3 replies to this topic

[Naked Shooter]

Hi. I'm trying to get the frustum vertices by using the perspective projection matrix's inverse. I'm trying to take a cube with the vertices { (-1,-1,-1), (1,-1,-1) ... } and multiply each vertex by the inverse of the perspective matrix, hoping to get the frustum. This isn't working, however. I am aware that there's a lighthouse3d tutorial, and I just might bite the bullet and go through it, but I'm just curious why this isn't working.

glm::mat4 projection(0);
glm::mat4 projectionInv(0);
//the frustum in normalized coordinates
vector<glm::vec4> NDCCube;
NDCCube.push_back(glm::vec4(-1.0f, -1.0f, -1.0f, 1.0f));
NDCCube.push_back(glm::vec4(1.0f, -1.0f, -1.0f, 1.0f));
NDCCube.push_back(glm::vec4(1.0f, -1.0f, 1.0f, -0.0f));
NDCCube.push_back(glm::vec4(-1.0f, -1.0f, 1.0f, -0.0f));
NDCCube.push_back(glm::vec4(-1.0f, 1.0f, -1.0f, 1.0f));
NDCCube.push_back(glm::vec4(1.0f, 1.0f, -1.0f, 1.0f));
NDCCube.push_back(glm::vec4(1.0f, 1.0f, 1.0f, -0.0f));
NDCCube.push_back(glm::vec4(-1.0f, 1.0f, 1.0f, -0.0f));
/* Get the current PROJECTION from OpenGL */
glGetFloatv(GL_PROJECTION_MATRIX, (GLfloat*)(glm::value_ptr(projection)));
//calculate the inverse of projection matrix. Hopefully, this'll get us the frustum when multiplied
//by coords of a cube.
projectionInv = glm::inverse(projection);

//push back vertices
for(int i = 0; i < 8; i++)
{
  glm::vec4 tempvec;
  tempvec = projectionInv * NDCCube.at(i); //multiply by projection matrix inverse to obtain frustum vertex
  frustumVertices.push_back(Vector3(tempvec.x, tempvec.y, tempvec.z));
}

Edited by Naked Shooter, 21 October 2012 - 04:16 PM.

[Ohforf sake]

The projection matrix uses a projective space with homogeneous coordinates. This means that you have a 4th component. To get a vector from your normal 3 dimensional space to this projective space you attach a 4th component and set it to 1. In your case it is 0 for some vectors which is incorrect.
After the transformation with the inverse projection matrix the homogeneous vector must be transformed back from projective space to your normal 3 dimensional space:
tempVec /= tempVec.w

Also note that OpenGL returns the matrix column major IIRC and glm _might_ store it row major (I have never used glm). In this case you would have to transpose the matrix.

[larspensjo]

Also note that OpenGL returns the matrix column major IIRC and glm _might_ store it row major (I have never used glm). In this case you would have to transpose the matrix.

I can confirm that glm stores it in the same order as OpenGL (column major).

[Naked Shooter]

Yes, that works! I knew it had something to do with w.. Thank you Ohforf!!