Making A Splash: Nvidia’s Real-Time Fluid Physics

By Adam Smith on April 25th, 2013 at 3:00 pm.

I feel like I should apologise for the headline but Nvidia call their middleware physics engine PhysX, for crying out loud. ‘Making a splash’ is almost Nabokovian in comparison. You may recall recent advances in convincing/crazed coiffures and I care about that about as much as I care about the latest floppy-fringed hair fashions in the real world. Not a jot. Fluid physics though? Ever since the invention of physics, which was sometime just before I balanced bricks on a plank to create a see-saw bridge in Half Life 2, I’ve been waiting for a game with proper water. The latest PhysX tech demo got my juices flowing and you can see it below.

Rapturous.

, , .

126 Comments »

  1. Premium User Badge

    Matchstick says:

    No entirely sure it’s fair to blame Nvidia for the Physx name as that came with the company when Nvidia bought Ageia in 2008.

    Course Nvidia are fully to blame for not managing to come up with a better name in the last 5 years or so…

    • Berzee says:

      It’s like how we inherited a tiny poofy dog from my great-grandma and its name was Muffy.

    • karry says:

      “Nvidia are fully to blame for not managing to come up with a better name”

      …or a proper use for it, at all, anywhere.

    • KeyboardGato says:

      I remember seeing this about 3 or 4 years ago, there was also a video of jelly bunnies being splatered into an invisible wall. It looked great but i can’t believe they haven’t implemented this into games yet…

      Was that done by this “Ageia” company?

      • Nihaopanda says:

        You might be thinking of this video from Lagoa Multiphysics . This was a softimage plugin and not connected to a games engine, but still very cool indeed – the creator of it now went on to work on an equally cool online rendering engine called Lagoa Multioptics.

  2. Yargh says:

    I have to admit, that does look very nice indeed. While one can spot the fairly large water particles in the end-result shots I don’t feel they take anything away from the effect obtained.

    I’m looking forward to seeing this in games soon.

    • The First Door says:

      It seems that every demo released just had smaller water particles to make the water look less oily and more like real water!

      Having said that, this version of it does look rather pretty!

    • cool4345 says:

      Don’t get your hopes up on seeing any technology even close to resembling this any time soon. Maybe even in our entire lifetimes we’ll never see this technology in actual games.

      There were already examples just like this way back in like 2006 of physics demonstrations. Yet almost no games in the last few years have shown any significant use of physics.
      Euphoria. the physics system that GTA 4 uses, was pretty much available to use back in 2007/8. Yet other than GTA, there weren’t any real notable examples of that engine being used anywhere else.

      • Didden says:

        In our lifetimes? Come on… things have come a long way in a very short amount of time. In my lifetime (and I’m not that old) I’ve seen us move from square pixels to – filled vectors – that was a biggie.

        I’d imagine in 20 years, Graphics will be basically close to life like. At which point… Matrix! :)

      • Lukasz says:

        Lifetime??? Assuming an average person here is 30 years old it is still fifty years before his or her life ends (on average)
        Are you saying that in fifty years we won’t be able to simulate water realistically enough in games?

      • v.dog says:

        There’s three different but related problems:
        The first is technical- being able to actually make something like this work in-game at a consistent framerate and in real-time (notice that water didn’t interact with any animated objects).
        The second is cost/benefit- whether the costs (performance hit, implementation time, financial) are outweighed by the benefits that it will bring to the game.
        The third is imaginative- can developers actually think of ways to use it in-game that will add to the gameplay?

        Whether we’ll see this in our lifetime is not a question of ‘can we’ but ‘will we’?

        • drvoke says:

          I noticed the water interacting with falling glass objects. Does that not count as animated for some reason I’m missing? No sarcasm.

    • Mctittles says:

      The programming side of this is actually not that difficult. Just give every particle a gravity algorithm and let it go. The problem is the power to draw the amount of particles needed; which I think will happen in our lifetime.

