The Scientists\' Opinions on Gaming Physics - page 2

96 answers Last reply
  1. Quote:
    This post has really frustrated me about 50% of the time. That 50% are the people who ignorantly claim that physics would not add to the game play.


    Any 3d game has to use some form of physics. Nobody said we dont need physics. However, some of us realize that there isn't a pressing need for _hardware accelerated_ physics. Very different concept there.

    Quote:
    Don't be a "12:00 flasher" (If you didn't get that it's the ones who cant set the time on their VCR (yes, VCR) and you might be one)


    I don't see how its relevent at all, but I don't own a VCR but have an annoying habit of setting other people's VCRs/Microwaves/etc for them.

    Trying to get back on the subject, you mention player collision detection, but does Ageia or Ati/nVidia do anything about that? I don't believe they do. Just because there is a specific physics problem in a specific game doesn't mean that slapping on a phsyics accelerator is going to fix that specific problem. In order to fix a problem with a physics accelerator, the game has to be written to use the accelerator to calculate the movement of whatever objects. In many of these cases, especially in the case of 1 object (in this case, the player) colliding with a solid object (in this case, the wall), I don't believe that offloading this detection into the physics accelerator is going to save any CPU time. In fact, I think in most 3D shooters, the collision box of the player is intentionally made smaller than the player model itself in order to allow a larger range of free motion for the player, and thus making the game more enjoyable to play, which is what's important in the end anyway.
  2. Quote:
    Just plain wrong, and gameplay IS physics (the story line is different). Physics in games is much more important than pretty graphics- it's probably the most important thing, and the hardest thing to do well. That's why the developers need new hardware help. Most people think it's just stuff looking like it's blowing up but that's wrong. Most of that is just pre-scripted events because they don't have access to real physics calculation horsepower. The building will "blow up" exactly the same way every time no matter where the missle hits it or how strong the missle is. Physics includes the whole "world" in the game. Gravity is applied to objects like jeeps, bullets, people, so they are drawn towards world-center. So when your character jumps, you come back down and don't behave like superman or a jet. Like the article said, it's simple to make a flat wall look textured like a stone wall. But once something happens to that wall you need physics. And the more the better.

    Right now cloth doesn't behave ingame like cloth very well, nor does water. Empty soda cans on the ground don't even move, let alone crush when stepped on. Physics would let me pick up an individual can and toss it in a room to distract the enemy so I can flank them. Or move/stack boxes so I can climb on a roof. Some games have varying degrees of these things. Bottles and shelves move in FEAR but you can't do anything with them. Let me pick up a flare in one room, then 3 rooms later find a gas can, and let me make a IED if I need too.

    Last time I threw a football and it hit the ground it randomly bounced all over. Right now in a computer football game the football behaves as if it was round, last time I saw. Of course they have to make it unrealistic to a degree so the gamer can easily get the ball, especially on consoles. In every game except maybe checkers, physics is very important but can't be done well yet. Even the original Pong game had a basic physics.


    You and the person before you both missed the point of what Zok was talking about. He, like many others (myself included) dont give a rats behind about physics, atleast not to the point developers, and maybe you (and the guy who posted before you) would have everyone believe.

    Game play is NOT physics, Physics is game mechanics (or atleast a portion of game mechanics) Gameplay is atmosphere, storyline, how well the game plays over all, etc. A well thought out game, with lots of atmosphere, a decent storyline, along with a well layed out UI, and probably something original (too many developers are just making sequels, or doing something someone else has already done to death).

    As an example, lets take the two wulfenstien games. The original as I recall (I bought it when released) was released in the Mid 90's ( 93-94?), and was fine for a mind numbing game. Now with the second version, what has really changed ? Other than newer graphics (including Physics behind said graphics I'm sure . . ), sound, and possibly different, or at the very least, different LOOKING levels, nothing . . .hell for all we know, they're using the same game engine for both (highly doubtfull, but there is a point here).

    How many FPS games did you play before they all started running together ? RTS ? MMORPG's ? The Idea what I think Zok was talking about is that we dont need more eye candy, but instead more originality. SO in effect, you're trading eye candy for good solid 'game play'.

    wait wait wait... so you're telling me that the physics engine in HL2 didn't affect the gameplay what-so-ever? Come on now. If you truely don't give a rat's ass about physics, then you are in the minority my friend.

    BTW, have you see Valve's new game called Portal?

    Nearly the entire game is based on an advanced physics engine. Here's a link to a trailer for it:

    http://video.google.com/videoplay?docid=1940781722688043082&q=valve+portal

    Now tell me this game doesn't look original.
  3. Quote:
    So yeah... I am all about being able to blow a hole in a wall to make a new path or hiding behind something solid or whatever is useful as far as the gameplay is concerned so that I can use the environment in my own ways to win... but I don't see how it's necessary for it to behave in exactly the perfect physical ways as long as its close enough to be predictable to the point where I can do those things.


    I think anyone with any perspective agrees with you; it doesn't really matter how perfect the effect is. What's important here is that effect is possible in a believable way.

    5 years from now, when we're all using video cards (or whatever) with accelerated physics playing Half-Life 3 or whatever, you'll know exactly what I mean. Physics acceleration is already in the works with ATI and nVidia. So whether or not you think it's necessary doesn't matter. It's going to happen. First it'll come on PC... then once Sony and M$ see the potential, it'll be on the nexgen consoles.

    You cannot have realism without physics. Period.
  4. I understand what you're saying.
    I just wish current problems get fixed before get worried about acceration.
    I think the Ape got it right about 1 month ago.
  5. Quote:
    This post has really frustrated me about 50% of the time. That 50% are the people who ignorantly claim that physics would not add to the game play.


