05-18-04, 06:34 PM
Can anyone tell me how you would go about calculating spherical collision patterns? I've been playing a 3d pool game that is very realistic and I am amazed at how life-like the ball collisions are.
Any info on this subject would be greatly appreciated.
05-18-04, 07:12 PM
Everything except the collision-detection should be pretty basic.
I think that if ball A hits ball B at point P on the surface of A, the direction of change in momentum on B will be normal to P on A. The magnitude of change will be proportional to the momentum of A times cos(v) where v is the angle between the momentum-vector of A and the normal through P also going through the center of A.
I'm not sure it works like this tho, but it seems reasonable.
05-19-04, 01:52 AM
Im no mathematician although the subject fascinates me so you may have to help me out here a little if thats cool. I think I understand what you are saying and it seems reasonable, however it must get massively complicated when 'spin' is applied to ball 'A'.
There must be a standard algorithm to calculate all possible outcomes of the collision of two spheres and there subsequent trajectories after collision?
You maths guys amaze me incidentally :tup:
05-19-04, 06:20 AM
Yeah, wow, i didn't think about that.
Working out a solution for the spin imparted on impact can be pretty complicated depending on how real you want the situation to be. I'm sure there's a good approximation tho that isn't so complicated. It might work if you let a fraction of the kinetic energy of the balls disappear at each impact and then let a fraction of that, depending on sin(v), be transferred into "rotational energy".
If you really want to do this, there's a part of highschool-physics that teaches what happens when two objects collide. What an elastic and unelastic collision is, etc. It requires only basic algebra.
05-19-04, 07:41 AM
It just fries my head. How can anyone be clever enough to work this out to come up with a completely accurate game???
Thanks Anders for taking the time to ponder this with me.
05-19-04, 08:35 PM
I found a really good link if anyones interested. Gives the whole algorithm.