      • RC-1290'Dreadnought' says:

        It is slightly more complex than you make it sound.

        If you only apply gravity to a particle, it’ll just accelerate in the direction of gravity, passing through everything it comes across.

        If you were to use standard physX sphere collisions, the ‘fluid’ as a whole would compress too much. (Among other problems)

        This video shows a technique where movement and collisions of the particles are handled in such a way that they don’t clump together too much, while requiring less calculations than previous fluid simulations, without having the particles going out of control.

    • jrpatton says:

      It definitely looks pretty, but also spongy. The volume of the water isn’t constant, and is noticeably expanding and contracting when under pressure.

  3. Premium User Badge

    darkChozo says:

    I’m glad that we’re finally making strides towards the rabbit-drowning simulator that every gamer dreams of. It’s about time.

    • McSasquatch says:

      And therefore heralding the development of that Fatal Attraction game-of-the-movie… though to be fair that sounds more like Nintendo’s scene.

    • scatterlogical says:

      Shame they missed Easter.

    • Keyrock says:

      I was quite disappointed in the water physics in Rabbit Drowning Simulator 2013. Hopefully they’ll add more realistic water physics for Rabbit Drowning Simulator 2014.

  4. Scare Tactics says:

    Makes me all wet

    • Robslap says:

      This should really make waves in the graphics industry.

      • bladedsmoke says:

        I’m just sitting here letting the realism soak in

        • Pray For Death says:

          Definitely a watershed moment in graphics advancement.

      • Premium User Badge

        c-Row says:

        I’m shore it will.

      • Premium User Badge

        CheesyJelly says:

        I can hear a ripple of applause spreading across the internet.

      • scatterlogical says:

        It’s high tide they started getting water simulations ship-shape. Makes HL2 physics look like a drop in the ocean.

      • Ultra Superior says:

        Are you all brainwashed? I won’t go with the flow here and I’ll dive right to it: you tube is flooded with streaming videos of watered down CGI realism, but it’s the same old trickle over and over again.

        • scatterlogical says:

          Aw, c’mon, we can never be over-saturated with too many swell videos like this.

        • drvoke says:

          I can already feel my interest start to ebb.

      • Premium User Badge

        skalpadda says:

        I just wish they’d focus on getting their tech into actual applications, instead of the constant drip-feed of tech demos of things that are not yet seaworthy.

      • analydilatedcorporatestyle says:

        I like a bit of fluidity in my Bitcoin mine!

    • Premium User Badge

      Thermal Ions says:

      Ah but how well will it stream in a practical application. Might be a while before they get it so pretty at a frame rate faster than the trickle their showing here.

    • Premium User Badge

      Vandelay says:

      Well, I think they might have tap-ped into something here. As long as my frame rate remains fluid, only good things spring to mind about this torrent of news.

    • I want to stab you to death and play around with your blood. says:

      Had 2 Opine about this. There’s a flood of pissimistic comments putting a wet blanket on the seatuation. Some would say this tech is circling the drain before it even sets sail. Let’s see what surfaces before shooting it out of the water.

    • Keyrock says:

      This is a drop of liquid smooth realism in an ocean of watered down physics effects.

  5. redredredguy says:

    Photorealism is an endless, money-gobbling hole. (This DOES look pretty cool though.)

    • Mr. Mister says:

      Hey, physics systems are not explicitly related to photorealism. They’re not even included in art style.

    • Koozer says:

      The thing with physics is that it applies to everything, everywhere. With art you have to draw a new wall texture for every wall you make, but they can all use the same physics to fall over and kill fluffy animals.

      I also have no idea whether what I just said is at all approaching the reality of game development.

  6. Premium User Badge

    distantlurker says:

    That’s a good wedge of graphics, sorry, graphx. Soggy graphx, but graphx nonetheless.

  7. Liudeius says:

    I always wonder how large a super-computer is necessary to make demos like this barely run.
    (Ok, it probably doesn’t need a super computer, but add in a game around it and it probably will.)