    Those of us you claim are ignorant of physics not playing a factor in games, may have been playing PC games since the mid-late 80's (such as myself)

    Quote:

    Tell, how many times have you followed a player in a online game (for instance CS:S) and saw when they turned, their rifle or gun went through the wall.

    If I turned with a scope up to my eye and hit a well I would expect a black eye...or a least a bit of disorientation.


    Only an Idiot (in real life) would run around with a scope up to thier eye, or even shoot a rifle with a scope up to thier eye.

    Quote:

    Physics can change the entire game. A concushion from an explosion can cause damage, sure. But, a chunk of flying concrete could end things right quickly!

    Please think of the possibilities before stating that physics is unneccesary. When you state such thing you clearly have not thought it thorugh. That type of thinking is, "things are fine the way they are" and there is no forward thinking, imagination or creativity (stroyline or technology related).

    Sure, Ageia may not have it right (now) but the is no mistaking that the idea is right.

    Don't be a "12:00 flasher" (If you didn't get that it's the ones who cant set the time on their VCR (yes, VCR) and you might be one)


    Again, you missed the whole point of what atleast I was talking about. Sure, physics is great, but adding physics to a game at the cost of removing Gameplay is un-exceptable (which is probably what is going to happen). If I didnt already know better, I would suggest that you get past the graphics of some of the older games on the market (from the late 80's to mid 90's), and see what I'm talking about.

    Oblivion is the perfect example, Theoreticly speaking, the game is fairly advanced, has alot implemented into it that is 'supposed to add to gameplay'. In real life however, if you compare the game to Dagerfall gameplay wise, the game is a dismal failing. Now, if you're thinking 'hey, the game has months, if not years of gameplay in it', then my answer / question would be, 'how often do you play the game? 5 minutes a week ?" . . .
  6. As much as you guys (and the scientists) seem to hate ageia, it's the way the industry is headed due to a lack of options. Sure, the video card approach seems better at first, untill you get into it. Here's what ageia has going for them:

    1. Ageia's SDK is free to anyone who supports the PPU. Thanks to this simple fact, there are numerous games with PPU support being launched within the next 6-9 months. This includes a big name game and it's accompanying engine, Unreal Tournament 2007. They might not have their killer app now, but it's not too far off. On the other hand, I haven't heard of a single game that supports havok FX yet.

    2. Ageia has a powerfull physics engine that performs well in software emulation. While this might sound like a bad business move for a company who has banked on selling PPU's, it's actually pretty smart. Look at cell factor. While it is a craptacular beta of a single map (the actual game is set for launch in Q4), it does prove that a PPU isn't needed for mass quantities of objects and all the accompanying calculations. As long as the more complex functions of their engine are left as effects (fluids and cloth atm), it is possible for people without PPU's to be on equal footing with those who have PPU's in a multiplayer environment while providing a benefit to those who have one.

    3. Enhanced performance via driver and engine updates for a longer product lifespan. Ageia claims to be able to enhance their hardware's performance through driver and engine optimizations. Based on the performance boost I recieved by updating from their 2.4.4 driver package to the 2.5.0, I'd say they weren't kidding. IIRC, pre-launch they were claiming to be able to add features through software along with increasing the particle limit from 35000 at launch to 50000 with mature drivers. Given the performance enhancements ATI is able to deliver to their high end cards through driver updates, this doesn't seem to be too far fetched.

    4. Time. Ageia has a massive head start on Havok for game development. Even if Havok FX can turn a GPU into something comperable to the PPU, it may be 12-18 months before we get to find out. That's long after the second wave of PPU supporting games starts hitting stores this holiday season.

    For the capabillities of the PPU, I highly doubt their as limited as that scientist heard. From page 3:

    Quote:
    He says that from a researcher's theoretical viewpoint, and from what he's heard so far, the PhysX card doesn't look promising, for two main reasons. First, the physics algorithms are locked into the hardware, which prevents programmers from changing the algorithms if they find better ones.


    I trust ageia's claims of programmabillity and what I have seen with my own physx card over a rumor that a scientist heard and his resulting theory. Given that the physx engine can be entirely run in software provided you have the power (how the 360 and PS3 will do it), I doubt there's anything hard coded into the card asside from some sort of primitive physics shader system. Also, you have to realize that this guy might be a professional in physics simulation, but not gameplay. With the current state of game physics, anything that resembles reality is better than what we have. This is what ageia offers for free and some shiny stuff if you're willing to pay the premium.

    Finally, increasingly complex game physics are here to stay along with the PPU in some form. Ageia might be a little early, but they're getting the ball HL2 created rolling. I'm willing to bet that within a year from now, you wouldn't think of building a high end, no budget pc without a PPU.

    EDIT: For the gameplay vs physics debate. You can have gameplay without physics, but not physics without gameplay. Having physics without gameplay is like having a car without wheels. There's no point to it.
  7. Quote:
    with out physics you could not move , fire .. hell there would be no game at all


    Again . . . Physics is part of game mechanics, not game play.
  8. Quote:

    EDIT: For the gameplay vs physics debate. You can have gameplay without physics, but not physics without gameplay. Having physics without gameplay is like having a car without wheels. There's no point to it.


    Couldnt have said it better myself, thank you :)
  9. What do you think, does ATI's merger with AMD give the company an advantage compared to NVIDIA and maybe even Ageia in optimizing physics related performance? Would AMD and ATI share their CPU and GPU secrets with each other to produce some sort of physics card? Come to think of it, ATI has expertise on GPUs, AMD on CPUs, and with Ageia they could be a company that can produce CGPPUs. 8O
  10. Once hardware physics proves it's worth, the ATI/AMD beast has the best chance at comming up with a competing product. ATI has the experience creating massively parallel chips and AMD knows how to make a mathmatically strong chip. All they need to know is the recipe.
  11. In reading the article, I thought not enough was really done to give the reader some background on the range of applications to various characteristics of game play both in terms of the visual responses and images the games show, and in the response and movement of objects that are of direct interest to the game player himself/herself. I think some of the main areas where physics comes into play include:

    (1) I as a game player take an action against another physical object within the game (another player, building, etc.) This could be with a weapon, my fist, etc.

