HOLOGRAMS
FROM EADWEARD MUYBRIDGE'S STUDIES
OF THE HUMAN FIGURE IN MOTION


Eadweard Muybridge was born in Kingston-upon-Thames, near London, England in April 1830.  As a young man he sailed to America to seek his fortune and settled in San Francisco and by the late 1860's he had established a reputation as a first rate landscape photographer and a prodigious inventor. Today, however, Muybridge is chiefly remembered for his extraordinary work with the photography of movement.

In the spring of 1872 Muybridge received a telegraphed message from Leland Stanford, Governor of the State of California, a renowned millionaire who was President of the Central Pacific Railroad Company. Stanford wanted photographic proof that a horse, at full gallop, took all four of it's hooves off the ground at one point in it's stride.

Muybridge accepted the commission and made his first attempts at instantaneous photography in May 1872, in Sacramento, using Stanford's famous racehorse "Occident" as his subject. The success of this venture led to a second more serious investigation, again financed by Stamford, which took place at Palo Alto in 1878. This time Muybridge used white screens in the background, which brought his exposure time down to 1/2000 second establishing him as a pioneer of high-speed photography. In May 1881 Muybridge published "The Attitudes of Animals in Motion"... "A Series of Photographs Illustrating the Consecutive Positions Assumed by Animals in Performing Various Movements". As well as horses, his work contained photographs of a variety of different animals including dogs, pigs, pigeons, goats and human beings.

Muybridge developed the device he called a "Zopaxiscope", which projected sequences of photographs on glass disks to give the illusion of motion. This was the world's first ever cine projector and the most sophisticated device of its kind until Thomas Edison demonstrated his "Kinetoscope" 14 years later in 1893.

Using his strange projector, Muybridge demonstrated his moving pictures to distinguished audiences of artists and scientists across Europe, with a tour of lectures, first in Paris in 1881 and then London in 1882. On his return to America Muybridge was given sponsorship by the University of Pennsylvania, where he undertook his most famous project. In one year he took more than 20,000 photographs of men, woman, children, animals and birds and in 1887 the photographs were published as a monumental work of 781 folio-sized plates, entitled "Animal Locomotion". It remains to this day the most comprehensive analysis of movement ever undertaken, and is still used widely as a reference source for illustrators and artists.

My own interest in Muybridge's work started exactly a century later, in 1987, when, on examining the photographs from "Animal Locomotion" I was amazed to discover that I could see the images in 3D.

A few months earlier I had joined the Stereoscopic Society in London, and had learned the technique of "free viewing": the direct perception of a three-dimensional image from a pair of stereoscopic photographs without the aid of a stereoscope or viewer. Normally when we look at a photograph both our eyes converge to examine a detail of interest. When looking at a pair of stereoscopic photographs, each eye is directed at the appropriate image examining the same detail of the scene, but from two view points, which the brain combines to form a three-dimensional mental image.

Free-viewing on to the Muybridge Photographs revealed that consecutive prints in the sequences were stereoscopic, producing clear three dimensional images. None of the many books written about the Muybridge photographs make any mention of the fact that they are stereoscopic sequences, nor did Muybridge himself consider that this was of any importance, although he was a skilled stereo-photographer.

The fact that the sequences have parallax information in them is a direct result of the way in which the photographs were taken: Muybridge was working several years before the invention of the cine-camera, so in order to expose his photographic plates in quick succession he designed a method by which a number of individual cameras were triggered by the subject's passing in front of each camera in turn. Because the cameras were arranged in a line a few inches apart from each other, they recorded different perspective views of the subject. This provide a wonderful opportunity for me because all the individual images related collectively as a stereoscopic sequence, so they could be brought together in a hologram and synthesized into a single three-dimensional, animated scene, which would show all the spatial depth of the original event as it took place, a century before.

