So it doesn't know anything about what's behind the player's viewpoint, nor does it actually understand what's on-screen.