Mirror or Prism assisted viewing
of stereoscopic pairs.

By John Wattie

Large stereoscopic image pairs on a computer screen can be seen by cross-eyed viewing. (X stereo). Beginners find cross eyes produces a headache, so an easier method has been devised.

Mirror assisted viewing (m stereo) requires one of the pictures to be flipped over. It returns to normal orientation when seen in a mirror. The normal image is looked at directly with one eye. The mirror is adjusted over the other eye until the two pictures superimpose.

This is far simpler than methods often described on the Internet such as liquid crystal shutters and flickering images. It will even work for moving pictures.

How to perform mirror assisted viewing.	Hyper-stereoscopy.
Image for mirror viewing	Image for cross eye viewing

Details of Mirror or Prism Assisted Viewing:

Mirror

One of the images should be flipped back to front. Here the left image will be the mirror image.
The same method works for viewing transparencies or radiographs on a light box. Just flip the left transparency back to front.
Mirror viewing also works with computed tomography [3D reconstructions] viewed, in motion, on a workstation. The reconstruction is run in two windows, 7 degrees out of phase, with the left image flipped.
The mirror should be made of thin glass. A ladies' handbag mirror is good, but her dressing table mirror is poor. A thick glass mirror gives a double reflection and distortion, which spoils the result.
Acrylic mirrors are very thin and work quite well.
A bigger mirror is better than a hand-bag mirror. The mirror should almost be in contact with the screen - which is impossible with a hand-bag mirror.
An excellent mirror is the front surfaced mirror provided with various 3D books from Dorling Kindersley (e.g. Microlife (1998) by Christopher Parks and Tim Hellier et al). It works well on the picture provided here. Front surfaced mirrors or prisms are the only way to avoid double reflections.

Hart Mirror Method

An excellent mirror rig for viewing two full size pictures in stereo has been designed by John Hart. It would work for two computer screens too.

When I explained the theory of mirror viewing at work, Richard Gee unscrewed the mirror off the typist's wall. It was 30cm square and worked well on a large workstation screen because the images were well apart, but Richard got into trouble when the girls found out next day!

Russian method for books

link

Prism

A 45 degree glass prism is better than a mirror. The prism from a broken pair of binoculars is excellent. Suppliers of scientific instruments have 45 degree prisms for sale ( Edmunds prisms are excellent optically but not cheap).
A prism reflection is bright and not distorted or double.
There is no need to have the prism in contact with the screen - it works properly at normal viewing distances.
Do not tilt your head to one side. The line between the eyes should be horizontal.
Look at the right hand image on the computer screen with both eyes.
Place the prism against the left side of your nose so the left image is reflected into your left eye.
Still looking at the right hand image, angle the prism, without moving your head, until the left image is superimposed perfectly over the right image. If your hand is unsteady, support the prism on something solid.
As the pictures fuse, you will see in 3D, without eye strain.
It is easy to tell when a person first sees these images in 3D. If they say "Oh yes - interesting," they have not got it. If they say. "Oh wow ...that's amazing...," they are seeing three dimensions.
Prism viewing works perfectly with the Dorling Kindersley 3D books, so do not despair when the front surfaced mirror they provide becomes too scratched.

Hyperstereoscopy, Mirror viewing, Central Otago, New Zealand:	View with screen minimum 800 pixels wide.

Left Eye, Mirror Image

Right Eye, Correct way around.

Hyperstereoscopy,
X stereo version (cross-eyes):

otagox.jpg (48857 bytes)

Right eye Image

Left Eye Image

U stereo version (parallel eyes) Pan to the right and use the middle two pictures.

RIGHT

LEFT

RIGHT

LEFT

The middle two pictures from this set of four can be used for U stereo: parallel eyes. Move away from the screen if the separation is too great or use a 4 mirror viewing system (Cazes viewer).

Go to Start

Hyperstereoscopy

Hyperstereoscopy means the two photographs have been taken with a separation greater than the distance between human eyes. Normally we cannot see in 3D beyond about 200 meters. By making our heads as big as a WWII battle-ship range-finder, we could see distant objects in 3D. As that is impossible (without genetic engineering), it is easier to use hyper-stereoscopy for this delightful effect.

This hyper stereo pair, showing Lake Hayes, was photographed by John Wattie. It is a "running stereo". A suitable site on the river terrace at the foot of the Crown Range was chosen. The left picture was taken first, then the photographer ran very fast and took another. The result is: the viewer seems to have a head as big as a small hill. The model effect then takes over, since we do not think of ourselves as giants. The brain assumes this is a small model of reality.
(Hyperstereoscopy without model effect is discussed here: PePax)

It was a fine Central Otago day - no clouds to move in the sky and spoil the stereo. The original transparencies show some sheep moved between the two exposures, but the version here has such low resolution you cannot tell the little white dots in the nearby fields are not quite right.

This valley is famous for its geomorphology, showing the effects of ice-age glacier carving. For a discussion of glaciers, press [here].

Shallow angle prisms:

The more common version of prism viewing is using shallow angle prisms over each eye, thin edge in. These bend the light sideways. U format stereoscopic pairs too big for parallel stereo can be fused.

The [Pokescope viewer] uses shallow prisms deviating light about 7� and is good for parallel computer stereo, as on this page (U stereo).

Close viewing of the computer monitor with a Pokescope causes keystone distortion, where the center part of the pair are magnified more than the lateral parts. This differential, vertical magnification spoils the stereo effect in the same way camera convergence during stereo photography causes keystone distortion. The pictures become hard to fuse. So you need to sit back in the normal place and not lean towards the screen while using diverging prism viewers.

The up side is: pictures taken with camera convergence have distortion in the opposite way, bigger laterally. So excessive divergence with a Pokescope potentially corrects keystone distortion caused by macro photography with convergence. (See [how to take stereo pictures] for more on keystone distortion).

Mirror stereo on a single flat screen also suffers from distortion, which is why John Hart's rig - mentioned above - uses pictures horizontally angled relative to each other. Mirror stereo on two computer monitors can also be corrected for distortion by angling the screens.

3D contents page