don't have the same limitation.

 

[PGHammer]

Suprise, suprise. There IS hardware Voxel acceleration. How long until nvidia buys these guys???
Volume graphics Inc and info on Voxels About volume graphics rendering

Doesn't anyone remember an old (and I mean OLD) company called NovaLogic? They were the ORIGINAL VoxelSpace Gaming Company (and were in fact roundly castigated for it). The vets of NovaLogic must be s!!!!!!ing up their sleeves right about now...

In fact, wasn't the reason the first two Soldier Of Fortune games were so roundly castigated was because not only could you target an individual limb (not Politically Correct, is it?) but because the acceleration model was so wildly different from that used by the games of its day?

In short, the idea is old. Several years old.

 

[PGHammer]

Eh, destructable environment isn't exactly new, up and coming, bleeding-edge stuff. You ask if it can be done with the next gen cards? Since you're asking a question like that, I'll give an appropriate answer; yes. Considering that, for the time, the destruction in Red Faction was pretty good and included most parts of the environment, it's not hard to make such a leap. Red Faction is quite old, remember. Polygon counts have gone way up, and will continue to do so.

Here's how I see it working; You want a "killer app"? Integration. When you shoot a zombie's arm with a shotgun, you only need two things for it to look physically real. First, you need the wound/damage to the arm to look real. With the right textures and high polygon counts, this would not be hard at all to do. Next, you need the spray/"removed" flesh to look/act realistically. Guess where Aegia's PhysX card comes in? This applies to anything, from walls and bricks to trees, dirt and, well, anything you want.

There you have it, all without completely revamping graphics hardware technology.

That is PRECISELY why the Soldier Of Fortune series was popular (and also why it was criticized). While I don't own the first game of the series, I *do* own the second (Double Helix). It was developed several years ago. But that sort of specific targeting offended the Politically Correct crowd.
Note this was pre-Ageia. In fact, this was in 1999. What once was old is apparently new again.

 

[nhusby]

that was nothing near voxels... it wasnt a dynamic function... there were something like 100-200 damage points that could take damage or break off. It was ingenius, but it was a pre scripted emulation of physics. Something really cool would be the ability to shoot holes through stuff, like someones head, and be able to look through the hole... LoL

 

[Firebat]

Yeah, like you said, and to be good it would take a lot of processor, plus in multiplayer you have to transfer all changes over the network. It would be so cool though. I guess if each brick just broke into 2 or 3 chunks, it would still probably be enough for a game to be really impressive.

Truth be told, Bandwith is a problem for many games now.

Say good by to online computer gaming within the next 5-10 years

 

[Terra]

Truth be told, Bandwith is a problem for many games now.

Say good by to online computer gaming within the next 5-10 years

We have the technology already:
http://news.com.com/Internet+speed+record+broken/2100-1033_3-5242144.html

The two organizations announced late last week that they had managed to send nearly 840 gigabytes of data across a distance of 16,346 kilometers (10,157 miles) in less than 27 minutes, at an average speed of 4.23 gigabits per second

Terra - Back to fiddling with my PhysX

 

[WARDOZER9]

If I understand the guys above me correctly they have a point. The use of voxel's and true physics like you guys state would play hell on multi player. It's one thing to drive a car in a circle with 40+ other cars that take no damage and have no debris of physics related calculations running thru dial up or low end broad band where the only location that has to be updated is the freaking block that is each car.

When you talk about blasting a single brick into particles the size of a grain of sand while having 16 people run and jump arround with bullet holes being created at the rate of 100's a second while the paths for the bullets in travel are being updated amongst all the other calculations and locations that have to be updated every 10th of a second.

You end up in the long run with a game that will be unplayeable on anything but an OC3. If I understand the guys above me corrctly that is : )

Is it me or have you guys noticed how 30+ person multi player games from the days of Unreal, Quake and CS have gone the way of the Dodo. Most games nowadays only list that to play a 16 player game you need 512kb - 768 kb just to freaking play. This isn't because of the graphical detail like some think, thats handled on the user end by your own PC. This increase in lag is due to all the extra freaking locations of decals/particles that come from "More realistic physics".

