Slashdot Mirror


Havok Team Interviewed

Chris writes "There's an interview up with the Havok Team on FileFront, talking to Chief Technology Officer Steve Collins about his company's physics engine. Questions are about development of the engine, getting developer support and the demands they have, and research. The Havok physics engine is responsible for allowing players to lob toliet bowls at unsuspecting Combine in Valve's Half-Life 2 and powers several other popular titles." From the article: "the realistic portrayal of characters is what we hope will define the next generation of games. You're going to see a lot more soft-body dynamics, hair dynamics, clothing simulation and all that cool stuff."

27 comments

  1. Way to dig! by suckamc_0x90 · · Score: 3, Funny

    I'm glad they asked the questions foremost on everyones minds:

    "Q: Could the engine be perfected, theoretically, to correctly display how the fat bounces around on a really fat woman during sex?"

    1. Re:Way to dig! by bersl2 · · Score: 1

      The question then arises: Are they going to do hands-on research, or are they merely going to review others' data?

  2. Halo2? by ShawnMcCool42 · · Score: 1

    Q: At what point were you most happy with the successes of the engine?

    A: The release of the first games with Havok was a huge boost for everyone involved, but I guess that this holiday season has seen the release of many titles that we've been eagerly waiting for, like Halo-2, Halflife-2 and Medal of Honor Each year we see developers getting more and more value out of Havok physics, pushing it harder, and incorporating it more and more into their game design, so it's hard to point at any one success.

    I thought Halo 2 used a proprietary physics engine.

    1. Re:Halo2? by zwaffle · · Score: 1

      yeah I was surprised too, I'm pretty sure Halo2 (like DOOM3) use their own physics implementation.
      I checked the Havok website and nowhere do they mention Halo2.

    2. Re:Halo2? by hollismb · · Score: 1

      Yeah, it does. As a matter of fact, that was pointed out in an episode of X-Play last night, where they showed the physics of Half-Life 2, Theif, Hitman, Halo 2, Psy-ops, and more. If the dumbasses at X-Play know, you'd think the people who actually developed the engine would too.

    3. Re:Halo2? by zwaffle · · Score: 1

      some reviews mention halo2 has its own physics engine, some mention havok.

      Again, go to www.havok.com (check http://www.havok.com/company/careers.php to see how much fun those guys seem to have), and show me where you see a reference to Halo2...

    4. Re:Halo2? by hollismb · · Score: 1

      Did you not actuall read my comment? Like I said, yes, Halo 2 uses it's own physics engine, which X-Play got right.

    5. Re:Halo2? by zwaffle · · Score: 1

      Heh, I did read your comment twice.
      But since you quoted games that do use havok (HL2, PsyOps) I thought your "yes it does" meant that Halo2 does use Havok.
      I thought the rest (about the dev knowing what they code) was some twisted third degree sarcasm :P

  3. Destructing Environment by superpulpsicle · · Score: 2, Interesting

    What would it take to create an engine that allow the destruction of the entire environment. Sure it's nice to toss toilets at a wall, but I'd like to see the wall collapse after 20 hits.

    1. Re:Destructing Environment by over_exposed · · Score: 1

      Mostly a bunch of mathematical calculations that can be translated into realistic graphics and physical reactions understandable by the human eye. Mostly...

      --
      "The object of war is not to die for your country, but to make the other bastard die for his." - Patton
    2. Re:Destructing Environment by TychoCelchuuu · · Score: 1

      Soldiers: Heroes of World War II or Soldner: Secret War let you blow up the enviroment. Warning: Soldner sucks. So I'd stick with Soldiers.

      --
      Against stupidity the Gods themselves contend in vain.
    3. Re:Destructing Environment by papadiablo · · Score: 1

      I wouldn't imagine making it collapse after 20 hits would be difficult, nor would making it collapse in a seemingly realistic way. The real kicker would be if you could take out enough of the supporting structures and make the building fall. And not only fall, but make it fall realistically as well, that is, make the top floors crush the bottom floors under them. That would be a sight to see.

    4. Re:Destructing Environment by Anonymous Coward · · Score: 0

      Probably easier than designing a game large and convincing enough that doing so wouldn't totally obliterate the rest of the storyline.

    5. Re:Destructing Environment by bynary · · Score: 1

      Ever play Red Faction?

      --
      http://www.bynarystudio.com
    6. Re:Destructing Environment by johannesg · · Score: 1
      What would it take to create an engine that allow the destruction of the entire environment.

