Spontaneous combustion/collisions(best I could come up for a title)

SUCCESSS !!! see below

This seems long, but it is not, it is easy... had to give all the steps.

1 Concoct a crater by raising a circular ridge of terrain high enough to keep objects within the crater.

. Make it big enough to be able to cut donuts within it running a simple vehicle(box with 4 wheels).

. Floor of crater is at x,0,z – Physics Material on terrain is ‘Metal’ or ‘Wood’.

2 Make a simple Vehicle mass 900, with rigidbody, BoxCollider(physics none), 4 wheelColliders.

3 Create 10 spheres(balls) .74 meter Radius. (<- corrected - had erroneously stated .65 Diameter) . Make balls oblong so they do not roll as much as round ones would.

. Balls mass 15 each, with rigidbody. CapsuleCollider(Physics: wood), radius .65.

4 Place the balls in a triangular configuration somewhat near the raised terrain area, within the crater.

. Stand them on their small end so as to acquire a tight formation, like bowling pins or billiards balls.

. The colliders of each of the balls do not intersect at rest in the triangular formation.

. Place vehicle within the crater as well, where you wish.

. Drive the vehicle like a ‘madman’(person) until you cremate into the apex of the balls triangle.

. Watch all the balls explode in all directions and roll back down the crater ridge.

. Note: balls tend to go straight out in front of fhe vehicle, those on the side(not in direct line of hit) may remain standing or move much less.

. All is well.

5 This is where starts the problem.

. Start pushing any and all individual balls with the vehicle to attempt to gather them

. in an as tight a flat pile as you can anywhere you choose within the crater.

6 All appears well until at some point for apparently no reason at all, you are barely touching

. one of the balls with your vehicle attempting to push it nearer the others and an arbitrary

. count of the other balls just blow sky high and disappear in the Aether possibly including

. the one you are touching???

You can see the objects ricochet in the air as if they are retriggered repeatedly into a series of

deflected collisions(with ‘nothing’) until they disappear in the distance(or the abyss).

Sometimes they come back down to be found on the terrain, sometimes they vanish even after having come down???

What am I doing incorrectly that causes this unwanted behavior?

Has anyone run into this type behavior in a similar scenario?

Can others do this test, and report any findings?

As always, my appreciation.

UPDATE: Nevermind, still does it... it is just unpredictable, except for the fact that the more you bounce the objects around the more chances there are they will blow into cybespace.

Original post

I changed only the colliders and now the objects are staying put. ...at the risk of speaking too soon, I may have the answer, or at least a step in the right direction.

I cannot be the only one who has or will encounter this problem, thus I hope this post will be valuable to someone.

The objects, all exactly the same, clones of one object, have a shape like a bottle of 'Perrier' that allows the collider to protrude outside of the object shape/mesh in one area.

As long as the objects were standing, that area of the collider never touched the canvas/terrain, but as soon as knocked around the collider portion that 'sticks out' of the objects mesh could come in contact with the terrain. I am wondering why they have to sit a period of time before PhysX kicks in? why do they not blow out immediately?

Now that I adjusted the colliders to not protrude at all, the objects do sit on the terrain slightly sunk-in, but none have blown up to the heavens since the change.

The collider protrusion can be eliminated by using multiple colliders on each object, scaling each collider to the area it covers, but how much of a cost would that impose on the game engine? 2 to 3 colliders where 1 should suffice... especially when I may have 100s of these objects concurrently in the view port.

A mesh collider should be a match to the shape but I am not sure I will be able to get the same behavior I get now without using any code at all.

...testing, more testing.