your mother said:
rocketpig said:
omgwtfbbq said:
Physics engines are generally locked at a particular "frame rate". I've done some programming with Open Dynamics Engine, and I've found nasty things happen when you don't lock the physics engine at a particular rate. The basic idea is that a physics engine will work out the forces on a particular object, and then work out their accelleration, from that work out their velocity (based on their previous accelleration and the timestep) and then work out their position based on the velocity and their previous position. The accelleration is only applied at one point and the velocity is the same for the entire timestep. Therefore a large timestep will mean that a large accelleration applied over a short time (such as being hit with a baseball bat) will be stretched over the entire time period, causing a disproportionate amount of accelleration. So not only does fixing the timestep make sense (otherwise the physics behaves differently when there are more items on the screen, definitely not something that we want), but also a small timestep will see a benifit even when the graphics are locked into a much lower framerate. Basically, smaller timesteps will mean much smoother curves for velocity, accelleration, and displacement, and much more realistic movement. As for the 360 steps per second number, I have no idea where that comes from @larry: thank you for saying that Forza won't even make 10k in Japan. This means it will be the first of your completely ridiculous predictions that will be proven false. Can't wait for the numbers! |
Thanks for the explanation. I wasn't sure about it but I remembered reading it was locked at 360fps. BTW, it's been repeated several times by Turn 10 that the physics engine is running at 360fps. They felt it was necessary since the speeds will be so high at tracks like Nuremburgring, anything less would mean that your car was travelling too far in between physics cycles. Here's one example where they state the 360fps claim: http://xbox360.ign.com/articles/766/766507p1.html |
OK, the explanation was pretty self-explanatory, and the article states 360 refreshese per second, which makes enough sense. Thanks for the explanation and link. What I can't understand is the following: If the physics is refreshed 360 times per second to make the simulation more subtle and realistic, the output (feedback) is still rated at 60fps, because that is what the visuals refresh at. In other words, even though physics-wise you get 6x the refresh rate than that of the visuals, you can only react to what you see onscreen, so your response time cannot be anywhere close to 360x per second, and should be lower than 60x per second. ? |
yes there is still a difference, because the movement will follow realism more closely.
imagine someone throws a ball, the movement of the ball follows a parabola (assuming no air resistance)
now imagine that instead of a smooth parabola, it follows a straight line at the tangent of the parabola for a small amount, then another straight line, and so on. This is how the simulation works. Obviously in this case, the ball won't land in the same place, the path of the ball will only be an approximation to the actual ball's path.
Now if you make the lines 1/6 as long, but only draw to the screen every sixth position, although the ball will move the same speed and it will look the same, the ball will actually be following a different (more accurate) path.
that's why the speed of the simulation is important even if not every frame is drawn.
