Hi,

I try to do Environment Mapping (in OpenGL). I have some questions about how it works.

The scenario I think I now how to do is the following:
If I have a simple mirror (for example) I need to calculate the virtual viewers position and render the whole scene from that position to a texture.

The problems I have now are the following:

- How can I compute the projection matrix if I have convex or concave or more complex surfaces, i.e. a surface with waves?

- If I render a cubemap for a whole object, should I render each side separatly or have a projection matrix which has a 360° FOV?

- Can I program this using a recursive approach:

SceneNode::draw(int pass) {
    switch (pass) {
    case 0:
        //activate texture framebuffer
        scene.render(pass+1);
        //activate standard framebuffer
        render();
        break;
    case 1:
        render();
    }
}

somewhat of a pseudocode... Would that work with OpenGL?

I hope you can help me clarify my issues and maybe solve them...