The key to stereoscopic vision is depth, and our brain will happily take care of that for us, providing our eyes are given the right information in the first place. This is exactly how those red-and-blue glasses work -- each color filters out part of the image, giving each eye a slightly different view. The brain puts the two different images together, and those blue-and-red blurry images turned into a fantastic 3-D comic, or movie, or TV show.
Stereograms, also known as Magic Eye pictures, use seemingly-random patterns of dots but rely on the viewer to cross his eyes in just the right way, or to look through the image until the eyes see just the right part and allow the brain to decode the hidden depth information.
Both methods have their disadvantages, of course -- the red-and-blue glasses make it difficult to show color in the 3-D image, and viewing stereograms is an art in itself. Neither method is entirely suitable for playing games.
Nevertheless, the underlying principle is exactly the same: creating and controlling those two different points of view. But just how easy is it to create these two separate images, one for each eye?
The answer is all about how games are created. Not so long ago, the graphics we saw on our computer screens were carefully drawn into the computer -- every single frame of animation, every different view of a character. If you wanted a dinosaur in your game, you sat down and drew the different views of a dinosaur into the computer.
Nowadays, games designers sit down with a 3-D graphics package and design their dinosaur in three dimensions. Once that's done, they needn't worry about the different views -- the computer has a 3-D model of the dinosaur in its memory, and the game simply works out where the player is looking and draws the correct view of the dinosaur using the 3-D model. In fact, everything you see on your screen in a modern 3-D game is produced the same way; the game is like a gigantic 3-D model. The computer works out what it needs to display on your screen and generates the appropriate view.
Since the computer is quite happy to create one point of view, there's no problem shifting the viewpoint slightly and creating another point of view. And after that, all you need is a way to get the correct image to the correct eye.