    (2) Some non playing character or inanimate object takes a similar action in a game that I can observe. This could include a tank I am observing. Or this could include a bomb that has been detonated. Multiple objects could be involved and both or all of them could be dynamic in some way.

    (3) I attempt to move myself or a transportation vehicle I am using in the game...such as driving, or running or jumping.

    For each of these categories one could work to define specific sub-categories of physical situation and responses which differ in significant ways.

    The next step would be to describe the physical or materials response phenomenon that would be expected or desired as a response. From there the mathematical options to represent those interactions could be considered. The need to solve realtime physics of motion equations, along with mechanical engineering materials response equations would have to addressed. Sometimes the use of pre-determined, canned responses would fine, other times it would be totally unsatisfactory. It really would depend upon the specific details and context of the action and how it really relates to actual game play and the overall experience of the player(s),

    Obviously all this goes well beyond the scope and intent of the article. My point probably really is that the representation of physics in games is in its early stages, and there much to be learned. One area that was not mentioned is the relationship between physics and the control inputs of the game player and specifically the use of joysticks and controllers.

    Over the last 30+ years a great amount of learning has occurred in the 3d graphics area. ( I remember when I was trying to build car following scenes in 3d on the ADAGE graphics processor in 1969...and the "hidden line problem" was the topic of university theses) The same will hold true for physics over the next 10 or so years. Eventually we will have names and approaches for all the physics phenomenon we believe are worth addressing.

    These current special purpose hardware will offer some solution, but soon give way to a more general capability enbodied mostly in a set of practices that are likely to be performable by general CPU power with some specially design pipeline processing. Arithmetic processing and logical processing are both needed. Display GPUs will probably be found to be sub-optimal too.

    chers...
  12. My point was not that there is a pressing need but a will and interest to move forward (really? you didn't get that? I should be more clear, sorry)

    Real time physics is not usually employed in games as a whole.

    What phsyx (just to annoy) we do have is largely scripted.

    I know Havok and all that is really good, in fact once I finished HL2 I started the game over immediatly!

    I thought this thread was about physics and not whether a dedicated card was neeed or not.

    Physics is what makes a game more real...and as for the 12:00 flasher it represents that new technology is not deemed needed so is not learned.

    Maybe Ageia doesn't have it right but it is another step in realism!
  13. imo, physics are not the nxt big thing, but something that needs to be factored into games these days along with gfx and sound
  14. one of these days we are just gunna walk in to games and live in a diffrent world.. then power goes out and your F*cked...
  15. Quote:
    one of these days we are just gunna walk in to games and live in a diffrent world.. then power goes out and your F*cked...


    :lol: Nice perspective! I think all of us needed that one!
  16. Quote:

    Trying to get back on the subject, you mention player collision detection, but does Ageia or Ati/nVidia do anything about that? I don't believe they do. Just because there is a specific physics problem in a specific game doesn't mean that slapping on a phsyics accelerator is going to fix that specific problem. In order to fix a problem with a physics accelerator, the game has to be written to use the accelerator to calculate the movement of whatever objects. In many of these cases, especially in the case of 1 object (in this case, the player) colliding with a solid object (in this case, the wall), I don't believe that offloading this detection into the physics accelerator is going to save any CPU time. In fact, I think in most 3D shooters, the collision box of the player is intentionally made smaller than the player model itself in order to allow a larger range of free motion for the player, and thus making the game more enjoyable to play, which is what's important in the end anyway.

    True but it colliding with emersion. Its a trade off.
    Well it what Devs focus on, more emersion at what cost. A Dev could choose to keep it simple. Or find a solution. With more CPU power and hardware acceleration solution comes closer with in range to use it in a game. It means more work but adds to emersion.

    Some games already using gun colliding.
    If cramped space the gun are point down if you hit a wall. You can't lay down close to a wall and unable to turn against the wall.

    This means not one cramped down bounding box. But a Largere one for fast colission detextion and that passes if true to a finer one. With Mesh collision but simplified by a complex bounding hiarchy consisting out of mutiple bounding primitives to aproximate the complex mesh.
    A charactor with ragdoll PhysX atribute already have this dynamic bounding boxes for each body part for Ragdoll Physx.
    Like if the Gun gives a collision the charracter put it gun back Up or below. If the head crash against low ceiling the character could duck automaticly. Or fel down stunned. If charactor comes to a small pass way it adjust to a smaller siluette by putting gun in line and move side way's.

    Its more work a bit more computational and more anime poses must put in.

    PhysX poor games wouldn't burn on such feature but a PhysX rich game are more likely to put this feature in and take the more work. To enhance the overall emersion though all the game.
    Its a choice for Dev's to utilize resourses to this feature and how important or not this feature is for the projects.

    //Oh my balls. :)
    This reply its not about ageia PhysX But in general PX so in this context
    Legenda
    X = ICS /ics
    P = Phys
  17. do you guys even realise how amazing a computer is? I mean the speed computers run now is just amazing the Mecipt ran only 50 instructions per second and had a external memory drum equivalent of 3 kb, its media was a tape puncher at a few tons, that was 1961. In 1961 they could not even dream of what a computer is today. And now we are adding physics jesus at this rate we will be able to walk into a computer soon :)
  18. Quote:
    with out physics you could not move , fire .. hell there would be no game at all

    Again . . . Physics is part of game mechanics, not game play.
    Well, according to Wikipedia:
    Quote:
    Generally, the term "gameplay" in video game terminology is used to describe the overall experience of playing the game excluding the factors of graphics, sound, and the storyline. The term "Game mechanics" refers to sub-elements of the gameplay, but particularly the primary control and movement features of the game (thus excluding things like level design or AI).