    • HadToLogin says:

      Well, unlike most renders that run on best cards, quite often in some akimbo version, this one was rendered by one GeForce 580. That gives hope it will be something usable “tommorow”, not in next decade.

      • karry says:

        “this one was rendered by one GeForce 580″

        Or so it is claimed. Let’s be honest here, none of the AMAAAAAZING tech we see in GPU companies tech demos actually find themself put in any real world product within at least 5 years. Nvidia demos from 2003 are still more advanced than almost anything i’ve seen in actual games to date.

        • analydilatedcorporatestyle says:

          Too true.

          It won’t be using the standard driver set, not a chance and to be fair PhysX has just been seldom implemented hype up to this point. Which developer, even if heavily sponsored by Nvidia wants to alienate half of the PC gamers. The whole PhysX debacle stinks, FUCK YOU NVIDIA hello AMD!!

          • Premium User Badge

            Matt_W says:

            Mirror’s Edge. Which was still a very good game without the cool broken glass and wind effects. It would have been nice to have had a warning though: “Turn off these options if you’re using an AMD card unless you fancy seeing your framerate drop to 2fps whenever a window gets broken.”

          • Dr I am a Doctor says:

            On the other hand, lol @ gaming on an amd.

        • VengefulGiblets says:

          Yeah, that’s the real issue. I’ve seen too many of these over the decade and next to none of it has manifested in actual games.

          Personally I think that these demos shouldn’t be reported on, as fun as they are to watch. If the companies (NVidia, AMD) want to generate hype and attention for the capabilities of their technology then they should be forced to do so through real world games. No game? No headline, and I’m not talking about interactive tech demos.

      • TaylanK says:

        Not so fast. It may look okay when it is run in isolation but an actual game could have tons of other geometry to keep track of and render. One could also imagine that the performance goes down as the volume of fluid increases. I don’t expect to see any oceans filled with those tiny balls anytime soon.

        • Premium User Badge

          Gap Gen says:

          There does need to be a way to change the resolution where you need it, even if for an ocean you’re only modelling the surface layer, or to switch from pre-rendered waves (say) to physical modelling around the hull of a boat / body of someone wading through a river.

  8. misterT0AST says:

    It looks a tiny bit too floaty and splashy, particles seem very eager to jump around and roll out of things maybe a little bit too fast, but I think they did it on purpose to show off the waves and splashes a bit more clearly.
    I just hope in games they actually use heavier, denser fluids though, rather than showing the “look at how cool this is” squirts of liquid that reach 2 meters as soon as you step into a few inches of water.

    • Premium User Badge

      P7uen says:

      The innocent tiny teaspoon that always positions itself exactly under the tap when I turn it on so that it creates a huge sheet of water and makes it look like I’ve wet myself disagrees with you.

      • Berzee says:

        So does the perfectly curved bit of ice cream forming a ski-jump-for-milk every time I attempt to make a shake. =P

    • CutieKnucklePie says:

      Yeah to me it looks sort of like the uncanny valley for water.

      • Josh W says:

        I aggree, there’s this strong sense of “I know there’s something wrong with this, but I don’t know exactly what”.

        In fact, it occurs to me that this could be the reason for the uncanny valley that is totally different to the usually assumed one (looks like dead/diseased/triggers racism). Perhaps instead of triggering reflexes that we use when making gut judgements to preserve our social groups, instead it slips into that gap between differences we recognise and ones we don’t. We see but cannot pin down why it’s wrong.

    • Koozer says:

      One thing it’s definitely missing is surface tension. If you look at the edges it looks like silly putty.

      • Premium User Badge

        VelvetFistIronGlove says:

        That was particularly noticeable in the shattering-fishtank shots, as the water never started dribbling down the edge of the block, but was still flowing over the edge when its velocity would’ve been too low.

        Though I guess proper dribbling would also requires hydrophilic surface simulation.

