Alright, I have a 3d project that I intend to use prerendered backgrounds in for aesthetic reasons.
So I started by observing what techniques were used back when prerendering was popular.
So far I've mostly paid close attention to the techniques utilized by chrono cross.
I've observed that chrono cross used several distinct methods to bridge 2d and 3d, one obviously being orthographic projection, and the other achieves a feeling of perspective along with a panning camera (but given the issues I've run up against I have a feeling it was still orthographic but they scaled the objects as they moved into the background).
I've done some tests of my own and I have a plan for tackling orthographic scenes, and scenes that justify using an entire 360 degree environment texture, but I haven't figured out how to efficiently approach a perspective scene that likely won't use more than a quadrant of the environment texture space.
My thought process is that if I can match a background to a 3d scene with a wide fov, then somehow I should be able to narrow the fov and pan around within the bounds that the original camera would have captured. In practice there seems to be some distinct problems. Technically I could make a large viewport with a wide fov camera that captures the entire scene and then crop it inside a smaller viewport and follow a character by moving around the inner viewport. But I feel like this should be able to be achieved without buffering more than what should be displayed on the screen.