I say forget physics in multi player games for the next 5 years while the internet plays catch up.

Oh, heres a good game that should be supported by Ageia. Red Faction . Probably the only freaking game that lets you blast appart damned near any part of the game world you please : ). And if you got a fast connection and a fast PC with a ton of ram use the crack that removes the decal limit, and use the crack that lets you use more than 32mb of your video cards mem. Then you can tear an entire level into little fragments with a couple of friends ( if you have a few spare hours or just too many rockets ).

 

[nhusby]

I've had this discussion under the thread
Multiplayer PPU (speculation)

if a programmer were to design a game from the ground up for physics and network play, then through paralell processing only the actions of the players would have to be transmitted and each PPU would come to the same reactions.

 

[Firebat]

We have the technology already:
http://news.com.com/Internet+speed+record+broken/2100-1033_3-5242144.html

But how's that effect your average joe?

The demand for bandwith, is exceding the rate of which bandwith is increasing.Thus in a few years gameplay online will become too laggy.

 

[WARDOZER9]

Okay, heres where I stray in and out of topic :

Thats the point. Why are we pushing for these companies to work on physics tech for mostly MP games when it's going to do nothing more than hurt our already deteoriating online gameplay. It's technology like this that is hurting multiplayer games as is.

If you ask me, people shouldn't be pressing for physics in most games instead of just pushing for it in SP games like oblivion ( Maybe not this one specifically ) where it can't hurt anyone.

Ageia is an awesome step in gameplay but it needs to stay the hell out of multiplayer games imho, then it can't be a bottle neck.

If you think about it anyways it would be more appreciated in SP games as most people that play an SP game tend to pay attention to whats going on. How many people do you know that are paying much attention at all to physics in MP games while other crack addicted players are chasing them with an AS12 ? Few, and those few die alot.

So if physics are causing problems with games being laggy maybe those games aren't meant to have physics just yet. If the industry would pay attention we'd see heavy physics and voxels in games that arent affected by any machine but our own.

Hell, I'd love to see a SP game like MAX PAYNE with destructible terrain and heavy physics worthy of the Ageia. But if the industry continues incoporated crap like this into good games with MP capability then yeah, it'll always be an issue when it scomes down to playing your buddy 2 states away.

In all reality though, if I wanna enjoy a MP game with my friends I have the same choice all of us have. I pop in a 5 - 10 yr old FPS and go at it with my friend and 28 other people and never have to worry about lagg.

Afterthought :
Why doesn't the game industry just offer us an option in the game menu to turn off physics and debris ? I just now thought of this. As it becomes more and more of an issue I think maybe this could be an answer.

 

[SuperGee]

Oh, heres a good game that should be supported by Ageia. Red Faction . Probably the only freaking game that lets you blast appart damned near any part of the game world you please : ). And if you got a fast connection and a fast PC with a ton of ram use the crack that removes the decal limit, and use the crack that lets you use more than 32mb of your video cards mem. Then you can tear an entire level into little fragments with a couple of friends ( if you have a few spare hours or just too many rockets ).

If you where impressed by redfraction you would be impressed by soldner to.
I'am totaly not impressed. Soldner is a bad game. And Redfraction is a decent FPS game but it's very limited and no link with gameplay destructable somewhere in the map the same loking walls. Didn't add much.

You could blow up some specific walls. With a specific texture. this is just like how Physics is used before like very limited.

Hardware Physics give you a lot more power, but it still far away to put a dozen high level of relalism physics effect in a game that works well simultainlously.
It's just the first big jump to do a lot more Physics. But the detail of it is still not realistic but a lot better then non.

Novalogic Voxels look nice from far up close it doesn't look nice. And are not vulumetric. It works more like a Texture and a bumpmap. But then on a large scale and in software(CPU) The problem with voxel is the detail level. A large map from far with 4096x4096 voxel is no problem it look reel nice .but if you move closer and standing on a area up close. Then a ingame surface of 1m^2 would have a voxel desity of 8. It would look very Blokky.