      • Premium User Badge

        LTK says:

        That’s part of it. I don’t think they simulate the adhesive quality of water properly, so it looks less viscous than actual water. Maybe it would simulate pure ethanol better.

      • snakeo_x says:

        or prandtl’s boundary layer. you can’t expect real fluids simulation from a game engine, even physicists still can’t do it that well for complicated stuff.

  9. realkruste says:

    It seems the Occulus Rift arrives just in time.

  10. gillesv says:

    So, hopefully this wil mean we’ll finally see an end to those ridiculously unrealistic “fill the room with water to swim to the exit at the top” puzzles we see in *every single FPS*. You know what I’m talking about, the ones where opening a tiny little valve can flood a stadium-sized room in under 10 seconds.

  11. Simon_Scott says:

    What I find interesting with physics cards is that they seem only to offer added eye candy, but promise more. I’d much sooner the physics took a bigger role in games, but until it’s the norm for people to have them, I doubt this will happen. And if all you get is more eye candy, then take-up will be slow unless manufacturers exclusively release integrated graphics/physics cards. It’s a lot of money to shell out just to be able to cut down the bunting in Borderlands 2.

    • Premium User Badge

      Harlander says:

      Isn’t PhysX all done on GPU these days? Can you even get a discrete physics card any more?

      • HadToLogin says:

        Well, it’s quite popular combo to get Radeon for cheapy card and some old GeForce for PhysX.

      • Simon_Scott says:

        I may be a little out of touch then! Not had to buy a new graphics card for years. Point still remains, though, that until a substantial number of physics-capable cards are out there, the tech won’t form part of the game beyond a bit of pretty.

    • Mr. Mister says:

      Well, since the only GPUs allowed to run PhysX are nVidia’s, incorporating PhysX into gameplay ratehr than just extra effects would be tricky, considering you’d most likely alienate the AMD population to some degree.

    • HilariousCow says:

      The way GPU architecture works, it’s mostly a one way pipeline. Once you fire graphics at the gpu from the cpu, it’s hard (slow) to receive the resultant data back, which is why this is unlikely to be useful for anything other than effect.

      I am very willing to eat my words on that. New hardware all the time… but until it’s ubiquitous and not specialist hardware, it’s hard to justify developing core gameplay features from these NVIDIA fancies.

      • Premium User Badge

        VelvetFistIronGlove says:

        The PS4′s architecture is going to make some interesting changes there.

      • RC-1290'Dreadnought' says:

        It IS possible to read data back from the GPU fast enough by now, but it’s still one of the slowest things you can do with a GPU. In the best situation, you’ll be doing unrelated calculations while waiting for the data to travel back over the motherboard to your CPU. That way you can ‘hide’ the delay.

  12. SwENSkE says:

    I want better games, not better graphiX, physX or rabbit drowning simulators.

    As long as devs/publishers use shiny graphics to hide shallow game play while throwing mountains of money out of the windows the looks of a game are not even secondary to me. Tertiary at best.

    • Spoon Of Doom says:

      Think about it. This is not just looks and graphics – proper, performant fluid simulation brings with it a whole lot of possibilities, not only for more realistic setpieces, but more importantly for game mechanics making use of that simulation. In 2D an interesting example of innovative use of fluids is Vessel. I can’t think of anything in 3D, but that’s the point – tech like this, used by a clever game designer, can create all kind of neat puzzles and emergent gameplay, just like other kinds of physics simulations.

      Edit: Fight the typos!

      Edit2: 3D example: From Dust used fluid simulations as gameplay element. Things like this, or a whole lot of other creative uses, can spawn from this.

      • bj says:

        Except none of that will happen as long as it remains exclusive to one brand. It’s incredibly irritating, because there’s plenty of hardware out there capable of running this sort of thing, but the arbitrary restrictions prevent it from being used in any meaningful ways.

