The idea behind stereoscopic 3D image is to present to each of your eyes a slightly different image. Your brain will (try to) interprete the two images as being two different viewpoints of the same object and deduce the third dimention (depth).
First you need a way to let your eyes see 2 different images.
The simplest is to use a device which allow you to see different image (this is call a stereoscope).
You can also place two (small) images side by side and educate your eye to look "behind" the screen as it is the case for autostereogram http://en.wikipedia.org/wiki/Autostereogram
You can achieve it by colored lenses (this is the case of your sample pic: a blue and a red glass. Through the red glass you can see only the blue image and it looks black and through the blue glass you see only the red image)
Another way is to have a screen presenting alternativally the right and the left image and glasses which sychronouzly hide one or the other eye.
You can also use polarized light and discriminate which part of the image goes to each eye with polarized filter
My guess is that in the case of a print (on whatever surface) you are restricted to the side by side images or the colored ones
Having choosen the technic how to produce the images? look through a window: closing an eye you can superpose on the window a drowing to the underlying image, now (without moving the head) switch eye and start again: you have your two image to be combined: of course you may take photography (replacing your eyes with camera) or use computer to project the 3D object on the window plane