The problem with voxels is detail. So every Object must have a kind of Voxel detail lod level This would become a very larg memory hog to be able to macht Texture polygon rendering.

That company that uses Voxel for folume rendere can do it why. The use it pure for One Large high detaild object. A car or Human body Wich can't be done on a PC but the use heavy Workstation/servers 64Bit CPU and Voxel CAD aplication like IA64 Itanium or EM64T/AMD64 Opterons and a lot of System mem on those Heavy workstations.

So the problem is not alone not Hardware accellerated but it eatup lot of memory. to put the detail on a level to mach Textured geometry rendering.

volumetric 3D Textured could be the way. It like being Voxel be hardware accelerated. But All volumetric stuf would eat a lot of memory. So These Tech must come with a compression solution.

If john carmack says it's easier. Like Easier to program doesn't mean easy to implement it on a high detail in a retail game. And I think he doesn't ment Volumetric Voxel but more surface Voxel. as a more realistic looking way of render method. More the way Novalogic uses it. Not like the medical solution in a volumetric way. Wich Physics could use to let is stuff going on at it.

Now with Physics, Volume metric Voxel or 3D textures comes in handy to do highdetailed stuff. Til G-card have 4GB of Vidram it may come some day. And DX11 support stuff to make it hardware accelrated an PPU are there in a 3th generation to compute such finer detail.

 

[vidor]

I don't think the problem in the future will be with hardware. The current hardware can probably support moldable environments and thousands of objects to interact with, although maybe not at game-performance speeds. Unfortunately, to continue this trend requires an exponential amount of work on the part of the designers. Can you imagine creating just a simple office (like the one I'm sitting in) with each piece being a functional, destructible object?

Just to name a few objects, I have wires and cords, a whiteboard, 3 dry-erase markers (different colors, and brands, different shaped caps), an eraser, some tape on the whiteboard, a hat rack, an old yellow plastic poncho from last week when it rained, any number of papers (some yellow, some white), some vendor catalogues, a binder, a tape dispenser, a computer, a few pens/pencils (all different shapes/sizes), a phone, a flashlight, a stapler, my hardhat, safety glasses....and thats just the stuff on or around my desk. I know thats not the same thing as making walls and buildings to destroy, but the wall in my office is similarly complex. Sheetrock, steel bracing, metal window pane, glass window, blinds, brick outside, surely some sort of Tyvek weatherproofing, insulation, carpet on the floor, electrical wiring...the dynamics of such things are very hard to reproduce.

In order to make something look "good" you have to make it sufficiently detailed, complex, and with pretty good mock-up of material properties. For example, sheetrock should blow apart into chunks and dust, while a steel wall should not. You're seeing this now with anatomically correct human models being built from the bones up, rather than wireframe shells. To accurately model that stuff game designers are going to have to get into things way beyond newtonian physics -- something along the levels of teaming up with materials scientists or engineers to get down to the material property level (i.e. modulus, yield strength, ductility, brittleness etc). It will be the anatomy of a building (or hill, mountain, tree, car, etc) instead of that of a person.

It sort of reminds me of the quagmire Sony got into with the in-game economy of EQ...and Blizzard hired some highbrow economists to help design their game system. Physics will be no different.

I think thats the wall that "realistic gaming" is going to run into. It will just get really really BIG really fast, and require a ridiculous amount of work on the part of designers.

 

[psychoace]

Your really talking about an ultra fine detail that most people with there untrained eyes wont notice the difference. A game is about fun and there needs to be smarts on the developers part to know what amount of time needs to be allocated for certain things and what can easily be half assed. It's a game so you don't need to have the correct calculation of how a wall will break and crumble when an RPG hit's it. Then keep calculating over time how the support beams are handling the weight and if it's stressing anything which might break and thus cause other problems. That's just overkill we just want stuff to blow up and graphics to look great. Real life physics are no fun in comparison to Movie physics.