        • Spoon Of Doom says:

          I agree it’s annoying that so much stuff is arbitrarily limited to PhysX so often, but with GPU computation becoming more and more important, I can’t imagine it staying that way for much longer, because it can be used for much more than gimmicks, like it has been. Let’s hope I’m not falling victim to wishful thinking here.

          Besides, I’m pretty sure this is more about the new(?) algorithm for fluid simulation, which anyone will be able to implement to create better or more performant fluids than they could with SPH. Or did I get that wrong?

          • SwENSkE says:

            Well, I hope you’re right, but my expectations are rather low.

            Since HL2 and Dark Messiah I can’t remember any games using PhysX for anything other than more ‘realism’. And more ‘realism’ falls in the same category as ‘shiny graphics’. At least in my book.

            I’ve got realism enough in EVE, thanks. Just on a very different level.

            What I want from games is more fun, more innovation, not more realism.
            Don’t misunderstand me, I like great looking games just like everyone else. I still remember buying a new PC for Wing Commander 3 or a 3DFX Card for the first Tomb Raider. But those games were also fun to play and innovative. Today we get eye candy and random shooter No. 378. And I’d rather abstain from the eye candy (and from random shooter No. whatever, of course).

        • Snakejuice says:

          Indeed. NVidia really should open up PhysX to be implemented by anyone. I think it would benefit the whole industry including NVidia.

      • Premium User Badge

        P7uen says:

        Wetrix 2: Men of Water

      • Jekhar says:

        There was also Hydrophobia, if i recall the name correctly. That slightly scifi game where you are trapped on a gigantic ocean liner under terrorist attack.

        • Premium User Badge

          strangeloup says:

          Unfortunately, for all its clever water tricks, Hydrophobia was a spectacularly terrible game.

          • Jekhar says:

            True, and also from a developer having problems accepting criticism. I got it for under 5 € during some steam sale, i still don’t know if it was worth it.

          • Spoon Of Doom says:

            Yeah… Also got it for 3 Euros or something in a sale. You know these occasions where people say “just pass on one time Burger King / one beer in the pub / one of something else and you can get the game”?

            This was the one time where I wished I’d gotten a beer instead of the game.

  13. edwardoka says:

    “From Dust 2″, please.

  14. Squirrelly says:

    Neat tech, but did they have to make such a mess?

  15. Spoon Of Doom says:

    I spotted this the other day and it indeed looks very mouth watering (durr hurr hurr), especially since it appearantly runs in realtime on a single GTX580, meaning it might be performant enough to make its way into actual games in the foreseeable future.

  16. Solidstate89 says:

    It may not be as good as these demos (obviously) but the water effects in Just Cause 2 were always some of the best I’ve seen in modern gaming thanks to its CUDA acceleration.

  17. Superpat says:

    Sort of reminds me of the lionhead studios intro logo in black and white 2.

  18. haowan says:

    It’s only incrementally better than the SPH stuff that’s been going on for ages now. I think I’d be more impressed if this were demonstrated in a more common video game scenario, or on massively larger scales somehow.

    • Spoon Of Doom says:

      I have no idea about the math behind both methods, but from what I gathered, this position based simulation is significantly more performant than the SPH approach in addition to behaving a bit more realistically.
      How much of that is real and how much is marketing speak, I cannot say, because, as I said in the opening of this comment, the math behind both is basically black voodoo magic for me.

  19. AlManiak says:

    Looks good, still implementing it in a tech demo and getting it to work in a full fledged game is a different story.
    I wonder if this thing will ever be possible in multiplayer? I remember that PhysX is not a deterministic physics engine, so to sync that up between players would require some fair bit of bandwith?

  20. Premium User Badge

    Wisq says:

    PhysX has always struck me as something that could have a lot of potential if it were an open standard that everyone could implement, instead of being locked up inside a single vendor’s video card.

    Unless everyone has PhysX, you can’t use it for “core” gameplay aspects, like “where does all the fluid go” (where fluid actually matters to gameplay — e.g. “electrify/freeze all enemies standing in that puddle”) or “does the player get swept away / drown?” etc.