A detailed description of the set up and equipment used by Muybridge was published by the University in 1888. The camera house was a long low structure with a white front and a white interior, inside the camera house was a battery of twenty four, four by five inch cameras. The camera lenses were spaced fifteen centimetres apart and focused at a distance of about fifteen meters, perpendicular to the subjects line of progressive motion. From this layout we can calculate the optical convergence and the parallax angle from one viewpoint to the next.

In texts dealing with stereoscopic photography, for example in Lenny Lipton "Foundations of the Stereoscopic Cinema" (1982, page 97) the angle formed between consecutive cameras and the subject is called the "parallax angle", which is twice the angle of convergence.

The angle of convergence may be calculated by trigonometry :

 

Tan θ%  = tc                            Where "tc" is the distance between the lenses and

2D                            "D" is the distance from the cameras' base-line to the subject.

 

 

One may calculate the convergence angles and parallax angle for Muybridge's optical set‑up as follows:

 

Where tc = 15cm  and D = 15 meters = 1500 cm

 

Tan θ%15

2 x 1,500

 

Tan θ%  =  15

3,000

 

Tan θ%  =            5.‑ 03

 

θ%  =    (5. ‑ 03) tan-1

 

θ%  =    0.286 degrees

 

Thus the angle of convergence = 0.286 degrees

and the parallax angle = 0.57 degrees.

A total of 20 holograms of selected Muybridge sequences were made as part of a study of temporal parallax at the Royal Collage of Art. "Arabesque" (above) and other holograms were exhibited in a one-man show at the Museum For Holography & New Visual Medium in Germany. The Victoria & Albert Museum selected and exhibited work in their show "Towards a Bigger Picture" which was also shown at  the Tate Gallery Liverpool.

"Arabesque" was the first image I chose as part of the Royal College of Art study, a nude man standing on one leg. In the original Muybridge plates, this is a rather austere image, the sombre black and white tones making the sequence vaguely reminiscent of Gray's anatomical drawings. There is a universal quality gained by the nudity of the figure, he is an Everyman, less a person more a symbol: the pose is redolent of Leonardo Da Vinci's famous drawing the "Canon of Proportion", of a man framed in a circle and square, except that the Muybridge figure has his back towards us, his self and identity hidden.

Permission was first obtained from the Victoria and Albert Museum to make photographic copies of the original Muybridge Plates, which are housed in the print room in several boxes. The copies were made on black and white negative film, Kodak TX emulsion, which produced good definition, high-contrast results with standard processing.

To get the best definition the hologram was made from positive film transparencies which were the same size as the final hologram image and back illuminated rather than projecting 35 mm positive film or slides onto a screen. Several types of film and developers were tried but Agfa "Copyline High Definition" HDV 3P gave the best results for a high contrast, continuous tone image.

In aligning the films, control of the vertical registration was straightforward using the grid which Muybridge placed in the background of the image to ensure that the films were at the correct height and avoid any angular rotation of the image. Establishing the correct stereo-window was a little more complex, as there was a choice as to where to position the image.

The first option would have been to register all the frames to the background grid, in the same way that they had been aligned to this grid for the vertical registration. This would have located the stereo window at the plane of the grid, the furthest part of the image. The figure and the ground upon which he stands would have been pulled forward through the stereo window. This might have looked a little odd, and would have been a contravention of one of the basic rules of stereo-photography: to keep everything behind the stereo window, unless it is a protrusion such as the branch of a tree, where the trunk of the tree is set behind the stereo- window.

A second option would have been to register all the frames to an image point which was at the closest plane in the image. And in fact Muybridge provides one of these in the form of small number-plate at the very bottom of the image. This would have set the figure into a little box of space, psychologically distancing us from him.

