after having gave up on various algorithms for computing contact forces, finding that even examples i found in some cases still got stuck and the general complexity of the whole thing, ive decided to go with a projection method. i succesfuly made a simulating with colliding spheres with a soft body aswell, with all of it coming to rest and ive started with non-spheres. Atm i have a simulation with multiple colliding boxes that comes to rest, the problem however is in my collision detection method: Basicly, im transforming one boxes coordinates into another boxes coordinate base and checking for penetrations with each edge, and choosing the smallest penetration to move the boxes apart. However when two boxes are on rest on eachover, and one is slowly sliding so that at one point with numerical innacuracy, they have the exact same coordinate, the boxes end up penetrating eachover as shown in the picture: http://denvish.net/ulf/080607/73855_interpen.jpg any suggestions on getting with of this problem? [Edited by - luca-deltodesco on June 9, 2007 7:11:18 AM]

fixed the problem by allowing for numerical innacuracies, rather then it having to be penetrating, it has to be 'almost' penetrating and it solved the problem perfectly.

ive now added circles into it:

http://denvish.net/ulf/090607/48408_ball4.jpg