    Singleplayer games can’t use it unless they have a reasonable facsimile for non-PhysX users. And since that would mean maintaining two parallel code paths, not many devs are going to want to do it, not many games are going to do it, it’s not as appealing as it could be, and that slows adoption, feeding back into the problem.

    Multiplayer games can’t use it at all because there would be major discrepancies between the two, and because some players could have an edge by enabling / disabling it. (Heck, I don’t even know if every PhysX chipset returns exactly the same values — it might never be usable for core gameplay elements in multiplayer.)

    The result is, it’s stuck in the “purely cosmetic” realm, possibly forever. Which sorta makes sense when you consider that nVidia is a GPU company, and graphics are ultimately the cosmetic part of the games in the first place. But it seems like it could have the potential to be so much more, if only nVidia weren’t holding on to it so tightly.

    • scatterlogical says:

      I totally agree with this – I think it was a bad move on Nvidia’s behalf to take the tech and only provide it as integrated with their own GPUs. If they continued to sell the PPUs in standalone cards, they could actually capture some of AMD’s customers. I have a crossfire setup in my system – no small investment and not one I’d be keen on replacing anytime soon – but after seeing that video, I would be handing them some cash for a PhysX card if I could. It’s pretty cool tech that seems to be starting to mature, and could become quite purposeful if only they let everyone have access to it.

    • karry says:

      Pretty much means we wait until Intel stucks a physics coprocessor onto the main core, THEN everyone will have it.

      • Premium User Badge

        Wisq says:

        … after the many years it’ll take for non-physics-coprocessor cards to finally go away.

        I mean, even 32-bit still hasn’t gone away, despite its extremely obvious limits (like RAM size) compared to 64-bit.

        (Are any games actually requiring a 64-bit system in their minimum specs yet?)

        • Snakejuice says:

          Planetside 2 does not require 64-bit in the minimum specs but I’ve heard it’s very crashy on 32-bit because of the memory limitation.

          • Premium User Badge

            Wisq says:

            Yeah, with the result that they still have to build and distribute a 32-bit version.

            There are certainly enough games that use enough RAM that they should be declaring them 64-bit only, but so far, nobody seems to want to do that.

      • CitizenDickbag says:

        Except for those of us with AMD CPUs because we can’t justify paying the same price for half the cores at a lower clock speed!

        • karry says:

          “same price for half the cores at a lower clock speed!”
          you forgot to add “70% of Intel’s overall performance”. Seriously, even if there is still a good amount of reasoning to choose between NV and ATI – who in their right mind would buy AMD CPU in 2010+ for a gaming PC ? For a computational workstation i’d give it a thought, but for a home gaming PC ? You are simply mad.

          • SwENSkE says:

            GPU power is important for a gaming PC and so is enough RAM, but the CPU power of an AMD processor is more than enough.
            I’m a (PC) gamer for more than 25 years now and since then I also built my own PCs. I never ever bought anyhing else than the parts with the best price/performance relationship. And Intel is usually very bad in that category.

          • Snakejuice says:

            @SwENSkE

            Have you ever used an Intel CPU? I was like you and always bought AMD because it was “more hertz/cores for the money”. Then I started playing some heavy cpu bound games like Rift and Guild Wars 2, my cpu was bottlenecking me so hard I could only get over 20 fps when no-one else was around.

            So I built a new computer using one of the best AMD cpus at the time (AMD Phenom II X6 1090T), however my cpu was STILL bottlenecking me hard, raids in Rift and WvWvW in GW2 still resulted in sub 20 fps performance so I gave up modern MMOs for a while.

            A year or so after that my computer broke down for various reasons and because I wasn’t happy with the performance anyway and had some saved money I decided to build new again from scratch using Intel (i5-3570K), because I had never had an Intel cpu before (not counting my first computer, prebuilt with a Pentium 120Mhz) and I had the extra money to spare.

