• Create Account

# Resolve Collision

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.

4 replies to this topic

### #1Mr.L  Members   -  Reputation: 125

Like
0Likes
Like

Posted 09 July 2012 - 04:20 PM

Ok I finally figured out how to use SAT 3d

But how can i resolve the collision.

I first wantet to just reset the last step at a collision, but that seams unnatural.

Does anyone have a better Idea?

### #2bpj1138  Members   -  Reputation: 113

Like
0Likes
Like

Posted 16 July 2012 - 07:20 AM

SAT is a good start, but eventually you'll have to figure out the features of the objects that are intersecting, like which edges and which faces. You need references to these features so you can feed that information to the Verlet system which will "deform" these features, and the collision response will follow from that, because the objects will subsequently "fix" themselves through their rigid structures. I'm posting the X-purge demo from 6 years ago. Hit the 'T' key. You'll see the rigid structures. Hit 'T' again, you'll see object features. Try laying the 'X' on its side (ijkl rotates obj) on top of the cube. You'll see periodic impulses to certain edges. That's when the deformation or response is applied. It blinks on and off with the impulses. Hope this helps you.
--bart

### #3eppo  Crossbones+   -  Reputation: 1682

Like
0Likes
Like

Posted 16 July 2012 - 11:33 AM

Assuming your colliding object has some sort of velocity-vector/tangent, you can interpret the separating axis as a plane of impact's normal.

e.g. an elastic response: vec = vec - 2 * axis * dot(axis, vec)

### #4bpj1138  Members   -  Reputation: 113

Like
0Likes
Like

Posted 17 July 2012 - 10:45 PM

right, except that's not gonna solve contact forces...
--bart

### #5Inferiarum  Members   -  Reputation: 717

Like
0Likes
Like

Posted 18 July 2012 - 06:35 AM

Assuming your colliding object has some sort of velocity-vector/tangent, you can interpret the separating axis as a plane of impact's normal.

e.g. an elastic response: vec = vec - 2 * axis * dot(axis, vec)

I guess when objects are colliding there is no separating axis.

And also if we take a very simple example of two spheres colliding and the axis is normal to the tangent plane at the point they touch, your calculations are not physically correct.

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