I don't completely understand the last part of the MVP matrix (projection).
In orthographic projection we simply scale the view volume into the default view volume (+-1 on all axis).
In perspective projection we scale and skew the view frustrum into the default view volume (+-1 on all axis) which causes non-uniform foreshortening giving the appearance of perspective that we're used to.
So now, regardless of which type of projection we have used, we have the default viewing volume (+-1 on all axis). So how do we get from this cube to an image on the screen?
I understand that we need this volume for the z-buffer. But how are points (anywhere in this volume) finally transformed from 3d to 2d and is it the same procedure for orthographic projection as it is for perspective projection?