    I think Physics counts as game mechanics, but like all game mechanics, it impacts gameplay.
  19. It would be neat to have a model of the physics card/ depending on the stteing of the game/ physics model might sensibly be selected from
    * orbital mech (ie spcae ships, asteroids, comet tails, solar wind)
    * fluid drag (plane in air, swimmer in water, bullets in water)
    * momentum transfer through semi elastic bodies (ie tackling someone, bullet himpact, .. etc)
    * deformation and faiiure of structures (ie how does a bridge break?)

    in some games on ly a few apply/ but I am sure that if the physics add-in has the computaion ability plus the memory bandwidth plus a programming model to solve even one of them at order of magnituyde faster than CPU speeds/ with open algorithms - the engineers working in these fields would be real interested!!

    Scientists are interested in the difficult/unsolved questions/ engineers would be happy with cheap approximations to eliminate some of the less useful designs before paying 50k for some time on the super computer to confirm the final result.
  20. 9/10 article :D
  21. :? In any given "real world" simulation there are to many variables involved to be accurate to even give a similar real world effect. Getting PhysX algorithms right is paramount to having Gods understanding of the world. So for the most part getting accurate representations all the time in games is impossible. Sometimes it might "look the part" but hand in hand with goes unforeseen variables witch will cause the opposite of the desired effect. Each will cause a game to alternate from feeling real to not feeling real. Einstein could not do it! How do you think anyone else will?
    PhysX is a TOTAL waste of money. Developers should just improve and evolve there current methods using CPU and GPU capabilities. This will more than suffice, especially in the future when they are powerful and include even more functionality


    PhysX Sux!
  22. Intelligent post there. Not.

    I think they are getting a little ahead of themselves with teh idea that people will go out and purchase a totally seperate card that will contain a physics coprocessor to emulate linear interpolations of higher order physical formulae, but the idea that one would help with gameplay and immersion is not a bad one.

    I think that they do need somethnig like this as the next step, but how a seperate processor would be any better than another core on a CPU is puzzling.

    I mentioned before about the old 486 SX/DX chipsets and their introduction of the math co-processor. This thing made the chip JUMP ahead of the 386 in regards to processing speeds on programs designed to use it. It was something needed and eventually just became part of the whole.

    Now we are looking at dual and quad core CPU's (anyone remember the quad GPU a few years back) and I think that the guys in software should be looking into things like this for future development. Let the hardware manufacturers find a better way to calculate where your debris is falling, just make sure your proggie has the ability to have it.


    I also agree that this would be a problem in MP. If a building collapses, how could you simplify it on other machines that do not have the power to calculate how/when/where? Do you make their algorithm simple and wait for a confirmation from the ones that CAN calculate it? That would be a major coordination issue, but I think it would be possible.


    Now, one step further. Has anyone even thought about EM? The fact that you would be able, eventually, to get a TRUE ac of electricity to shoot through the air rather than a "health damaging field" programmed into the machine?

    What about radiation? What about electromagnetic pulse and resonance? What about even simpler things like blast wave reflectivity (echo) which is being used in todays analysis of buildings?

    So many possibilities.


    Back to reality though. I think the idea is great, and I am looking forward to it, but I do not think that what we are seeing now is all that important or will make enough of a difference to wararnt going out and getting it. I think that this would be solved better by doing it in software first, utilizing the second core of the new CPUs, and then having hardware companies come out to try to make a product that will be optimized fot the work and free up system resources.
  23. I dont think physics are the bottleneck here, I think our polygon
    rendering systems are the problem. In a current engine a special case
    needs to be made for every deformable object in the game. In order for you
    to break, shoot, burn, or smash a simple chair in the game the designer
    has to make a deformed version of that model to display once you have
    smashed it. To complicate things, what happens when you smash it on the
    left side instead of the right side? Or even the bottom or the top? Then
    imagine doing that for every object in the game, every building, window,
    car, tree, door, and even the ground. The problem is, you couldn’t. You
    would have to have tens of thousands of extra models and lines of code to
    address the behavior of all those objects. I think the solution is voxels
    or a voxel like rendering engine.

    A Voxel can be thought of as a Volume Pixel. Instead of being a flat two
    dimensional pixel, its a three dimensional one. Like a cube. So if your
    modeling an office chair you build it out of Voxel cubes like your making
    a chair out of Lego bricks. What you end up with is a solid model of a
    chair inside and out.

    Once you place your 3d "Lego chair" in a game it can be easily broken,
    smashed, shot, or burned, simply by altering the shape of the object by
    removing or moving the individual voxels to deform the object. Using
    Voxels for every object in the game means everything in the game can be
    physically acted upon, just like the real world. Voxels can be as small as
    one pixel in your display, so they dont have to be big chunky blocks.

    Carmack has said before, he wrote a voxel rendering engine for Quake2 and it ran really well. He has never spoken publicly about it again or released any source code.

    The directX team has stated there are several functions in DX9/10 that will accelerate a voxel rendering engine, but nobody has stepped up to write one.

    Please, ding dong your favorite game developer and tell them to do some voxel experimentation.
  24. It's not so much of physics being a bottleneck, it's the CPU. Even with multi-core chips, you are still wasting about 40-50% of your cycles on the directx API. Once you've done that, there's not too much room left for physics after running background tasks and the rest of the game engine. By moving physics to a specialized chip, you've got plenty of computing power with some dedicated memory at it's disposal while removing a very CPU intensive task. However, as I think ageia has discovered you're still increasing CPU usage by having a seperate chip. Since directx lacks a physics element, you have to run you're own API to coordinate the CPU with another co-processor. To make matters worse, this co-processor is located on the antiquated PCI bus, increasing the number of wasted CPU cycles.

