Decomposition of a 6 points / 2 triangles giving strange results

[Pindrought]

I am decomposing a window of a model of a vehicle I have.

For some reason, this is resulting in 8 convex hulls. I don't understand why this would ever be anything other than 1 convex hull.

It also seems to run slow in debug mode, but maybe that is just inevitable due to the STL. It takes about 13 seconds to process this mesh that is just a single face even when I am only using a recursion depth of 2.

I cannot provide the asset, because I do not have the rights to distribute it, but here is an example of code to process the same vertices from the primitive.

VHACD::IVHACD::Parameters params;
params.m_maxRecursionDepth = 2;
VHACD::IVHACD* vhacdInterface = VHACD::CreateVHACD();

std::vector<VHACD::Vertex> vertices = {
	{-0.068349301815032959, 0.77032583951950073, 0.23267269134521484 },	
	{-0.068349301815032959, 0.77032583951950073, 0.23267269134521484 },
	{-0.068349301815032959, 0.24485522508621216, 0.075129270553588867 },
	{-0.57871705293655396 ,0.24485522508621216 ,0.075129270553588867 },	
	{-0.57871705293655396 ,0.24485522508621216 ,0.075129270553588867 },	
	{-0.57871705293655396 ,0.77032572031021118 ,0.23267269134521484 }
};

std::vector<VHACD::Triangle> triangles = {
	{4, 2, 1 },
	{5, 3, 0 }
};

std::vector<double> points;
for (int i = 0; i < vertices.size(); i++)
{
	points.push_back(vertices[i].mX);
	points.push_back(vertices[i].mY);
	points.push_back(vertices[i].mZ);
}

std::vector<uint32_t> triangleIndices;
for (int i = 0; i < triangles.size(); i++)
{
	triangleIndices.push_back(triangles[i].mI0);
	triangleIndices.push_back(triangles[i].mI1);
	triangleIndices.push_back(triangles[i].mI2);
}

bool success = vhacdInterface->Compute(points.data(),
										points.size() / 3,
										triangleIndices.data(),
										triangleIndices.size() / 3,
										params);
if (success == true)
{
	int n = vhacdInterface->GetNConvexHulls();
	//^n = 8
}

Am I doing something wrong here or any ideas? Could all of the points lying on the same plane be causing issues here?

[jratcliff63367]

I am currently in the process of moving, and have very, very little bandwidth. If I can I will try to look at this but it might take a while. You're correct, it does not seem like correct behavior. That said, feeding something into the library that doesn't actually need decomposition isn't a typical use case. I'll let you know if I have time to look into it.

[Pindrought]

Thank you very much for your reply. I hope your move goes well.