Mounting slides or prints for a stereo viewer or projector
by John Wattie
Mounting stereoscopic slides for projection introduces a further window problem which does not worry computer viewing because of the strange ocular gymnastics used to see the images.
The virtual stereo window can be placed at any depth in the picture, as we have seen.
If slides are being mounted for viewing in a simple viewer which does not have prism lenses, it is vital that people do not have to diverge their eyes to see any part of the picture. In fact most people prefer to always slightly converge when looking at stereo pairs.
If the slides are projected onto a screen, everything is magnified and divergence can become an impossible problem.
This means the most distant planes must be set up, when mounting the slides, at exactly the same distance apart for the whole slide show. If they are not, the projectionist will be adjusting the projectors during the performance and this will cause bad eye-strain in the audience.
A simple solution is to have the most distant objects superimposed on the screen. This has the disadvantage that distant mountains seem only as far away as the screen.
A more advanced approach is to have the most distant objects diverged on the screen by 6.3cm.
In scenery photography, the distant mountains will be separated by 6.3cm horizontally (but of course they must be exactly the same height vertically)
In macro-photography the most distant object may only be a few cm away, but it should also be set at 6.3cm separation on the projection screen. The distant object then seems to lie behind the screen. This avoids excess convergence of the eyes on the nearest object as the stereo depth in macro-stereoscopy is often very great. This arrangement causes less optical strain when distant scenes and close-ups are mixed during a program, realising it is not orthoscopic.
If the slide masks of the two images are not superimposed, a side bar occurs where one slide illuminates the screen but not the other. Some workers prefer the masks to register perfectly, so there are no side bars. This means the stereo window must be set at a particular distance for all the images, usually 7 feet (2 meters).
2 meters comes from the 1/30 rule or the 2 degree
(Inv tan 65/2000 = 1.9 degrees).
Close up objects are less than 7 feet away and stereo masks may be adjusted to allow for greater convergence, yet still avoid projection side-bars.
Since stereo pairs will be seen in a non-prismatic viewer sometimes and projected at other times, a simple rule can be presented:
The most distant objects in the stereo pair must not be separated by more than the inter-ocular distance.
The average inter-ocular distance is 65mm, but to allow for small people and because many like to always converge, 63mm is sometimes suggested for the separation.
When mounting the stereo slides, a jig of some kind is very helpful.
Set up the most distant points at 63mm separation.
a) The slide mounts are placed the same distance apart they will be separated in the viewer.
c) Use a home-made jig over an illuminated view-box to keep the mounts precisely aligned.
d) Wear 2 diopter or stronger hobby glasses - they improve precision and also work as stereo viewers.
e) Homologous furthest points are located on each of the pairs - use stereo viewing to locate them.
f) The distance between these far points is measured precisely with dial callipers and set at 63mm.
Ensure the slides are not tilted - the pictures must be aligned exactly along the line they were in when taken, parallel to the bottom of the slide. Otherwise your eyes will have to move vertically or rotate in their sockets to fuse the images, which is possible, but many people in the audience will not be able to do it.
Choose a pair of windows separated by the distance between the nearest object on the two slides or very slightly less. This means you will need a choice of slide mount windows.
If the window edges are further apart than the nearest object, then that object will seem to project through the window. This "poke in the eye" technique is fine, but should not occur from the edges of the picture, as discussed above.
Special window shapes and arrangements where one edge is closer than the other can be set up, including the "falling frame" to reduce edge stereo failure. This means you will have to cut your own stereo windows, which is not easy. (Some purists object to non-standard window shapes. Usually they do not realise that edge stereo failure is more annoying than a slightly unusual window shape, which many viewers will not notice anyway).
To bring the window closer, as is often desirable for close up stereo, mask down each outer side of the stereo pair. This can be done with either separate 35mm slides or dedicated stereo camera pairs. It is possible to include infinity objects, but they must still be placed 63mm apart.
Commercial masks adjusted for window are available. These masks are no good for images bigger than the Stereo-Realist reduced size format. They spoil full size 35mm frames produced by shifting a single camera, as advocated here. These masks are not designed to include infinity with nearby objects.
- The three Stereo Realist masks are:
- 7 feet to infinity (Normal)
- 5 feet to 20 feet (Medium)
- 3 feet to 10 feet (Close-up)
35mm or 6x6cm transparencies mounted together in a single slide mount can only be projected in a dedicated stereo projector.
Only some stereo projectors will handle full frame 35mm, and the author has never seen a stereo projector for 6x6cm format.
It is easier to project these big, separately mounted slides using two standard projectors. Each projector will need a polarising filter properly aligned to match the polarising glasses worn by the audience. The screen must be metallic, to avoid scrambling the polarised light. A board painted with aluminium paint will do the trick in an emergency. A parabolic aluminium surface is ideal (e.g. the projection screen used in some front projection TV systems, or a Kodak Ektar screen). The parabolic screen gives a very bright image but with a narrow viewing angle. That is ideal since the acceptable viewing zone for stereo is in a 60 degree - just the same as the brightest image from the screen.
In a viewer, ortho-stereo occurs when the focal length of the taking lens is the same as the viewer. In stereoscopic projection, the "ortho-stereo seat" is directly in front of the screen at O mm from it where:
O = F W/ w
F = Focal length of the camera
W = Width of the screen
w = width of the slide (before it was masked down for stereo window)
If a telephoto lens was used for the stereo pair, the ortho stereo seat moves away from the screen.
In front of the ortho stereo seat the stereo effect is "squeezed" while further away it is "stretched".
Out to the side there is deviation shear or worse, loss of brightness.
Viewing stereo by projection is never as as satisfactory as a good viewer, especially for audience members nowhere near the ortho-stereo seat. (Ghost images, big people in front of you, problems if you tilt your head, insufficient blackout...)
This means each transparency is mounted as a separate slide. However, they should be mounted for viewing side by side in a simple hand viewer, with two non-prismatic lenses. When in this hand viewing mode, the most distant objects should once again be placed 63 to 65mm apart. Now when the stereo pair are projected, the two projectors are turned until the most distant objects are 63mm apart on the screen (and exactly the same height). Once this is done at the start of the show, correctly mounted slides will retain the correct separation.
It is more practical to have the two projectors one above the other to get the lenses as close together as possible to reduce keystone distortion. (However, if you are presenting macro stereo performed by camera convergence, it is better to have the projectors converging too so the camera and projector keystone distortions can cancel out.)
Standard mounting of each of the stereo pair means slides can be projected either stereo (two projectors) or mono (one projector), making a more versatile system than a dedicated stereoscopic mount allows.
There are quite a few stereoscopes available for seeing separately mounted stereo slides in 3d The simplest is two plastic 35mm viewers stuck together with tape. Many simple plastic 35mm viewers are fixed focus for closer than infinity, which is not good when the eyes are set parallel for seeing the stereo pair. Parallel eyes expect to focus at infinity.
Agfa used to make a simple 35mm viewer with focus. Two together make a stereo viewer.
35mm viewers taped together allow inter-ocular adjustment by tilting them at the hinge joint between the two.
The author has made a stereoscope with non-focussing viewers by arranging for the transparency to insert further away than normal, at the focal length of the lenses, so infinity focus is possible. A simple sliding inter-ocular adjustment also helps suit people with various inter-ocular separations.
Digital projection with liquid crystal glasses is not discussed here - this page is for amateurs with limited finances. When digital projectors fall in price, and work at high refresh rates, that is probably how we will all see stereo.
Setting up prints for stereo viewing in a viewer using magnification, non-prismatic lenses also demands the distant points be separated by 6.3cm. Background separation can be increased to the full inter-ocular distance (6.5cm) if there is a smooth transition from near to far. But if a near object is in front of a distant background, with nothing at intermediate distances, the jump in separation from near to far can be hard to achieve for some people.
Experts can diverge to at least 8cm for free viewing of Holmes card pairs, but only from 30cm. If they are looking from the close position allowed by magnifying viewers, precious few can fuse 8cm separations. The viewer is needed even by experts to see Holmes cards at full size.
Separation more than the inter-ocular distance is fine when prism lenses are used (as in a Holmes viewer), depending on the prism power available.
The classical Holmes card is 3.5 by 7 inches (89 x
Each picture is 3 inches square. (76.5mm).
Actual measurements by the author on antique Holmes cards shows not all followed this rule. "Rose Stereographs" of New Zealand were sometimes 87.5 by 73mm on a 4 by 7 inch card (101 x 178mm)
Holmes stereoscopes have different degrees of lens divergence. You can
check this by taking the stereoscope out in the sun and using it as a
"burning glass". When the sun image is smallest measure:
a) how far it is from the lens (viewer focal length)
b) measure the distance between the two images. This distance is the correct infinity separation for your particular viewer.
the near point deviation from the window is zero, giving a separation of 3 inches
Infinity deviation from the window is usually 5mm or less, giving a total separation of 76.5 + 5 = 81.5mm.
The total deviation can be increased above 5mm by having near objects poking forwards through the window, which needs special handling as described already.
Experiments by the author found 10mm total deviation could be viewed by most people, but only if infinity deviation was kept at or below 5mm relative to the frame. Splitting the deviation into 5mm positive and 5mm negative is the only sensible way to include objects both at infinity and closer than 2 meters, when using a 35mm focal length lens. This split technique can work very well on Holmes cards, especially if combined with tricks to reduce edge stereo failure.
Modern commercial postcards come in various sizes.
4 x 6 inch is common and can handle a Holmes pair if there is no side border, but it is not wide enough for a standard Holmes stereoscope card holder. It can handle 4x3 inch stereo pairs in portrait format or is cut down to 3x6" for standard pairs. For a Holmes viewer it needs pasting onto a 5x7" or 3.5x7" card, which was the antique method anyway. Mounting results in a bulky item, which curls when the card expands more than the picture. Curling is largely avoided by sticking a picture on both sides of the card, which also increases the cost.
Two 4 x 6 cards are standard for over and under viewers, but people are not used to purchasing two postcards for one view.
5 x 7 inch is a common larger postcard, which is ideal for a modern Holmes card.
5 x 7 inch is too big for 3 inch square pictures, although the extra space can be used for text. Text on a stereo card should also be in stereo, making it readable in the stereoscope. This sensible rule was not always followed on commercial cards. Printing under just one of the pairs was common, but not as good as stereo printing.
5 x 7 inch is ideal for portrait format stereoscopic pairs each measuring 4 x 3 inches, leaving a 0.5 inch border all around. 4 to 3 ratio is identical to 35mm film frames and fits an 800x600 pixel computer screen perfectly too.
5 by 7 inch cards can be seen in an antique Holmes-Bates stereoscope, but look better still if the lenses are set a little higher. 3.5 by 7 inch standard cards are still seen perfectly in such a modified viewer.
A modern standard for Holmes cards is described here, along with advice for cutting the pairs before pasting onto a card.
Lots more tips on how to cut and paste Holmes Card prints here
This reproduction is JPEG compressed with much reduced resolution compared with the original card.
Designed for viewing in a Holmes stereoscope, it uses parallel stereo (U).
This is a stereoscopic abstraction by the author, rather remote from reality, although based on a Southern Hemisphere Nebula just West of the Southern Cross. Astronomical stereoscopy outside the solar system is impossible: even the earth's orbit is not long enough as a stereo base. However stereoscopic machinations such as this are a load of fun (and much more spectacular at full resolution.)
adapted from a spectacular mosaic image taken by Nathan Smith, University of Minnesota/NOAO/AURA/NSF. He used narrow band filters to detect single emission lines from gases, then coloured the combined images arbitrarily. His colours have been changed here.
4x6 Holmes stereo cards are available commercially. along with a simple viewer.
A rather large 10.5mm total parallax, from the front desk to the mirror reflection, is spilt unequally into 2.2mm in card space and 8.3mm in air space. Air space parallax is easier to fuse for most people, who do not relax their eyes properly on first peering into a stereoscope.
The main window is asymmetrical, closer on the right, preventing edge stereo failure on the walls (the camera site not in the mid line of the room).
A modified "falling frame" avoids serious edge stereo failure on the desks, which is a problem with air space parallax.
Neophytes are further helped by a gradual change of parallax into the picture, rather than a sudden jump.
Resolution on the original photographic version is good enough to read the black-board without difficulty.
Sunshine through the window combined with the darker school interior provided a photographic exposure problem - but I am not going to give away every secret!
Because of all these precautions, beginners have found it easy to view, despite the considerable depth. In the stereoscope, people think they are seeing a rectangular picture and only experts realise the subtle tricks used to make viewing easy.