    Under vista, the situation should be substantially improved. First off, the directx API is getting a significant performance improvement. IIRC microsoft is promising about 20% CPU usage with the new API. The GPU is nearly cut free from the far slower CPU (the CPU no longer has to pre-chew data for the GPU). Also, rumor has it that microsoft might be planning on adding physics, or at least additional co-processor support, to the directx 10 API at some point in the future. Combined with moving the PPU to a faster PCIe bus (hopefully something like 4x), performance should finally begin to benefit from having a PPU.
  25. Quote:
    lol, i bet you went and edited it just now. wikipedia is written and edited by people and is not to be used as fact on subjective topics.

    I have a Wikipedia account. It's Twile. Though if you knew much about Wikipedia, you would've checked the history, and seen that the only edits since mid-July have been vandalism and a revert, neither of which were by me.

    True, Wikipedia isn't the difinitive source, but it's a pretty good gauge of a lot of things. I'd like you to tell me, then, what Gameplay and Game Mechanics are. What they said made sense. Just because you pay an editor to do something doesn't mean it's any more accurate or well-written than what a collaborative effort can achieve.

    Quote:
    :? In any given "real world" simulation there are to many variables involved to be accurate to even give a similar real world effect. Getting PhysX algorithms right is paramount to having Gods understanding of the world.

    Gotta agree with Ninjahedge. You're taking this way too far, and as such, it makes the whole concept of using game physics seem absurd. We're not asking for the simulation of every single electron, proton, photon, etc etc to be done in real time with the very first PPU card and game which supports it. We're not talking Star Trek Holodeck-grade software and hardware which simulates things down to the finest detail (though if it were possible we would likely, as in Star Trek's holodecks, use the simulations to build and test prototypes absolutely free of charge and danger). First-gen PPU hardware and software features might be as simple as having water that ripples and splashes in a semi-realistic manner, cloth that tears and bends over character bodies, and deformable or destructable objects. Maybe you can only smash a wood board into 30 splinters in the game or the water flows but doesn't bead and permeate the soil accurately, it's still a huge step up from "invincible" boards and water that only splashes as a crappy 2D-looking effect. You're right in saying that it would be, for now, impossibly difficult to do a "real world" simulation but that's beside the point, we're only looking for a decent approximation of physics we take advantage of. Saying that physics stuff is a waste of time because it's not perfect is like saying "We don't know exactly how the human mind works, and even if we did, simulating it on a computer would be insanely processor-intensive, so let's not do AI for the enemies, they'll just stand there, or maybe walk back and forth and shoot anything that moves."

    Quote:
    I also agree that this would be a problem in MP. If a building collapses, how could you simplify it on other machines that do not have the power to calculate how/when/where? Do you make their algorithm simple and wait for a confirmation from the ones that CAN calculate it? That would be a major coordination issue, but I think it would be possible.

    Discussed on another PhysX PPU related forum but rather than making you go through 5 pages of flame-filled posts, I'll just summarize some stuff I said, right here:

    There are a couple approaches I can think of. First is to say "This game requires a PPU" as a flat rule. Very simple, but it limits the the consumer base to a dangerously small number of people. Until (or rather, if) PPUs become standard, this isn't a good choice.

    Second, you could make PPU acceleration an option with a check box. If you've got the card, your single and multiplayer games benefit from destructable objects and physics puzzles, as well as other fun physics effects. In multiplayer games, the server creator would determine to use PPU acceleration or not. A game without PPU acceleration would work for everyone, maybe PPU-users could have the software physics put onto the PPU for a minimal performance gain, but the point is it wouldn't stress physics to the point of making it hard to play without a PPU. A game with PPU acceleration would be unavailable to non-PPU-users, guaranteeing that everyone there had the physics horsepower to make sure the building collapses the same way for each person. The problem with this is, if PPUs became popular, people without them wouldn't have many games they could play in, and if PPUs were unpopular then the people with them wouldn't see any benefit in multiplayer, and stuff they did in single player wouldn't work in multiplayer. It also means that levels would have to be designed to take advantage of physics and no physics (so you couldn't make blasting your own cave a requirement in a level, but it would have to be an option otherwise the PPU guys wouldn't see a point to own the card).

    A third way to do this would be making physics more of a slider, with varying degrees. Think of it as between methods 1 and 2. Or it could be a series of check boxes. Low-level physics would be just the core stuff we see today: Objects fall and roll, there aren't many of them, and only a few objects can be destroyed, in pre-determined ways (think shooting a flower pot in CS:Source or breaking a box in Half Life 1 or 2). High level physics would be all the stuff we've talked about, breaking objects in non-scripted ways, having many more objects, fluid simulation, cloth, etc. All the settings between those points would gradually increase the realism level by increasing object count (they have this feature in FEAR, lower levels leave out less important props), adding physics-modeled object destructability, the fluids and cloths etc etc. With single player you could use whatever level you wanted, though at a certain point, if your CPU couldn't keep up with the physics demands you'd naturally see a performance drop. It's the same as picking your resolution as a tradeoff of quality versus fluid framerates. In multiplayer, the server creator would pick a physics detail level, or check off the physics features he/she desired, and those would all be simulated. For the people with the PPU hardware it would be offloaded there, for those without it would be done in software. This would allow anyone to play on any server, and the performance they got would be influenced only by the server settings and their own hardware. It would also allow for games to get better with age, quite literally, because newer hardware would allow you to crank up the physics to levels previously unatainable (like how getting a more powerful GPU lets you play your older games at higher resolutions, framerates, and detail levels). The downside, as above, is that some users would realistically be alienated because settings were too high for them to reasonably run, and that PPU-users would see some effects disabled, effects they might've been used to and counted on while playing single player. Also, as above, it would require more variety of gameplay, so that there would be more things you could do with various physics options enabled, but they wouldn't be required to get through the game.

