Fig 1: Rays from eyes converge onto object.
Man has used other forms of representation for the purpose of modelling his situation (imaginary or otherwise), such as sculptures and carvings. But drawing on two-dimensional surfaces is his main form of expression whether it be sketching, painting or even writing.
Most of the information here has been culled from the following references, [sgram,tiw,ns,faq].
Sir Charles Wheatstone invented the stereoscope [1838] and thus began a wave of interest in stereography. Though the ancient Greeks had some knowledge of stereography, the Victorian renewal of interest was further sparked by the invention of photography [1839]. [Euclid did examine the nature of seeing with two eyes and perceiving three dimensions in his treatise "Optics".] This enabled Wheatstone's invention to be combined with photographs (instead of hand-drawn images) resulting in a stunning capture of the real world in fine detail and great depth. Such was the rage that stereo-viewers were produced commercially in the 1850s and the London Stereoscopic Society was founded in 1893. Even today there is remnant interest in 3D photography. (Viewmaster stuff.)
More recently, anaglyphs were introduced where the viewer is required to wear a pair of cardboard glasses which had a red filter over one eye and a blue or green filter over the other eye. Films were made so that audiences with these glasses could experience a 3D effect.
Holography was the next development though its progress was hampered by the technical difficulties of the process. Polarization has been used to deliver different images to each eye, either using polarized light and filters orientated in perpendicular directions or using liquid crystal filters which alternately block the eyes of the viewer in time with the showing of left and right images.
Last year, the Pulfrich phenomenon was used in some TV programmes for 3D effects. This effect requires one eye to be covered with a fairly dark filter and relies on the psychological fact that the brain appears to process dim images at a slower pace than brighter ones. Thus with a moving series of pictures (such as a video sequence on TV) earlier frames from the filtered eye are fused with the present frames from the naked eye.
The importance of his result was that without any apparent visual clues, the human brain could reconstruct the 3D information. In someone with normal binocular vision, each eye has a slightly different view of the world around because they are separated by about 6--7cm. When both eyes look at a distant object, two processes cooperate to bring two images into the brain.
Fig 1: Rays from eyes converge onto object.
The first process directs the eyes to the object (so that the images are on the centres of the retinae) and the second focuses the individual images. For an object directly in front of the face, the first process turns the eyes inwards towards it (if it is assumed they start off pointing in a perpendicular direction from the face) and the process is known as convergence. See figure 1.
The brain must then fuse the different images together to produce the perception of depth; this fusion is known as stereopsis. Julesz with Miller showed that the sense of depth could arise solely from stereopsis, using patterns of random dots.
The next step came with the discovery that a separate pair of images is not needed for stereopsis. Tyler and Clarke [1979] made the first single-image random dot stereograms (SIRDS). They relied on the "wallpaper effect," discovered by Sir David Brewster [1844] who noticed a 3D effect from staring cross-eyed at wallpaper, see figure 2. This effect is due the brain matching up different units of the repeating block pattern in the wallpaper. The "mismatch" translated into perception of the plane of the wallpaper at a closer distance than what it really was.
Fig 2: The wallpaper effect: the horizontal lines represent the wallpaper
(solid line with ticks to show the edges of each strip) and apparent
wallpaper (dashed line); the thin lines emerging from each eye show the
direction in normal viewing mode (dotted) and cross-eyed mode (dashed).
The important detail was that the apparent depth was controlled and determined by the period of repetition of the wallpaper. Tyler and Clarke using an Apple II computer wrote a BASIC program to generate a repeating image in which the 3D information was embedded by changing the repetition interval in parts of the image. The result was a mess of random dots which only yielded its secret when viewed in the correct manner (cross-eyed).
The same technique was later applied by other people to coloured patterns and images thus creating single-image stereograms (SIS) or autostereograms.
The only disadvantage is that the eyes must lock on to different parts of the repeating pattern. There are two techniques available: wide-eyed (deconverged) and cross-eyed viewing.
Both required the decoupling of the two processes mentioned above that man utilizes to send useful images to the brain. The convergence mechanism must be separated from the focussing which remains fixed on the flat image.
Fig 3: Wide-eyed and cross-eyed viewing.
The two viewing methods differ only in the position of the point at which hypothetical lines drawn from each eye, in the direction of their targets, intersect (i.e. the point of convergence). For wide-eyed viewing, the convergence point is behind the autostereogram and for cross-eyed viewing, it is in front; see figure 3. The decoupling can take some practice to achieve because it is so against what is the unconscious norm.