International Stereoscopic Union















 

Making a Start in Stereo

Derived from a series of articles published in the Journal of The Stereoscopic Society

Many people, having had their appetites whetted by viewing somebody else's slides, or by reading Pat Whitehouse and David Burder's excellent introduction to the art of stereography - especially those who are remote from clubs and societies which hold meetings - are left floundering without advice on how actually to get started. Or, if they do, maybe the results produced cause so much eyestrain and headache that they are tempted to give it all up as a bad job.

Clearly there is a need for a guide to the basics of stereography which is not catered for in one book.

Two of the main comments made by beginners are "Where do I start with stereo equipment?" and "How much will it cost?".

It is also not immediately obvious that a silver screen is necessary to view projected stereo slides or why.

Another problem is the terminology bandied around by many of the experienced workers, which is enough to scare the daylights out of the average beginner. Terms like "hypo-stereo" and "hyper-stereo", "giantism" and "lilliputianism"  - to say nothing of "pseudoscopic", "orthoscopic", "planoscopic", "homologue separation" and the like - are calculated to send all but the boldest spirits back to their corners with their tails between their legs. What they ought to do, of course, is to ask the people who use such terms to properly define them; the results might be interesting! It is a fact that after a certain amount of experience many people just drop into the bad (?) habit of using these terms without thought. This is fine for communication between experts but, like any specialist jargon, it ought to be banned in mixed company!

This guide to the basics therefore sets out to "diminish the mystery" surrounding stereo photography and to provide a plain man's guide to current techniques. It is definitely not intended for experienced workers, and for those that do read it despite this warning and feel that we are over simplifying matters, please think twice about commenting on this score! Of course, we would be delighted to receive constructive comments so that we can improve this guide.

 

The basic concept of stereoscopic imaging

AN INTRODUCTION TO STEREO GEOMETRY

The Information In this section can be applied to any format, except where specifically stated otherwise.

The essential feature of stereoscopic vision is that each eye sees a slightly different image of a scene, which is focussed by the lens on to the retina. The images are not in themselves three-dimensional, and it is the brain which 'reads, the information from the two images and processes it into 'stereoscoplc' form. This does not by itself provide a full impression of stereo depth: to achieve this the brain relies on stored information based on recollections of past experience of distances between objects (spatial separation) and other inputs such as perspective, which are combined to create normal stereoscopic vision. This complex process of converting the two different retinal images into a single three-dimensional picture is known as stereopsis.

Stereo photography, or stereography, sets out to 'freeze' the two images as seen by the eyes, by using camera lenses in place of the eyes' lenses and film in place of the retinas. The two resulting film images, called 'chips' (in customary transparency form) are mounted so that when presented to the eyes either directly - in a viewer which focusses the images and ensures that each eye sees only its appropriate image - or indirectly by projection on a screen using polarised light and polarising spectacles to ensure the two images on the screen retain their separate identities, which the brain then integrates into a single stereo image in the normal way. The corresponding points in the left and right views which fuse into a single image are often referred to as homologues.

Whether looking at a scene by normal vision or through the stored images of a stereo pair, the best clue given to the brain as to distances in depth in the scene is the horizontal differences in the lines of sight recorded in each image, known as parallax.

It is these small differences in direction and angles upon which stereography depends, and which explains why a rudimentary knowledge of stereo geometry is desirable for successful picture taking, mounting and viewing.So, that is the basic idea behind stereoscopic image making. Read on for more detail, or move on to the section on actually making some images!The distance between adult human eyes (interpupillary separation) varies between possible extremes of 55 mm and 75 mm, with an average of around 63.5 mm, and it is this latter measurement on which most stereo taking and viewing is based. Since the viewing of stereographs requires the brain to carry out the same processes as in normal vision, it is essential that the information supplied to it matches so far as is possible normal binocular vision. Any excessive separation or misalignment of the two images during viewing causes the brain great problems, which manifest themselves as double vision (diplopia) eyestrain and headaches.

Conversely, all viewing of stereographs in which the rules of stereo geometry have been observed should produce a result indistinguishable from normal vision. It is for this reason that the maximum separation of homologues on the two film chips when mounted for direct viewing should not exceed 63.5 mm and the spacing of the viewing lenses in a hand viewer should also be 63.5 mm.

For the same reason, the maximum separation of homologous points when projected on to a screen should not greatly exceed 63.5 mm in normal viewing situations - a point which is elaborated in the section dealing with stereo projection - due to be published at a later time.

When taking and mounting a stereo pair for normal viewing, it should be kept in mind that the vertical edges of each of the viewing apertures in the mount or mask will coincide in space when seen in a viewer and similarly will coincide on the surface of the screen when projected. See Figure 1 below, which shows the effect as seen through a viewer.


Effectively, the edges of the mask or mount form a 'frame' in space (E - F). This is usually called the stereo window, and a correctly mounted stereo pair will be seen entirely beyond the window except when a special technique is used to bring part of the image forward - i.e. through the window. Even then, only those parts which do not touch the mask edges should be allowed to project.From Figure 1 it can be seen that the size of both mask apertures must be identical, otherwise the edges of the stereo window will each appear at a different distance from the viewer (E and F'). Thus the distance A B must equal C D, although the actual size of the apertures can be altered according to the dimensions of the film chips or for the purpose of cropping in order to improve composition.

From Figure 2 it will be evident that the separation of the two mount or mask apertures in relation to the mounted views will influence the distance at which the stereo window appears. Accepting the convention that the separation should not exceed 63.5 mrn, and that this would represent a window at infinity, the window can then be brought closer by progressively reducing the spacing, with a spacing of 61 mm giving a window set at 1 metre. In practice, to ensure the stereo window always remains on the plane of the screen when projecting, the standard window distance 1.5 set at 2 metres (the 'seven foot window', which means that the centres of the mask apertures are set at 62.2 mm.

Note that this is independent of the actual dimensions of the apertures (provided they are equal): increasing or decreasing the size of both apertures will merely increase or decrease the size of the stereo window, but will not alter its apparent distance. In the so-called 'close-up' masks, the apertures remain at the standard 62.2 mm but are narrower so that the film chips can, if necessary, be positioned further apart to re-position near objects behind the window.

The width of the window will vary according to the distance at which it is set, the width of the mask apertures and, to a lesser extent the focal length of the viewing or projection lenses. At the conventional 'seven foot' window distance, the width will vary from about 1 metre using the 'Realist' format mask of 21 mm aperture width to about 1.8 metres using the full frame 36 mm aperture width.

The eye and brain can comfortably accommodate only a limited range of subject distances; this distance varying with individuals and with practice. In normal vision, the eyes are constantly changing focus according to the object distance in order to stay within this range. In a stereograph viewed on a single plane, there is an enforced limit to the depth range which can be viewed comfortably, corresponding to a parallax displacement of 1.2 mm between the nearest and most distant object in the scene. If the depth range is too great, there is a natural tendency for the eyes to change focus whilst converging on excessively near points in the scene.

The function of the lenses in a viewer is to collimate the images to a comfortable distance near infinity for the normal relaxed eye, and the conflicting demands of converging the eyes to fuse excessively close near points within a scene while simultaneously focussing on infinity leads to excessive strain on the brain's focussing and convergence mechanisms and once again quite literally results in headaches.

There is a further requirement for successfully viewing a stereograph: namely that everything from the foreground to the most distant part of the picture should be pin-sharp. To ensure this at the same time as avoiding problems with accomodation it is important that the depth of a scene in a stereograph should not exceed the depth of field available

This table gives some guidelines for near and far distances in any given scene:

Near point Far Point
0.75 m. (2.51) 1.2 (4.01)
1.0 (3.51) 2.0 (7.01)
1.5 (5.01) 5.0 (161)
2.0 (7.01) Infinity

When using a normal stereo camera with lens separations of between 65-70 mm, it is not advisable to attempt scenes with a prominent near point closer than 2 metres; otherwise the image will appear distorted, together with a tendency towards dwarfism,. With two-camera or 'Siamesed' systems having lens separations greater than this the nearest point should be correspondingly further away. This can be calculated using the 'One in Thirty' guidelines i.e. the minimum subject distance should not be less than 30 times the lens separation distance. This will be explained in greater detail in a future section of this series dealing with the taking of stereographs.

In order to create a correct stereo window, purpose-built stereo cameras have the axes of the lenses positioned slightly inwards from the aperture centres, imparting an effect similar to the use of a shift lens in mono photography.

Users of twin-camera systems can only achieve a satisfactory stereo window by contrived means:

1. By moving the two chips apart relative to the mount apertures, thus duplicating the effective masking of the purpose-built stereo camera (remember, the images are inverted when they are formed in the camera, so having the camera apertures closer than the lenses (and thus masking the inner edges of the chips in the camera) is equivalent to having the standard camera chips masked on the outside edges when they are mounted for viewing), in which case it is necessary to use mounts with an aperture narrower than normal (the GePe 32 mm mount is recommended for this purpose when using full-frame 35 mm chips),

2. By 'toeing-in' the two cameras so the lens axes converge at 2 metres. This gives a slight trapezoid distortion to the top and bottom edges of the picture which, however, is often acceptable in normal viewing and in any case tends to be swamped by the opposite trapezoid distortion during projection.

3. An alternative method of achieving the shift effect with two cameras is to slightly mask off the inner edges of the film aperture inside the camera body; ensuring of course that the masking width is identical each side.

MAKING THE EXPOSURES

TECHNIQUES AND EQUIPMENT

For the purposes of this article it is assumed that the end product is a pair of transparencies for viewing in a hand viewer or by projection, although most of the techniques apply equally to the making of stereo prints. The text contains some generalisations and a lack of precision which may offend purists, but the guidance is of course intended for beginners.

The techniques described are in increasing order of equipment sophistication (but not necessarily of improved end result).

 

1. Sequential exposures with a single camera

This is the simplest method of all, since no special equipment is required and almost any type of camera can be used. The disadvantage is that it is limited to static subjects, since any movement in the scene between exposures sets up an interference pattern between the two images which the viewer's brain cannot cope with. Also, it is vital that the two exposures are kept in absolute register, horizontally and vertically, or else the brain will again find it difficult or impossible to fuse the two images into one, and headaches will result.

For normal topographical scenes, a sideways movement of 5 cm to 10 cm, between exposures with wide-angle (28 mm) to normal (50 mm) focal length lenses produces the best results. Subject matter closer than about 2 metres should be avoided. When using a hand-held camera, a useful tip is to stand with the feet about half a metre apart and put the weight of the body on the left leg when taking the left-side exposure, and on the right leg when taking the right-side exposure. If possible, the centre of the viewfinder should be aligned to a particular distant point m the scene and kept fixed on it for both exposures. A camera with motorised wind-on is a distinct advantage, since it not only speeds up the time between exposures but also avoids the need to take the camera away from eye-level, thereby helping to keep the two exposures in register. Shutter speed should always be fast enough to avoid camera shake during exposure, and care should, he taken to press the release button with a smooth action, again in the interests of keeping the two images in register, as well as sharp.

More reliable results can he achieved if the camera is mounted on a tripod topped with a slide-bar which facilitates accurate movement between exposures. Suitable devices to serve as a slide-bar can he made up in wood or metal, but a very convenient ready-made piece of equipment is the type of slide-bar designed for accurate adjustment when taking close-up mono photographs. A simple scale in millimetres glued to the base is an aid to more consistent and accurate results. Typically, such a slide-bar provides for movement of up to 15 cms, which is ample for most purposes.

It is particularly useful for close-up work (macro-stereo) when a reduced and fairly carefully calculated separation is necessary. A variant of the slide-bar particularly suitable for normal stereo work, is a 'swing-over' platform which shifts from left to right by mains of parallelogram, bars: the advantage of such a device is that the separation can be precisely determined and remains constant for each exposure.

Note: When hand-held, and the viewfinder fixed on a particular point, the camera tends to he automatically 'toed-in' between the two exposures, so that the optical axes at right-angles to the lens converge at infinity, or nearer. When the camera is on a slide-bar, the axes remain parallel. In the latter case, there is less likelihood that the full width of the pair of transparencies will provide proper fusible stereo imagery; hence it probably becomes necessary to reduce the frame width when mounting them, in order to provide a suitable stereo window: this will be more fully explained in a later section on mounting techniques.



2. Simultaneous exposure with two Cameras

With this method, the range of subject matter is increased enormously, although problems can still arise if the two cameras are not lined up accurately or if their shutters are not fired at exactly the same time. At its simplest, the method involves mounting two cameras side-by-side or base-to-base on a support bar, and making a simultaneous exposure with the aid of a twin cable release or some other means of triggering the shutters simultaneously.

There are some shortcomings to this technique;

(a) The lens separations with the cameras placed side-by-side range from 10 cms with the smallest compact 35min cameras to about 17 cms with a large SLR: in all cases greater than the ideal separation of 6.5 cms (equivalent to the separation between the human eyes, which produces the most natural stereo effect with standard lenses for most subject matters) and resulting in a degree of hyper-stereo in which the images appear somewhat miniaturised compared with real life. In many cases, however, the exaggerated stereo effect has greater appeal to a non-critical viewer and can usually be accepted as long as there are no obvious distortions to, for example, human features.

To reduce the lens separation, the cameras may either be overlapped one behind the other (the resulting variation in image size being insignificant except for very near objects) or mounted base-to-base, but this latter course has a number of disadvantages cheif among which are the vertical format, which is not well-suited to many stereo subjects, and the limited scope for reducing the width of the images to form a suitable stereo window.

(b) Even with two cameras of the same type, differing focal lengths and other variations may occur unless the lenses are specially selected as matching pairs, resulting in discrepancies between the two images which, although not too troublesome in a hand-viewer, can make for difficult viewing by projection. Slight differences in the mounting of the cameras on the bar can likewise cause problems, especially if they are not in exactly the same horizontal plane or if the optical axes diverge ("toe-out"), As with the use of a single camera, it is usually desirable to reduce the width of the transparencies during mounting in order to fuse the whole of the images behind a stereo window; but if there is a wish to avoid this (eg, when only hand-viewing is contemplated) the cameras should be slightly toed-in on the mounting bar so that the optical axes converge at around 2 metres, thereby forming a slightly deformed stereo window at that distance.

(c) When cameras with mechanical shutters are used, or if the shutters are triggered by direct pressure on the release buttons, it is extremely unlikely that they will open in sufficient synchronisation to permit the use of flash or to stop any moving subject at exactly the same time. Even cameras with electronic shutters fired in unison by means of a coupled release mechanism may not respond identically, especially at high speeds. Under no circumstances should separate flash units be used on each camera in order to overcome lack of synchronisation, as the resulting variations in shadow positions render it almost impossible to view the stereo pair satisfactorily.

Many advanced stereo workers permanently couple a pair of cameras, often with wired-in circuitry to facilitate simultaneous shutter release. A popular 35 mm compact camera for this purpose has been the Olympus XA series. At the simplest level, all this type of camera needs to achieve a reasonable level of synchronisation is for the two shutter releases to be connected in parallel. Once this is done, pressing either shutter release will trip the two cameras together. The drawback to this approach is that the two cameras are still independantly setting their exposures, and can show differences in the chip densities. There are ways of making one camera work as a "slave" of the other, thus using just one set of exposure electronics. These methods are, however, outside the scope of this document.

 

3. Using a beam-splitter

The beam-splitter (more correctly called an image-splitter) is a device which attaches to the front of the lens of an ordinary mono camera and by means of mirrors or prisms produces two images of the same scene side-by-side on a single frame of film. The advantages of this technique are: (a) simplicity - the bearn-splitter simply screws on to the threaded front of a standard camera lens and can he put on or taken off at will, enabling stereo pairs and mono exposures to be mixed on a single roll of film; (b) economy - no special processing or mounting is required as the stereo pairs are contained within the ftame of a normal mono exposure; (c) reliability - provided the attachment is parallel to the run of the film in the camera the image pairs will always be in register with each other, and of course are completely synchronised.

There are, however, substantial disadvantages as well. As a result of one eye-view effectively passing through one side of the lens, and the other eye-view passing through the other side of the lens, the device introduces distortions in opposite directions in the two images.Also, as transparencies, the stereo pairs can only conveniently be viewed in a complementary hand viewer. The picture size, on a normal 35 mm transparency, is small - no more than about 16 min in width at most, after allowing for the overlap of images in the middle - and although it is feasible to remount the half-frames in a small (eg, Nimslo-size) stereo mount the fusion is often rather unsatisfactory. The vertical format and the relatively long focal length of the small images do not lend themselves to satisfactory composition of many pictures: the best applications being for portraits and other near-to subjects. There is a risk of ghosting or flare due to the scattering of fight in the mirror system.

Nevertheless, the beam-splitter is a useful tool for experimentation with 3D photography prior to commitment to purpose-made equipment. It is particularly suitable for the production of stereo prints in the form of normal en-prints from colour print film; such prints can either be viewed without modification in the Loreo (or a home-made) viewer, or cut up and mounted on board for use with a Holmes-type viewer.

Availability of equipment: In the past, beam-splitter attachments and viewers have been made for specific high quality cameras (eg, Leica and Contax) and for general use (Stereax): these are occasionally available on the second-hand market.

A Japanese-made system with the trade-name of Stitz, produced in the 1960s-70s, comprised a beam-splitter with adjustable mirrors to suit a range of focal lengths, a projector attachment (not very effective) and a viewer for stereo en-prints. The relatively recently discontinued Pentax Stereo Adapter Set, comprised a beam-splitter attachment and viewer, selling for around £60 (either component was also available separately). It could be used on any 35min SLR camera with a standard 50 mm-55 mm focal-length lens: it came with a 49 mm filter thread but step-up rings were obtainable to adapt to other lens mount sizes. It could also be used, with limitations, on 120-size SLRs with standard 75 mm-80 mm lenses (where the wider film gave greater flexibility for remounting in stereo mounts).

The Franka Beam-splitter, a lower-cost alternative to the Pentax model, was available from Reel 3-D Enterprises in the USA. It did not include a viewer and was designed primarily for en-prints (for which a 3.5x5 inch print viewer was separately available).The Loreo Company have offered a camera which is also designed on beam-splitter principles, incorporating angled mirrors on its front panel and having many of the characteristics of these devices. Although it can be used experimentally for the production of stereo transparencies, its simple point-and-shoot design (with fixed aperture and shutter speed) mean that it is only really suitable for the production of stereo prints from colour negative film with intrinsically wide exposure latitude.

Loreo have also introduced a "3D Lens in a cap" which is basically two lenses with a mirror spreading/transposing box, which is produced with camera mounts for several popular SLR bodies. The two-lensed approach does not suffer from the distortions inherent in the beam splitter designed for single lensed cameras, but the Loreo version is designed to work with their viewer, and so the images are transposed on the film through the design of the mirror box, and unfortunately this does cause some keystone distortion.



4. Commercial stereo cameras

Stereographic cameras have been manufactured for almost as long as photography has existed as a technique. Those made before the First World War are now museum pieces and collectors' items; those made in the Inter-War period and which survive in working order (mainly for use with glass slides or roll film) are only recommended for experimental use.

During the last 3D photography boom period of the 1950s, hundreds of thousands of stereo cameras were manufactured, mainly in the USA and Germany, and many survive in good working order. They still form the backbone of the amateur stereo photography movement. Because of their solid and relatively simple construction, the better models from this period are still worth acquiring. A good camera repairer is able to service and patch-up a well-preserved camera of this type almost indefinitely.

Werner Weiser's book 'Stereo Cameras since 1930' lists about 70 different cameras produced in this period. Almost all utilise 35 min film and were designed primarily to produce pairs of colour slides. The vast majority adopted the image size of 24x23 mm pioneered by the American Stereo Realist camera launched commercially in 1947: known familiarly as the Realist or American format. A small but popular minority adopted the alternative image size of 24x30 mm utilised by the French Verascope F40 camera in 1946 (the last in a long line of prestigious stereo cameras produced by the firm of Jules Richard under the name of Venascope since the 1880s): this is often referred to as the Verascope or European format.

Among the more popular of the Realist-format cameras are the Stereo Realist itself (in two main versions: Model 1041, with f3.5 lenses and Model 1042 with f2.8 lenses) and the Kodak Stereo Camera from the USA, and the German-made Edixa Stereo and Iloca Stereo (made in several model types). Any one of these models in good working condition can be relied upon to give satisfactory results. A simpler camera in the Realist format is the Stereo Graphic (with a single fixed shutter speed of 1/50 sec), made in the USA but also sold in the United Kingdom with Wray lenses and known as the Wray Stereo Camera.

In the Verascope format, the Verascope F40 itself, early versions of the lloca Stereo, and the Belplasca (made in the former East Germany) are the only models likely to be found from the 1950s and 1960s, but the Ukrainian Fed company produced a camera for this format (based generally upon the Belplasca design) in the 1980s, with fully automatic exposure. Despite its relatively austere specification, the Belplasca is arguably the most sought-after of the 1950s cameras because of its high-quality Tessar lenses and general ruggedness in use.

Two cameras with reduced lens separation and thus mainly suitable for close-distance work have also been produced in some quantity: the Italian ho-Duplex 120 (using 120 roll film running from top to bottom of the camera to produce Realist format images) and the recent Japanese/British Nimslo (designed originally in order to produce lenticular stereo prints viewable without glasses; but capable of making acceptable transparencies in half-frame format). A clone of the Nimslo with some design modifications (for marketing rather than photographic purposes) was the Nishika, which was available only by direct sale from appointed agents.

Availability of equipment. Apart from the Loreo beam-splitter camera - now sold under the Vivitar brand) cameras of this type must necessarily be bought on the second-hand markets. Models can he found from time to time at general photographic dealers and camera marts such as the Portobello Road market in London, but the most dependable source of supply is now via the Ebay on-line auction, or through the Sell-3D on-line mailing list

There are also usually Some working cameras (condition not guaranteed) m the periodic photographic sales held by Christies, South Kensington and other auction houses; but prices are likely to be much higher than through a sale by private treaty. Guideline current prices by the latter method for a camera in good working order (and according to condition) are:

 

  Wray Stereo Graphic £80-£150
  Nimslo £60-£100
  Stereo Realist £80-£140
  Iso Duplex £150-£300
  Iloca II & Edixa II £120-£180
  lloca Stereo Rapid 2.8 & Edixa III £200-£250
  Verascope F40 £250-£350
  Belplasca £300 - £500

 


5. Hand-built stereo cameras
Over the years, a number of individuals and small companies have constructed all-in-one stereo cameras out of two ordinary production mono cameras. The precision engineering required for this is of a very high order, reflected in the high price of such cameras when offered for sale commercially. In all the models which have come onto the market, two camera bodies are spliced together so that a single roll of 35 min film runs across the two film apertures, which are positioned so that their centres are spaced as nearly as possible to the optimum of 65-75 innis, the film being advanced by the Collardeau progression (named after the pioneer of this method in early Verascope cameras) of alternately one and three frames, in order to produce an economical series of pairs.

Shutters are also linked, mechanically or electronically, so that a high order of synchronisation, suitable for flash, is achieved with a single release button.

In the most advanced cameras of this type, the two lenses are also coupled by means of rods or bands, so that focussing, aperture and (where appropriate) zoom focal length can be set in tandem with one movement. As camera design develops, latest features such as powered automatic wind-on and automatic exposure control are also being incorporated.

Availability of equipment. Two such camera designs achieving limited commercial production runs in the 1980s were the French-made Hectron and Twinolta cameras. At the end of the 1980s, production for commercial sale had concentrated m Germany, where three firms sold cameras by mail-order:

1. Peter Kato, Düsseldorf

2. Stereoskopie-Werkstatt Allgäu (Franz & Hermann Miller), Leutkirch im Allgäu

3. RBT-Raumbildtechnik GmbH, Stuttgart

The last was the most successful, continuing to this day to offer a wide range of equipment.


   

International Stereoscopic Union
8008 Zurich
Switzerland

Phone: +41 (44) 508 1024
e-mail: ISU@stereoscopy.com

 

© COPYRIGHT 1997 - ALL RIGHTS RESERVED INTERNATIONAL STEREOSCOPIC UNION