    Overall a difficult question I think. Between now and if/when PPUs become standard, I'd say option 3 provides the most flexibility and support for single and multiplayer, unfortunately at the expense of level and game design times increasing (but on the plus side it would force devs to give multiple ways of doing things!), but overall I'd vote for option 3. If/when PPUs do become standard then it's option 1, like how today you must have hardware graphics acceleration for basically all 3D games.

    There is a 4th option I didn't mention, one that I don't recommend. That is to have the server do the advanced physics calculations via a PPU (or not do advanced physics if there's no PPU) and send the results to everyone. The reason I don't recommend this is that, if you think lag is bad now, imagine when you have to send not just the positions, speeds and orientations of a couple dozen players and objects, but of a couple dozen players and a couple thousand or tens of thousands of objects. Also, there would still be the issues of gameplay styles supporting both PPU and non-PPU-users. Rather, I opt for each person to have the physics hardware. Sure, the cards are all doing the exact same calculations, but streaming out the information on every object in real time would require a huge amount of bandwidth. Instead, let each computer do the simulations, and the only data you have to synchronize is the unpredictable stuff, by which I mean the players.
  26. Well, altought i got tired of reading the books people have written in this post, I'd like to add just an arcticle posted yesterday at gamespot, regarding Cellfactor:

    CellFactor Interview

    Quoted form the developer of the game, here's a proof that PhysX might not be a waste afterall:

    Quote:
    In the tech demo [shown at E3], there are several thousand rigid-body objects within the environment that can be used as weapons, and there are some cloth and fluid elements, but beyond that, you're basically in a static environment. Not so for Revolution. Our goal for this game is now to make the environment as interactive as the objects within it, and our primary way of doing that is to construct architecture from breakable-jointed dynamic objects rather than static geometry (as was done in the tech demo). This allows explosions to shatter stairwells, psychic powers to collapse pillars (the debris from which can then be used as weapons), gunfire to chip pieces off concrete, and all manner of environmental destruction. The downside is that joints are very complex computations, so the performance difference between software and PhysX hardware is more extreme. Thus, we find that some environments simply don't run in software anymore, but that's a worthy trade-off for the havoc you can unleash on a destructible environment. It really makes some of the psychic powers rather insane to see in motion, as everything around you collapses and shatters.


    Eventough GPU manufacturers states they'll be able to do physics, I'm rather cautious with them, as for graphics, they usually mess up with drivers, imagine what they'll do with physics (will they give enough support or will they toss around the blame for issues with havok?).

    I won't buy a PhysX now, but if all the promised games really do come up and have things that can't be done real time without it, i'd give it a try.
  27. Thanks for confirming my point.
    Idiot.
  28. im sure its already been said, but the new partnership between ATI and AMD could yield full AMD processors on GPUs as well as a separate main CPU for the other more generic tasks of running a computer. This would mean we could have a really fast processor dedicated to the physics that is required when running a game. Of course the GPU processor would still exist on the card.
  29. Quote:
    im sure its already been said, but the new partnership between ATI and AMD could yield full AMD processors on GPUs as well as a separate main CPU for the other more generic tasks of running a computer. This would mean we could have a really fast processor dedicated to the physics that is required when running a game. Of course the GPU processor would still exist on the card.


    It's a possibility, but I think we won't be seeing that at least for a year (as both companies said, that at least for 6 months they'll not change their product lines).
  30. I know this is out of the way and will be flame bait but....
    Doesnt out door games have true-to-life physics already? without the need for an addon? What if god made us upgrade for $300 each and the upgrade was inserted though the anus just so we could kick over boxes? Seriously lol. technology is improving so fast that we keep looking further and further into the future, but i still dont see hover cars.
  31. Quote:
    im sure its already been said, but the new partnership between ATI and AMD could yield full AMD processors on GPUs as well as a separate main CPU for the other more generic tasks of running a computer. This would mean we could have a really fast processor dedicated to the physics that is required when running a game. Of course the GPU processor would still exist on the card.


    It's a possibility, but I think we won't be seeing that at least for a year (as both companies said, that at least for 6 months they'll not change their product lines).

    After all it takes a while to design the new parts, make prototypes, test them out, see how the yields are, mass produce, etc. I think it'd be nice if they cooperated and worked on some co-processing possibilities. I mean, given the speed of GPU components (the X1800 XTX has RAM which is just shy of 50 GB/sec vs 3-10 GB/sec for desktop RAM, and we all know GPUs are darn powerful for certain operations) if the two worked out a way to offload some tasks onto the graphics parts, we could see some really nice performance boosts. This could give AMD+ATI a huge performance boost for non-gaming aps which don't use much 3D rendering power, and I'd imagine this would be a tremendous asset to people working with image manipulation like Photoshop. Resizing images and applying visual effects is like small potatoes for a graphics card ;) Or maybe they'd work on a hybrid processor for the graphics card and/or processor slot. Who knows! Whatever transpires between these two companies will probably be interesting to watch either way.

    ...what's the topic again? Game physics or something?

    They rock :D
  32. Quote:
    Doesnt out door games have true-to-life physics already? without the need for an addon?


    No. They don't. I've never seen a game which has "true-to-life physics" and frankly it would be hard to simulate, given that we don't even fully understand all of physics.

