Convex Hull shape questions.

[prasoc]

There is little to no documentation on how to use btConvexHullShape, but I assume you use it as a shape and add vertices using ->AddPoint() right? This code doesn't seem to work at ALL, the objects meet the ground plane, and collide successfully, but its like the collision area is a tiny sphere in the middle of the object, which means the object itself (mesh) clips the plane quite a bit

Code: Select all

	btConvexHullShape* fallShape = new btConvexHullShape();

	dbPerformCheckListForObjectLimbs(1001);
	int limbCount = dbChecklistQuantity();
	int totalVertexCount = 0;
	for(int i = 0; i < limbCount; i++) {

		int vertexCount = GetVertexCountForLimb(i);

		for(int j = 0; i < vertexCount; i++) { 
			btVector3 position(dbGetVertexDataPositionX(j),dbGetVertexDataPositionY(j),dbGetVertexDataPositionZ(j));
			fallShape->addPoint(position);
			totalVertexCount++;
		}
	}

It has some of my own engine and DarkGDK code library implemented, but the code is pretty self-explanitory. Any reason the collision just isn't working? The "totalVertexCount" reaches like 2900, so I know vertices are actually being added.

[Erwin Coumans]

convexHull->addPoint should work as expected.

If you can reproduce the problem in a standalone Bullet demo, someone might be able to help out with the issue.
Thanks,
Erwin

By the way, 2900 vertices in a convex hull sounds extreme, usually there are only a handful of vertices in a convex hull shape.

[prasoc]

I imported a shape that has a smaller number of vertices (like 150 or so), it still does exactly the same thing. Attached is a simple demo showing my problem. Maybe I'm using the wrong shape for the object? a btConvexHullShape should bound the object, is that the "easiest" shape for polygonal collision? I was playing around with sphereical and box collisions and they worked fine. Also, if you click the objects in the demo, it moves the objects using setLinearVelocity()

[Mako_energy02]

In your second for loop there, you appear to initialize a new variable "j", but only check and increment "i" from the previous for loop. Is that intended?