      Lots. For each object in the game you would need to know how it breaks apart. All parts stay in the game world, so they must be separately and correctly textured (i.e. each brick in a wall becomes a real brick instead of rendering lots of bricks as one object). And the game engine must keep track of all this information too, and save it when you leave the area if there is any chance if you coming back (people quickly rebuilding their houses when you cross a load-boundary is not very realistic). This would lead to massive save games, and much longer load/save times when changing areas. In other words, it isn't very practical to do this.

      Which is sad, because it would be pretty cool...

    7. Re:Destructing Environment by irc.goatse.cx+troll · · Score: 1

      Havok can do that now, its just not feasable on the machines we use now. Currently you can create a wooden structure with 'real' boards that when taken appart, it falls entirely accurately. But to do so to every object in even a single room would create a massive load on your processor. I expect the tech will advance within the next 10 years to the point where at the least, dual cpu is standard for gaming, and one cpu can be used for more dedicated physics work. Maybe even an onboard FGPA or two that can be reprogrammed as part of loading the game to work as a physics coproessor or a facial expression engine like the PS2 has.

      --
      Pain lasts, kid. Its how you know you're alive. Sometimes I think this growing up thing is just pain management-TheMaxx
    8. Re:Destructing Environment by alphaseven · · Score: 1
      What would it take to create an engine that allow the destruction of the entire environment. Sure it's nice to toss toilets at a wall, but I'd like to see the wall collapse after 20 hits.

      I read an interview with John Carmack where he commented on entirely destructable environments. He said you could do that now but you couldn't use current lighting systems, so graphics would look very dated.

      I've been reading reviews of the new game Mercenaries, everything is suppossed to be destructable there but I haven't played it myself. And Katamari Damacy has whole towns where you can pick up every object, not a lot of "physics" though.

    9. Re:Destructing Environment by Anonymous Coward · · Score: 0

      I don't think this is a technology problem so much as a game design problem. Think about all the FPS games you've played. Practically all of them involve finding your way around various obstacles fairly regularly. If you could just blow a hole through a wall to get to where you want to go, you're eliminating a large portion of the game.

    10. Re:Destructing Environment by zwaffle · · Score: 1

      I'm actually trying to implement a game that does just that.

      For a fully destructible environment, you have to:

      - detect contacts between thousands of objects.
      Most of the classic sorting structures like BSP are optimized for static world only.

      - handle those contacts:
      either impacts - easy to solve
      either resting contacts - hard to solve

      E.g it's easy to model a thousand pieces flying up in the air, the hard part is to have those pieces come to rest realistically on the ground as a big stack.

      - when you have impact, figure if and how objects break. That's relatively easy.

      - when you have resting contact, objects can still break due to static load. That's very hard to do in real time.

      E.g., It's easy to model a bullet piercing a wall. But it's difficult to model a shelve that collapses because you've put a weight on it.

      "Mercenaries" uses a mix of semi-realistic environment destruction (some buildings break into a few predefined pieces that fall realistically and come to rest on the ground) and fake destruction (some buildings break down using a realistic animation, but no physics is involved, very much like in MechAssault-LW).

  4. Fat bottom girls make the rockin world go round by Anonymous Coward · · Score: 0

    More cushin for the pushin!

  5. How hard would it to get a limited function SDK? by happyhippy · · Score: 1

    I would love to incorporate some basic physics in my small non-commercial demos. For example just basic gravity and limited objection interaction.

  6. Re:How hard would it to get a limited function SDK by orion024 · · Score: 1

    You would get that by going to http://www.ode.org/ ;)

  7. in-game animation by Anonymous Coward · · Score: 0

    The next big this I'd like to see in games is No Pre-recorded animation of game characters. Of course, cutscenes are ok, but everything else should be simulated in real time. Every character is a ragdoll, only alive and using artificial intelligence to stand up, walk around, etc. Kind of like real life robot programming. Instead of a recorded animation where a character walks up a set of stairs, it has to actually move it's legs onto each step, without losing it's balance.

    1. Re:in-game animation by radio.cgt · · Score: 1
      Like this?

      right now it only works for pre-rendering stuff (afaik), but it works on the principles that you're describing.

    2. Re:in-game animation by fixmyship · · Score: 1

      Yes. (I'm the AC). Naturalmotion does a good job at what it does, but right now, as you mention, it's only to record animation, mostly falls and collision between people or objects.

  8. Re:How hard would it to get a limited function SDK by Anonymous Coward · · Score: 0

    or http://www.tokamakphysics.com/

  9. Mac port by ComputerSherpa · · Score: 1

    Interesting article. The guy doesn't mention the reason why they backed out on the Mac port of Havok, resulting in the cancellation of Uru for Mac.

    --
    Information wants to be anthropomorphized!