    Some games have nice physics-like effects outdoors. You leave footprints and tire tracks on the ground, trees and foliage sway in the wind (like in Oblivion), clouds drift lazily across the skies, stuff like that... but it's not even really physics. The footprints and tire tracks are just textures slapped on the soil, so if for example it were to start raining while you were driving, you wouldn't see water collect in the tracks, or they wouldn't get washed away. Trees and foliage generally move based just on a wind direction and speed, a sort of scripted movement, so if you were to slap a building up in front of a bush which blocked it from the wind, it wouldn't become still. Clouds also move according to those scripted movements based on direction and speed, even if you could get up as high as they were you would find yourself unable to change their shape. And let's not forget that almost all games to date still don't let you blow a crater in the ground with a grenade or rocket. And see, those effects I just described wouldn't even be true-to-life, they'd just be low-grade simulations. If the foliage and soil and wind and clouds and water all did just what you would expect when you or other objects interacted with them, then we could say they looked very realistic, but again, all these things are only skin-deep.
  33. Twile you crack me up! By outdoor games I ment like Tag, soccer, Red Rover, kick the can, Football, tenis...
  34. The world call that sports games.
    outdoor is more like a large terrain map. with some structures on it.

    Quote:
    I know this is out of the way and will be flame bait but....
    Doesnt out door games have true-to-life physics already? without the need for an addon? What if god made us upgrade for $300 each and the upgrade was inserted though the anus just so we could kick over boxes? Seriously lol. technology is improving so fast that we keep looking further and further into the future, but i still dont see hover cars.

    Well i try'ed CF:CT the PhysX demo the latest version. R36
    And it has a Mod tutorial CHM with it with links to tutorial vids and documentation.

    I remember one tutorial in the CHM was about making a hoovering vehicle. It's also online.
    http://reality.artificialstudios.com/twiki/bin/view/Main/CreatingVehicles#Brief_Discussion_of_Hovercraft_P

    Because for a lot of games, hoover cars don't fit in the story or game theme.

    A starwars Pod racing game would be something just about hoovering things.
  35. Quote:
    I know this is out of the way and will be flame bait but....
    Doesnt out door games have true-to-life physics already? without the need for an addon? What if god made us upgrade for $300 each and the upgrade was inserted though the anus just so we could kick over boxes? Seriously lol. technology is improving so fast that we keep looking further and further into the future, but i still dont see hover cars.


    imho the adding of "true-to-life" physics in games would add a lot of fun and a lot of unpredictability as, with a fully interactive environment, users would be able to do almost anything possible inside that game universe, and you might even see gameplay changes...but i think that the greatest effect of physics would be the immersion, the nearer to reality things are, the easier for your brain to accept it (even in an unconcious way) and really bring you into the game.
  36. Bakka! You guys missed my joke. 10 years ago, an out door game was a game played out door, like kids games, sports..... I cant believe you are making me explain this.
  37. Quote:
    Bakka! You guys missed my joke. 10 years ago, an out door game was a game played out door, like kids games, sports..... I cant believe you are making me explain this.


    Sorry 'bout that. I get the joke now, ha ha. But as we spent a good deal of time talking about computer games, and you said outdoor without specifying "sports" or "real-world", I too assumed you meant... well, games that took place outdoors. Like, free of synthetic-feeling "crates" and "guns" and "explosive barrels" and such. Understand that the variety of people posting on this forum ranges from knowledgeable and well-worded to clueless and flame-happy, and... damnit! You made me write up a big paragraph about something you already knew! :P
  38. I just posted a blurb on Graphics and Physics and mentioned how if the 2 merged together (RPU anyone?) , you would have a product worth selling.

    I wont repost, just provide link:

    GPUs versus RPUs and Physics Accelerators

    Im looking for constructive criticism or corrections. Im also hoping to see someone from THG do a more indepth technical article on the subject of RPUs.
  39. Here is a Google Video demonstration of the weaknesses of 3D physics engines.

    http://video.google.com/videoplay?docid=-1279309767827721998


    This 30-second clip was created using Second Life, and shows a fully physics enabled 4-wheel cart, with real physical ball-bearing races, exploding apart due to physics calculation errors.

    The force vectors pushing on the bearings are very powerful but should cause little or no movement of the frame that is balanced across all the bearings. The physics engine cannot deal with the amount of precision required and starts generating positional errors that cause the frame to move around in an unnatural way.

    The cart-frame position errors add to the forces on the bearings, the frame starts to jump around violently, and finally the added vectored energy from the errors is so great that the bearings explode out of the races and blow the whole thing completely off the edge of the Second Life sandbox simulator.


    Note the cart is very large in this example to even get it to stay stable for these few seconds. The demonstration platform is 20 meters x 20 meters x 20 meters. If the wheeled cart had been built smaller the error rate would have increased exponentially immediately and it'd have blown apart in the blink of an eye.


    For the brief moment where the frame is shaking around rapidly, the Havok physics engine is generating heavy CPU load due to a condition known as Deep Think.

    The Havok I engine cannot deal with overlapping/interpenetrating objects since in real life that is impossible, and calculating the impossible is hard on the engine. When jumping around the frame ends up occupying the same physical locations as the bearings, which the engine is working very hard to try to resolve by moving the frame back and forth, compounding the explosive energy errors further yet.
  40. ok now you can do all the cool water effects and such but until you can alter the way things are its worthless. this is why the ppu is so important. its not about how it looks. its about what you can do with what you can see. and that things can happen and effect things when you arent even there. the true path of what physics should be doing in a game. Thats why a PPU is important, and a GPU is crap for Physics processing couple with CPU.
  41. Quote:
    Bakka! You guys missed my joke. 10 years ago, an out door game was a game played out door, like kids games, sports..... I cant believe you are making me explain this.


    Actually, when I first read what you said I thought of Paint Ball. I just did not think it was verry funny... ;)


    Anywho, little things can be done here and now, but some of the others really need a more intensive look into classical physics to be able to emulate them properly and efficiently. To do a proper dynamic finite element model on even some of the simpler objects, such as the cart you saw in the previous example, you need either increased precision, or simplified representation or it will either become unstable, or take an hour to calcultare the first few seconds of motion.