            Now I did expect the performance to be a bit better this cpu being newer and all but I was BLOWN AWAY by it’s performance! With the same gpu moved from the old gomputer I was still getting more than double the fps in Rift, GW2 and Minecraft, all very cpu bound games. And all my source games including Counter Strike:GO was pushed to a stable 120 fps vsynced. Even my maniaplanet games now goes at 120 fps unless there are lots of players on the screen.

            TL;DR: If you have a high end gpu lots more games than you might think is limited by your cpu, especially if you have a 120hz monitor. If you are spending lots of money on your gpu, don’t compromise on cpu performance, go Intel.

          • Premium User Badge

            Wisq says:

            Yeah, It’s been my understanding — and as far as I can tell, the general understanding of the PC building community — that Intel has been utterly kicking AMD’s ass for years now. Until now, I didn’t know a single person who would buy AMD unless they were looking for an ultra cheap CPU for a tiny Linux server somewhere.

            Clock speed doesn’t mean as much when one machine is doing way more per clock cycle than the other, and number of cores only really matters for extremely parallelisable things like video encoding. And even there, Intel has acceleration that gives them the edge.

            Essentially, AMD owns the low end and Intel owns the medium and high end. Tom’s Hardware agrees, with the “best gaming CPU for your money” article giving everything under $100 to AMD and everything higher to Intel.

      • SwENSkE says:

        *implying everyone uses Intel

        The last Intel CPU I used was in my 286/12 Mhz about 25 years ago.

    • Spoon Of Doom says:

      I think (or hope, rather) that with GPU calculations increasingly being seen as The Next Big Thing, that sooner or later it will find its way into gaming independently of hardware vendor. There are massive gains to be had, open source solutions for making calculations on the GPU are gaining popularity (not in gaming yet as far as I know, but still), so I can’t (don’t want to) believe that it will stay in this purely cosmetic, Nvidia reliant realm forever.
      One can dream, right?

      • Brun says:

        Realizing the gains from GPGPU usage would require a better mastery of multi-threaded programming than what is currently shown by the game programming (and general software engineering) industry. Most games barely provide proper support for parallelism and those that do often do so inefficiently.

        Remember the biggest advantage of GPU architecture is that it is *massively* parallel, so until computer science figures out how to make parallel programming more intuitive the potential advantages of GPGPU will be limited.

  21. bovine3dom says:

    I was reading the paper for this yesterday: http://www.matthiasmueller.info/publications/STAR2013.pdf

    I don’t quite understand why this approach wasn’t used before; it seems pretty blindingly obvious that you would want to make sweeping approximations in these kind of models. You just want to get the “feel” of a fluid rather than an atom-perfect simulation.

    • Premium User Badge

      Gap Gen says:

      I assume because graphics cards couldn’t handle that many particles, and that bigger particles make the fluid look like jam (jelly)?

      EDIT: Ah, sorry, so it is a different method to the one before. I suppose research takes time to do, so people keep coming up with new techniques. It’s not that we don’t know the equations of fluid motion, but like you say, finding the best approximation that looks right without using too much compute power.

    • Premium User Badge

      Fontan says:

      The paper actually deals with solids simulation, not fluids, and the behaviour is quite different. I admit I still haven’t read it whole, but do they mention a way to adapt that to fluid dynamics?

  22. callmeclean says:

    Could someone explain to me how exactly “middleware” works/is used. Couldn’t all the calculations for physics engines and other middle ware stuff be done in the game engine? Or is it like more efficient to do it at a level closer to the hardware or something. I am really clueless on how graphics cards, drivers and games all really work together. A link to a detailed explanation would be cool if there are any. Thanks to anyone who can help me understand it.

    Oh, and the water physics are bloody amazing.

    • Premium User Badge

      darkChozo says:

      Not terribly familiar with parallel processing, but in general, the reason to use middleware instead of just doing it yourself is that doing it yourself is expensive, takes time, requires expertise that you may not have, and may end up with something lower quality than what the middleware provides. It’s not necessarily better performing, but it often is.

