Hi.
I wonder if its possible to render scenes/objects to texture memory (video card memory) without render to system memory first?
I want to simulate reflections in this way by moving the view port to the polygon that should reflect and render the scene from that view port to a texture, later using this texture from the origin view port to render the scene.
Is better for me to calculate the intersection between the ray projected from the eye to the reflecting polygon and from the reflection polygon to the other objects in the scene.
It seems more logical to take advantage of the 3D hardware to this kind of thing, but is it fast enough if I cant render direct to video memory?
/Dew
I think if we gather up all the people that want to do this we could put enough pressure to get this in the next version.
http://www.opengl.org/discussion_boards/ubb/Forum7/HTML/000145.html
And to think I was told I was the only one that wanted pointers. 
If you render in OpenGL, then it’ll automagically end up in video mem. Just use glCopyTexSubImage2d() to update the texture from the framebuffer. It’s fast on nVidia hardware at least.