German pages.


Page up
Go to higher level.


Stereo photography
Taking pictures in 3D.

Modified: 05.11.2003

Many years ago (when Germany was still split in two parts) I got an optical experiments set from "the other side". This set was a real treasure chest, among other things you could create a stereo slide viewer from two single slide viewers. And there was one stereo slide of an amazing quality that you could watch with this.

Some years later I managed to buy a "Viewmaster", a small device to watch very small slides that are arranged on a disc of paper. The disc is rotated so that you can always see a pair of stereo slides through the lenses. This is also one of the very old (and quite famous) examples of stereo photography.

I had the idea that it should be relatively easy to create such stereo slides, well, at least they are just two photos that were taken with two cameras mounted in the distance between the human eyes. After a short search in the internet I knew two more things:

  • I was not the first to have this idea.
  • There are many more possible solutions to create stereo images, each of them in its own range of costs and quality.
Here I will present the common techniques, publish some do-it-yourself descriptions and give you some practical hints on how to create good stereo images.



The three dimensional impression that we get when we look into a landscape arises from the evaluation of information from both of the images that we see with our two eyes. The brain evaluates each single images as well as the differences between them. Even from having a look at one single picture we can gather information about the depth. The following hints are evaluated by the brain to extract this information:

Texture Gradient:

The density (or the size) of the single objects on the picutre. With an increasing distance from our point of view the objects look smaller. Also the distances between to far away objects seem to be smaller. This leads to the impression that far away areas are far more dense and compressed. The image to the right show this effect.

Parallaxis of Movement:

If you move your head then the images of nearby objects move faster over your retina than those of far away objects. With a growing distance the movement seems to slow down (for example if you look out of your driving car). The evaluation of images that have been taken after each other can provide information about the distance to the shown objects. The picture to the right shows two objects that are shifted in front of the eyes. The angle between the both positions of the nearby object is much bigger than the angle between the far away one though the distance the objects have been shifted is the same.


The differences between both single pictures provide an essential part of the depth information of what we see. The distance between the human eyes is usually about 6.5cm. If we watch some object we always see two images from a slightly different point of view. Each of them is a two dimensional image of the environment so from each single picture we cannot create a complete three dimensional representation. But if you compare both single images you can conclude some depth information from the differences between both of them. The principle is the same as with the parallaxis of movement but it is not necessary to move the head because we already have two images with a different viewpoint.

The upper image shows that the distance to both objects can be derived from the different viewing angles in the single images.

The second figure shows that a single image is not enough to do this. Different distances to some objects can produce the same viewing angle to a human eye (in this example the left one). Only the other eye provides the missing information that is necessary to estimate the distance (red line).


When viewing a single picture the parallaxis of movement cannot be used to extract information of depth because the two dimensional image does not change when we move our head (holograms are special in this regard). The texture gradient provides us even in this case with some information about the depth of the scenery. But obviously this is not enough to create the impression of a threedimensional scene. Only the additional information of a second image provides us with a realistic impression of depth.


But how can you get two slightly shifted images of the same scene? There are many practical methods to achieve this. These have very different levels of limitations and costs so I will discuss some of them here.

Using one camera, shooting two images in a sequence:

The most simple way is of course to use an existing camera, shoot the first picture, shift the camera by a few centimeters and shoot the second one. The limitations of this method are easy to see:

  • When shifting the camera you have to be carefull to do preserve the viewing direction of the lenses. If there is a slight turn in the direction you get what we are used to if we squint: If we turn each of our eyes more to the outer rim then both of the images do not overlap. We see two different images that are not being correlated by our brain. If we look cross-eyed our brain refuses to correlate the different images too. In both cases we tend to get a bad headache after a few moments. The same will happen if you watch stereo images that were taken with a non-parallel viewing direction.

  • Obviously there should be no movement of objects between both of the shots. Such differences are very disturbing for the three dimensional image (because our brain uses exactly this kind of differences between the images to derive the information of depth). This kind of differences between two images destroy our impression of depth.

So this is only the easiest and cheapest solution to create stereo images. You can use this for a short test, but it's not suitable for taking serious photos.

Using one camera and a beam splitter:

For normal camera lenses there exist some prism devices called beam splitter wich fit into the filter thread. The splitter has two lenses from wich the entering light is splitted so that the light from the right lens falls onto the right half of the film, the light from the left is lead to the left half of the image. That way the camera takes a stereo image that has only half of the width of a normal image. To view this kind of image correctly a viewer is necessary that ensures that each eye only gets to see its own half. These viewers exist for slides as well as for paper prints.

With a beam splitter it is easy to take stereo pictures even from moving objects. The devices have a strictly parallel viewing direction for each half so there is no problem with this either. A disadvantage is the format of the images: It's only half as wide as a normal photo, so it's a quite narrow upright format. There exist no equipment for the projection of slides that were taken using this technology.

Using two cameras:

If you own two cameras you can attach them next to each other on a bar. If the cameras are small it is even possible to mount them with the distance of the human eyes (6.5cm) from lens to lens. If the cameras have zoom lenses it can be difficult to make sure they both show the same detail of the scene. Even fix focus lenses of the same type do not always show exactly the same clipping. Often even lenses of the same type differ in the brightness of the produced images (this can be adjusted using the setting of the film sensitivity or using some correction setting for the aperture).

The remaining main problem is the synchronous release of the cameras. There exist cable releases for two cameras that can do the job. Much more precise is the use of electronic synchronisation mechanisms if the release of the cameras is electric. I use a pair of Olympus XA2 cameras that I enhanced by leading the electric contacts of the release to the outside of the camera and connecting them to a small box with a common release for both cameras. The necessary circuit can easily be made by yourself. Here is a  description for building a synchronous release.

You can even mount both cameras on a bar so that both of the bottom sides point to each other. That way you create a stereo camera for upright format photos. In this case you have of course to keep to the distance of the human eyes too. At the moment my stereo camera uses three small cameras from wich I can always choose two for the actual picture. So I can choose between upright and landscape format. The following figures show the construction.
Upright format Landscape format

Using two (or more) synchronised cameras is the most flexible solution. If you choose taking slides you can watch them using just two normal viewers (one for each eye). For projection of the stereo slides you need two projectors, two polarization filters and a pair of glasses with polarization filters. The quite small costs of small rangefinder cameras with fix focus lenses (especially if you buy them second hand) enable the construction of a cheap and universal stereo camera.


Here I will proceed...

 Encouraging, praising & criticizing eMails are highly appreciated.