      When it comes to Physx in particular, Nvidia also specifically tweaks their architecture and API in order to improve Physx’s performance on Nvidia cards. Physics is also something that’s rather suited to parallel architectures like those used in graphics cards (basically, physics calculations involves a lot of matrix manipulation, and linear algebra involves a lot of parallel operations), so GPU-based physics is often going to get better performance than CPU-based physics.

      As for a more general explanation, graphics cards are basically a combination of lots of small processors acting in parallel and high speed RAM, with a bunch of special things for handling video. Drivers are manufacturer-provided (well, usually) bits of software that act as a middleman between the hardware and the OS, usually also exposing a specialized API to other bits of software as well. Middleware, if it’s used, sits on the OS and makes calls to either OS or driver APIs, depending on what exactly it’s doing. And then the game engine sits on the OS and middleware and maybe drivers, making calls to all sorts of things.

  23. Quine says:

    Good to see white rabbit in the tub…

  24. rei says:

    I got infinitely more excited by this video when I realized the music sounds like it could be from Mirror’s Edge 2. The first one even used PhysX. I bet I’m onto something!

    • Premium User Badge

      megabear says:

      The music actually is from the Mirror’s Edge soundtrack! =D
      Track name is New Eden.

  25. Jupiah says:

    Perhaps next they can solve the riddle of waterproof protagonists. Hopefully soon the day will come when your video game will climb out of a pool of water and his/her clothing will actually be *wet*.

  26. UpsilonCrux says:

    Reminds me of that family holiday we took as a kid to this place with a wave pool. This has less soiled bandages, verucas and shameless old men though.

    As an aside, neato tech demo.

  27. analydilatedcorporatestyle says:

    GIVE ME HI RES BLOOD CUBES!!!

  28. Premium User Badge

    Continuity says:

    Looks like a load of balls to me.

  29. NthDegree256 says:

    It looks gorgeous, but orthogonal to that: Every 5 months or so, gaming news sites manage to once again frustrate me with the reminder that there is no Mirror’s Edge 2 yet. This time, they did it by sneaking the Mirror’s Edge soundtrack into a physics demo video. Well played.

    :(

    • Premium User Badge

      Continuity says:

      Fear not, Mirrors edge shall be reborn in the glorious VR future heralded by the Oculus Rift.

  30. Premium User Badge

    sonofsanta says:

    I am sad that even our tech demos have to have guns and bullets in them :(

    It also felt a bit weird staring so determinedly at a minotaur-thing’s crotch because that’s where the splashing was happening.

    • Koozer says:

      “It also felt a bit weird staring so determinedly at a minotaur-thing’s crotch because that’s where the splashing was happening.”

      RPS tagline of the week.

  31. Wonko the Sane says:

    > Rapturous.

    I see what you did, there.

  32. Shadowcat says:

    Ever since the invention of physics, which was sometime just before I balanced bricks on a plank to create a see-saw bridge in Half Life 2, I’ve been waiting for a game with proper water.

    Hydrophobia. How did you miss it? People may or may not think it’s a good game, but it certainly is a game, and it was completely based around “proper water” physics. Did you miss it? (How did you miss it?)

  33. Wedge says:

    At last, we can have tampon commercials in realtime.

  34. drcancerman says:

    I think we won’t be seeing this anytime soon. Why? Because of the consoles, has anyone seen the depressing xbox specs? Well, at least we would get more access to a 64bit executable… :/

  35. MadTinkerer says:

    Hmph. Why can’t they use magi-technological seeds to turn that water into puzzle-solving organisms? That said, I’m hoping for Vessel 2 to have even better fluid physics now.

  36. MeestaNob says:

    That looks super impressive, but I still happily bought an ATi card the other day as I don’t like supporting proprietary tech.

    PhysX should be for everyone.