 

[nonlnear]

I don't think the problem in the future will be with hardware. The current hardware can probably support moldable environments and thousands of objects to interact with, although maybe not at game-performance speeds. Unfortunately, to continue this trend requires an exponential amount of work on the part of the designers. Can you imagine creating just a simple office (like the one I'm sitting in) with each piece being a functional, destructible object?

...

It sort of reminds me of the quagmire Sony got into with the in-game economy of EQ...and Blizzard hired some highbrow economists to help design their game system. Physics will be no different.

I think thats the wall that "realistic gaming" is going to run into. It will just get really really BIG really fast, and require a ridiculous amount of work on the part of designers.

And what's so wrong with that? There was a time when 3D rendering was the stuff of supercomputing. The idea of rendering games in 3D was the stuff of fantasy. But then it got "really BIG really fast, and requires a ridiculous amount of work on the part of designers" - at least compared to Pong.

It's rather comic to think about the remark you make with incredulity:

and Blizzard hired some highbrow economists to help design their game system.

Imagine the nerve of them, designing a game that has an economy so complicated that it required - gasp - an economist!

Fully imersive physics is coming. It is still a long way off. There will be many generations of game engines between now and then that use physics selectively. Even long term, I think the use of full physics will always be selective - at least good implementations. But ridiculing comprehensive physics models simply based on the cost of coding and processing is just silly.

There will be libraries built. Virtual universes that game developers will be able to buy/sell/modify/customize. With the appropriate layers of abstraction*, it's conceivable that fully immersive physics titles could even become the stuff of homebrew.

* That's a large caveat, but not inconceivable. In fact, I think it's likely that there will be a GPLed "GNUniverse" available for OSS game physics development within a decade or so.

 

[nhusby]

I don't think the problem in the future will be with hardware. The current hardware can probably support moldable environments and thousands of objects to interact with, although maybe not at game-performance speeds. Unfortunately, to continue this trend requires an exponential amount of work on the part of the designers. Can you imagine creating just a simple office (like the one I'm sitting in) with each piece being a functional, destructible object?

Just to name a few objects, I have wires and cords, a whiteboard, 3 dry-erase markers (different colors, and brands, different shaped caps), an eraser, some tape on the whiteboard, a hat rack, an old yellow plastic poncho from last week when it rained, any number of papers (some yellow, some white), some vendor catalogues, a binder, a tape dispenser, a computer, a few pens/pencils (all different shapes/sizes), a phone, a flashlight, a stapler, my hardhat, safety glasses....and thats just the stuff on or around my desk. I know thats not the same thing as making walls and buildings to destroy, but the wall in my office is similarly complex. Sheetrock, steel bracing, metal window pane, glass window, blinds, brick outside, surely some sort of Tyvek weatherproofing, insulation, carpet on the floor, electrical wiring...the dynamics of such things are very hard to reproduce.

In order to make something look "good" you have to make it sufficiently detailed, complex, and with pretty good mock-up of material properties. For example, sheetrock should blow apart into chunks and dust, while a steel wall should not. You're seeing this now with anatomically correct human models being built from the bones up, rather than wireframe shells. To accurately model that stuff game designers are going to have to get into things way beyond newtonian physics -- something along the levels of teaming up with materials scientists or engineers to get down to the material property level (i.e. modulus, yield strength, ductility, brittleness etc). It will be the anatomy of a building (or hill, mountain, tree, car, etc) instead of that of a person.

It sort of reminds me of the quagmire Sony got into with the in-game economy of EQ...and Blizzard hired some highbrow economists to help design their game system. Physics will be no different.

I think thats the wall that "realistic gaming" is going to run into. It will just get really really BIG really fast, and require a ridiculous amount of work on the part of designers.

that is the exact kind of gaming environment I would love.
I realize how rediculus it may be, but thats what I would like, and if someone can offer that to me I would dish out some money.