As an essential preamble i'll say that you have to know that, to take a 3D picture, you need a camera able to take two shots at once from two point distant a few centimeters (the distance between our eyes).
The elaborations given by every shot don't give 3D images, but at most images with a slight relief effect, concretely they can't realize images with additional informations ad the true 3D images.
But what if we don't have a camera that can take 3D pictures, that are those with two lenses, how can we do?
The process is simple but can be applied just with stati object, such as landscapes, views, statues, monuments, etc...
During the shot we must reproduce the distance between our eyes. To obtain excellent results there shouldn't be any movement in the framed scene, and that's guaranteed ony by two synchronous shots. Moreover the sliding must be on the same plane, absolutely, or we'll have a poor result.
Thre's a support called "macro rails" thanks to wich is possible to move your camera on the same plane of a maximum of 10 cm,
In post production, once realized the two pictures, we have to process them with a simple software able to create 3D images called ArBaGlyph, it is a freeware and you can find it on this website: http://www.arba3d.com/
After the application of the 3D effect we'll be able to see our pictures with the red/blue glasses, you'll need exactly those with this lens color, those that you get at the cinema (with black lenses) doesn't work for the purpose to see 3D pictures realized with photoshop.
The final result is spectacular, you may watch the upper image with your glasses to understand what i'm saying.
I've got to thank Francesco Marzoli who realized the original tutorial that i used to realize this, in his useful guide, to be simpler, the effect is explained with a picture with a simple background, let's see what it changes in a picture with a more complex background.
Here you can find Francesco Marzoli's video tutorial http://www.youtube.com/watch?v=uxiKHRSIejM
Steps to realize 3D pictures:
1°) Let's open a picure with Photoshop CS5, if your picture was taken with a modern reflex and you haven't a very powerful video card i suggest to resize the image at 2000 pixels (or less) or at a certain point in the tutorial you'll be blocked by an error message.
2°) Theoretically: We need to move away the main subject from the background, with a "map in grayscale", the background should have a darker color while the elements that will have to appear closer should have a brighter color. for example in the picture above the sky should be in a pretty dark grey, almost black, the sea a medium gray and the main subject (that it's me) and the rock of a lighter gray.
Pratically: In addition to the background level you'll have to create another layer (level 1), then select again the "background" level, with the rapid selection tool, deblur the main subject from the background, if you selected the picture background go to "level 1", take the bucket tool and fill the selection with the dark grey, then go back to the level called "background" and invert the selection, go back on the new level and using the bucket fill it with a brighter gray.
In my picture this is the result.
3°) Having done this, to avoid haveing too sharp outlines and get a more realistic effect, select the new level, go on Filter -> Blur -> Gaussian Blur (set the value at 2).
4°) Now, always on "level 1" selected let's go on 3D -> New Mesh From Grayscale -> Plane, as you can see it creates a new advanced object with inside the 3D options.
5°) Now let's paste the original picture inside this result, deactivate level 1, select "background" level, let's click on Select -> All, let's go on Edit -> Copy, let's go on Select -> Deselect, to remove the selection once copied, now we have to paste our contents, reactivate level 1 and double click lower in the menu on Texture -> Diffusion -> Level 1 (as in the picture on the right).
6°) Now the file of the advanced object it's open, let's paste here our picture, go on Edit -> Paste, now go on File -> Save, again on File -> Close.
Pay attention, we are not closing the project we are working on, but just the file of the advanced object of photoshop, if we don't save and close this file we can't go on.
7°) Let's go on the 3d panel, if you can't see it just go on Window -> 3D
8°) From the 3D panel on the rendering settings click on edit, on face style choose unlit texture, click on ok, now go again on edit and enable the last option, on stereo type choose Red/Blue and on Focal Plane set 0 while in parallaxe set 100, this is the essential value i was talking about, in practice the higher the value the greater will be the depth of field. The focal plane determine the distance of the subject, forward or behind regarding the focal center of the frame.
9°) Now we have just to deselect the background level, cut the image in the way to avoid haveing a white outline and save the file in the format that we prefer. This way if you originally charged an image of 2000 pixels on the bigger size, you'll have a 3D image of 1200 pixels on the bigger side.
The origins of 3D photography go back to 1600, if you want to to know more read this article: Stereoscopic vision, the first 3D was made in 1600!  - Hstory of photography