Brian Beckman: The Physics in Games - Real-Time Simulation Explained

Play Brian Beckman: The Physics in Games - Real-Time Simulation Explained

The Discussion

  • User profile image
    I like all the download options. Is this new?
  • User profile image
    mcampbell wrote:
    I like all the download options. Is this new?

    Yes and no Smiley The options have been there, just not fully implemented. Now they are. Big thanks to the great Duncan Mackenzie for implementing this.

  • User profile image
    Brian is great, loved the video
  • User profile image
    Thats AWESOME! Big Smile

    Very good video.
  • User profile image
    He should thrive on "Many Core Era!"
    Nice video!
  • User profile image
    1. that's right !!
      when I created a small simple balls engine using XNA using some
      of my (unfinished yet) high school physics I saw that all the time I
      had to make some stupid decisions...
      and even worse, when stuff are moving fast (relatively to the
      anount of time we are looking at) you really need to see when
      were you supposed to stop because of a wall or something and
      calculate everything back there and than simulate what would
      happen until the time you are at now...

      in other words, very simple code turns into a monster...

    2. very very cool post !!!

    3. what's that fractal on the second screen ?
  • User profile image

    That's the best ting I've seen on C9 in a LONG time. It was fascinating and informative.

    We've been discussing the vehicle handling of the DIRT demo on the Codemasters forum this week (because it's horrible on some machines and great on others and nobody can tell us why) and this just gave me more information about game physics.

  • User profile image

    has an endless supply of supernal wallpaper

    and this is the screen-saver companion

  • User profile image


    And what about the mechanics part of the car? The stress in components can be also simulated in these games? A bad move and your gear breaks loose? Or the suspension deformates and changes behavior?

  • User profile image
    Yuppers, the Rigs of Rods approach is a great, easy, simple way to simulate stress, strain, permanent deformation, and even breakage. There are also more traditional "continuous media" approaches -- tensor-based approaches. A bit more complicated mathematically, ok, a LOT more complicated mathematically, but which are the only way to go when you have too many rods in your rig Smiley  The continuous-media physics is actually very beautiful: check out "Theory of Elasticity" by Landau & Lifshitz.
  • User profile image
    Near 44:48, there is talk about the vector function that plots like a salad bowl. I went to race school a few years ago, there was a salad bowl mounted on the dash with a small ball in the bowl. If you pulled too many G's in any direction the ball would fall out at the same time you would lose grip. Around 55:10 there is talk about Bill not being a racing fanatic. I've heard stories about all kinds of moving violations including vehicles from race cars to bulldozers! I'm sure he has a cool driving rig somewhere.
  • User profile image
    Really interesting interview. It's so true that sometimes often people with no legacy in a certain area are able to do the breakthroughs Smiley
  • User profile image
    Excellent video, I loved the demo at the end...
  • User profile image
    That plasma-pong game mentioned at the end is worth checking out as well.  Even if you are not into games, it is a very impressive fluid dynamics simulation.

    Fluid dynamics + silverlight 1.1 = demo based off of the same idea
  • User profile image
    A truely informative and interesting video

    great stuff

    look forward to more like that
  • User profile image
    Very interesting, engaging and even visual.  Nice one Dr. Beckman and Mr. Torre!
  • User profile image
  • User profile image
    This is awesome.  I love hearing from Brian.  He has a great way of talking about a problem from a higher level and still giving a lot of the detail.  I also love the fact that he doesn't start every sentence with "So..."  That must have been difficult for him being around so many people that have adopted that annoying habit.

    In any case I want more from Brian.  You could spend a whole week with him and I would watch.  Thanks Charles for a good video. Try to figure out the formula of what you did right on this video and repeat!
  • User profile image
    I get the feeling a lot of this went Woosh over the interviewers head (or boredom)!

    When talking about divergence, how is this accounted for when the formula is used for tyre manufacturers for cars going low speeds?
  • User profile image
    mrshrinkray wrote:
    I get the feeling a lot of this went Woosh over the interviewers head (or boredom)!

    Boredom? No way. I like to listen to Brian and learn, so that's what I did. Over my head? Sure, some of it. I'm not a physicist and I don't write simulation software. That said, one of my majors in college was math, so I in fact understood what he was talking about...

  • User profile image
    I have known about Rigs of Rods for several months and I like the idea behind it, as it has some potential.

    I have to mention some facts that other may not know or simply forgot.

    First of all, in 2000, two games have used exactly the same methodes to introduces very good game physics.

    The car action-racing PC game "1NSANE":

    ... and "BridgeBuilder" (a strategy game around building train-bridges):

    Both of them use the very same methodes as RoR uses, and both have been released back in 2000. So neither the idea nor the result that it is a reasonable technology are new.

    Why no one else is using better physic engine implementations is beyond me. Most games use a trivial physic engine and that's why most games (car, flight, etc. games) feel unrealistic. For example Flight Simulators physic engine (as of v. 2004) still feels as unrealistic and simple as v. 1995.

    btw., I am still waiting for "Loose Cannon", a GTA-alike game which has been developed by Digital Anvil, now part of MSFT Game Studios.
  • User profile image
    using particles for simulation of other stuff is a cool concept Smiley if you find rigsofrods impressive, you should check out maya8.5 and and its new solver, autodesk nucleus.. (implemeted in maya as ncloth) its a bad (I need to watch my language) cloth solver that is also ment to be used for other stuff..what has this got to do with realtime physics you ask? well maya and nucleus can be used in real time.. its really ment to preview and setup scenes but it also great fun to play with Smiley check it out:

    jos stam is the guy worth remembering as well.. he created nucleus as well as alot of fluid dynamic stuff.. (also implemeted in maya btw) if you look at the silverliht sample above, you'lls se that jos wrote the paper that that demo is build on Smiley
  • User profile image
    Can you please convert these videos to a compress formatt[H]
  • User profile image
    Yeah, Jos Stam did some fine work for the Maya fluid solver. The idea of the semi-lagrangian approach is simple and effective. But he isn't the only one who worked on ideas for real time fluid dynamics. Smiley

    I like NextLimit's particle based fluid/multiphysics approach.

    As someone above mentioned, using rod frameworks in games is older than RoR. Bridge builder and the 3D successor Pontifex were games using the same idea long before RoR. But RoR has a quality of it's own and the new refreshing idea of using these rods for cars and airplanes.

    Somebody said that MSFS is missing a real physical approach for simulating flying objects - that's why I'm flying with X-Plane that uses "proper" physics (blade element theory) instead of lookup-tables.

    What I'm afraid of is that all the CPU power we have now will only be used for destruction modeling in first person shooters. I'm waiting for a game that makes intelligent use of the possibilities. Smashing crates realistically with a crowbar can't be everything Smiley

Add Your 2 Cents