Jump to content

  • Log In with Google      Sign In   
  • Create Account


Swept sphere, how to properly detect multiple contact points?


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
No replies to this topic

#1 pondwater   Members   -  Reputation: 191

Like
0Likes
Like

Posted 28 February 2012 - 04:05 PM

I originally used Kasper Fauerby's swept-sphere algorithm many of you are familair with: http://www.peroxide....n/collision.pdf

I'm now attempting to fix the issues associated with multiple contact points such as jittering in corners and getting stuck where surfaces 'pinch' together.

I'm attempting to store multiple contact points but my current problem is that if I keep my sphere a small buffer distance from contact (method used in Kasper's algorithm), my system will rarely (if ever) detect more than 1 contact point.

So I have to remove this buffer distance, but then how do I deal with the potential intersections that may occur do to floating point precision error.

At first I was thinking about only allowing movement if the velocity is parallel to the surface. But this wont work for multiple contact points.

My general algorithm will store the lowest t valued contact point per triangle. So Each triangle can return a maximum of a single contact point. And then I will sort the list and use all the contact points sharing the lowest t value.

Can anyone suggest any decent resources for swept volumes and multiple contact points?

Sponsor:



Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS