Stereo Vision is the primary method that the human visual system uses to perceive
depth. Stereo vision is common to most predatory species, including man, and it is very effective
at judging distance. The principle is actually very simple. If you are focused on an object far away, both
eyes are looking almost straight ahead. If you are focused on an object closer, then
both your eyes are pointing slightly inward. If you are trying to focus on something right on
your nose, then your eyes are pointing severely inward and that is when the person doing it is "cross-eyed".
What your eyes are actually measuring is the disparity of corresponding images on the two retinas.
Your brain can sense how far inward your eyes are pointing and it knows how far away the
object of focus is. You brain basically triangulates the position. Have you ever wondered how
astronomers know how far from the Earth the moon is? well they view it from two different locations
on the Earth simultaneously and by recording what angle they viewed the moon from they can use trigonometry to calculate how far away the moon is.
Creating A Stereo Image
Creating an image for stereo vision is easy. For photography you can take who photographs
simultaneously from two cameras that are side by side. Or you can use a special stereo-camera, which
is basically the same thing in one box.
For computer games, you simply render the scene from two slightly offset locations.
The end result is a stereo picture, or picture pair. Here is an example Quake picture which I will
explain how to view later.
The above two images look the same, but they are slightly different in the position from which the
screenshot was taken - one picture was taken for each eye. If you can display one image to one eye, and the other to the other eye, then the brain will put both images together to make one 3D picture complete with depth. To
explain, lets super-impose the two example images above.
In this example, the focal point of the picture is the window frame farthest away.
The focal point of the image depends upon what the two cameras are pointing at when the
stereo image is taken. After the two pictures are merged together, there is very little
disparity on the window frame far away (close to the focal point), but the arch closer to us has
a double-exposure effect, indicating higher disparity. For this example, ignore the gun and the
HUD, because these move with the player and therefore don't show any 3D data in the stereo image.
The main problem with stereo imagery is trying to display one image to one eye, and a different
image to the other eye, without the two getting mixed up. Let's check into a few methods...