5.1. Determining Camera Occupancy
Camera placement is constrained to a set of acceptable positions in the game. It is desirable to position the camera so that it is out of the way of any geometry objects. When using a dynamic stereoscopic camera, you want to accurately reproduce the movements of the user. For this reason, the camera cannot be placed in locations where it would be possible to locate a static base camera but not the final eye cameras. This placement would lead to results that do match the movement of the player.
Take these constraints into consideration before deriving the two eye cameras from the base camera. To avoid collisions with geometry objects, you must secure regions for the camera that are out of the way.
Some particular camera positions are acceptable. In these positions, there are no geometry objects in the camera FOV between the camera and the near plane. Otherwise, rendering produces artifacts.
For a dynamic stereoscopic camera, you must confirm that a particular camera is in an acceptable position for all possible user positions. For this purpose, the camera occupancy volume can be defined as the frustum from the camera plane to the near plane. The sides of this volume are the planes that stretch from the extreme reachable positions of the camera to the edges of the window. See Figure 5-2.