For real 3d effect, some trick must be applied to supply slightly different images to the right and left eye. Several techniques were invented for that purpose, they all have their pros and cons; I will discuss some of them.
In order to create a stereoscopic image, you normally need a left/right image pair, one for each eye. (There are exceptions, e.g. random dot stereograms (RDS) need real depth information.) Creating such image pairs seems easy at the first moment but it turns out that you need to respect some rules and gain some experience in order to get decent results.
It is important that stereoscopic images should be designed with the applied projection (size and view distance) in mind: images which look nice on your computer display may do really bad on an IMAX-sized cinema screen and vice versa.
When creating left/right image pairs, artificial graphics (using e.g. POVRay) has some advantages: You can easily place the camera at exactly defined locations pointing into an exactly defined direction, you can do a lot of experiments fairly fast and cheap and you need not use perpendicular camera angles (as with real cameras) which can be used to avoid cut-off if the left and right images are to be mapped onto the same projection plane.
In order to actually create left/right image pairs using POVRay, you need to supply the proper camera parameters. Read more...
After having created stereoscopic left/right image pairs, some special projection or display method has to be applied in order to get the desired 3D effect. Various methods have been invented for that; see the list of the most common ones; some of them are discussed in more detail below.
Anaglyphs are these usually red-greed or red-cyan images which have to be viewed with colored glasses (details on principle).
Anaglyphs can easily be created from stereoscopic left/right image pairs and viewed on any normal computer screen as well as printed paper; the only requirements are (fairly cheap) anaglyph glasses. In order to get an anaglyph, simply join the left and the right image using different color channels. Read more...
Small full-color 3d images can easily be presented using the parallel-view or cross-eyed view approach (details on principle) using next-to-each-other image pairs at a distance of less than 6cm. Read more...