    But other stuff, like holes in the ground, basic fluid flow, environmental interaction (like blowing up cover that someone is hiding behind, etc etc) would be one more step into adding realism that will make you become less aware that you are in a game.

    Whether this needs a seperate processor or not depends on how fast the other processors progress. If the GPU/CPU merge actually occurs, yuo may see the PPU as the next addon card.

    Maybe not.

    Ah well. I hope we see more of this. I am rather tired of seeing an enemies rifle butt sticking through the door I just slammed shut on him!!! ;)
  42. Quote:
    I am rather tired of seeing an enemies rifle butt sticking through the door I just slammed shut on him!!! ;)


    Are you familiar with bounding boxes? Typically 3D games optimize to the extreme to get maximum performance out of the game player's lousy three year old hardware. :)

    So for example, a complex vase with all sorts of curves and handles and so forth may just merely be represented to the physics engine by a simple box that is the rough shape and size of the vase.

    Since the engine looks at the box only, the complex vase is really just a decoration, like a detailed texture of a control panel slapped on a flat wall, except in this case a complex surface slapped on an invisible cube.

    Sometime parts of the detailed surface may protrude outside the bounding box. These parts are ignored and treated as phantom.. two objects with bounding boxes slightly smaller than the object can stack closely together with parts overlapping, without the physics engine caring.

    Your rifle example is much the same. The environment has been optimized to such an extreme that the bounding box is ONLY for the body of the soldier, and does not include the gun.

    As long as the bounding box of the body of the soldier is behind the door, the physics engine is happy.... even if you're not because you see that rifle hanging through the door.

    This is not a failure of the ENGINE, but of the game designer for not taking the time to include the gun in their bounding boxes.

    -Javik
  43. Picky picky.

    But what happens when you close a door and a bounding box is there preventing the door from closing? Does it move the guy away or is the guy an immovable object? Does it deform the box and its contents?

    Now we are starting to get into how the physics engine works.

    As for the interaction, things liek the brush and leaves from Oblivion would be a good example. I believe they have no collision boxes set at all, so they flow through whatever they are near. Trees, walls, the ground. Having the boxes defined AND a physics/engineering engine that would be able to have the leaves bend when they hit the wall, or move (as they do from shock waves in Crysis) according to physical realities rather than scripted movement would be what I would love.....

    ;)
  44. Quote:
    But what happens when you close a door and a bounding box is there preventing the door from closing? Does it move the guy away or is the guy an immovable object? Does it deform the box and its contents?


    That sort of capability has to be explicitly specified to the engine, and often with objects consisting of tiles/triangles held together that will break on a predefined line. There's no way to simulate an actual door frame, because the engine power is not there.

    Sure, you COULD theoretically simulate a solid wooden door, where it is billions of short 3D tubes held together with a breakable bonding force. That would accurately simulate the fibers in wood grain. Firing a fast-moving simulated lead bullet at it would break through the tubes and tear off chunks of the adhesion, ripping out a realistic bullet hole through the thick layer of tubes.

    Heck, there's probably room for some flex to the tubes and the adhesion, so the 3D door could bend, dent, splinter, and be torn right off the hinges just like a real door. Now the scattered chunks on the ground can still be further bent, blown up, and splintered, or picked up and used as a shield or for ablative protection.

    But the processing power? You'd have to continuously calculate the forces and positions of all these tubes all the time to do an accurate door simulation. And this is just ONE door, in a game world potentially containing 50+ doors just for realism's sake. Now we're talking about processing possibly 10E+15 of tubes in doors, and simply as a background process occurring 30 times a second that always runs.... besides whatever else is REALLY going on in the game.

    I suppose if you made the tubes fairly large such as the size and length of a pencil rather than actual size, you'd bring the calculations per door down from trillions of tubes to perhaps a few hundred thousand, but it's still a ridiculous amount of processing power just to approximately simulate ONE wooden door.

    The power is simply not available to do that, and I don't think it ever will be unless we figure out a way to tap into the insanity of qubits and quantum computing.. and apply it to something as unimportant as 3D gaming physics calculations. ;)

    -Javik
  45. Maple's slow though - partly because it's exact. Approximations need to be made all the time in order to speed up physics processing and most of the approximations don't change the dynamics enough to move objects more than a pixel anyways.

    zomg i answered a post that was like 5 months old. sorry.
  46. Quote:
    Here is the solution. Make a Physics card that has a licensed version of the Maple (maplesoft) math engine. by doing this you would allow developers to bypass lengthy coding and just pass simple strings of math to the PPU to be calculated.

    so if you want to solve a system of equations you send the string

    solve( 'equation 1', 'equation 2', ... 'equation n')

    and the maple engine along with the hardware calculated the solution . it would make vector math and linear algebra easy and would reduce the overall bloat of the game

    the hardware exists , so does the software why not put them together to maple a ppu


    I don't think maple has anything like the code efficiency (read performance) you would need for the millions of calcs a second you need for realtime physics in gaming, at least because its a software not hardware solution.

    Gaming physics give away a *lot* of accuracy for speed, which is OK. I mean who really cares (or can even tell by lookinng) for example whether some ripples on water have precisely accurate behaviour? Accurate fluid dynamics calculations still need a supercomputer anyway for realtime results.
  47. Quote:

    ...
    But the processing power? You'd have to continuously calculate the forces and positions of all these tubes all the time to do an accurate door simulation. And this is just ONE door, in a game world potentially containing 50+ doors just for realism's sake. ...



    The trick there is to only perform recalculations based on stuff that you have already indentified as possibly changing state. you wouldn't continuously calculate each door, only those that you already know something may have happened to. the real smarts is in the hueristic that you use to determine that.
Ask a new question

Read More

Memory Gaming