The third option was to find a suitable point between the other two extremes. This is conveniently provided by the figure's left leg which, as he remains standing upon it throughout the sequence provides an "anchor-point" to the image. Locating the stereo window at the figure's left ankle means that only his right leg extends out beyond the stereo-window. Once the positioning was established, over a light-box, the location was maintained using a Kodak registration punch and pin-bars. Exposures for the laser transmission H1 hologram were made on an Agfa 8E56 (green sensitive) holographic emulsion on glass plate, cut to 40 x 15cm using an "Inova 90" 5 watt Argon laser at 514 nanometres.

The exposures on the H1 plate were all made sequentially by hand: the length of the H1 plate was divided into twelve sections, each of which was 25mm wide, rather than 65mm, as the inter ocular distance, in order to compensate for the fact that replay would be with a standard spotlight, rather than a conjugate of the reference wavefront, thus causing the sections of the H1 to dilate laterally. Limiting the slits to 25mm was calculated to permit each eye to see consecutive frames of the sequence at a viewing distance of 1 to 6 feet.

Because Muybridge shot each of his sequence against a grid of lines in the background, it is possible to see very accurately the degree of motion that the subject has undergone from one frame to the next. This proved very valuable in analyzing the relationship of the stereoscopic parallax to temporal parallax in the sequence.

Not a great deal has been written about "motion parallax" or "temporal parallax" in books about stereoscopic photography or holography, however, N.A. Valyus in his book "Stereoscopy" defines temporal parallax as:
"The displacement of any image point which occurs in a time corresponding to the visual inertia of the eye relative to the image of the moving fixation point".

A more straightforward description would be: "how a one-eyed man knows the world is in 3D" : that is, the ability to perceive the solidity of the objects and their spatial relationships by their relative displacement, rather than by stereoscopy.

Temporal parallax might be described as "memory parallax", because whilst stereoscopic parallax is derived from a comparison of two views both seen at the same moment with different eyes, temporal parallax is perceived by a comparison with the view seen by one eye with another view seen by the same eye a moment later.

Perhaps one of the reasons that there is not much written about temporal parallax is that since the early days of stereo-photography, one of the golden rules has been that there should be no subject motion between the left eye view and the right eye view. From the 1840's onwards special cameras have been made with two lenses, to ensure that both photographs would be taken at the same instant.

The early pioneers of multiplex holography were very careful to obey the rule that there should be no subject motion in the scene. Holographers such as R.V.Pole, McCrickerd & George, Kasahara, Kimura, De Bitetto, Groth & Kock, Redman, Haig and others, proposed different multiplex systems which would display footage taken either by a camera on a rail or by a static camera and a rotating subject which did not have any localised motion. In keeping to these constraints, the sequences of frames they shot were equivalent to a series of classical stereoscopic photographs.

Only when Lloyd Cross introduced subject movement in the early 1970's did the rules of classical stereo-photography get set aside, and there was disparity between consecutive images in the film footage. Whilst holographers were aware of problems of "time smear" occurring if the movement was too extreme, there was no analysis of subject motion in terms of temporal parallax. Apart from Mike Teitel's excellent paper on the "Time Depth Paradox" (Michael Teitel, "Animation in Holographic Stereograms: The Time Depth Paradox" SPIE Vol.1051 page 205. 1989) in computer-generated images, the influence of subject motion on the three-dimensional perception of the stereoscopic image appears to have been ignored by stereo-photographers and holographers alike. Temporal parallax is as real a phenomenon as stereoscopic parallax if there is local as well as global subject motion whilst the scene is being filmed.

In the "Arabesque" hologram, the fact that the figure is standing on one leg, anchored to the same spot throughout the sequence, means that this portion of the scene contains only stereoscopic parallax, caused by the lateral displacement of the cameras used to record the sequence.

In the sequence, the figure moves his right leg clockwise through 180o. This may be regarded, for the purpose of parallax analysis, as two progressions of 90o each. What is fascinating is the very different way these two motions are interpreted by the brain.

Initially, the figure rotates the limb so that motion gives rise to a disparity which cannot be resolved, as is commonly the case where movement has occurred between two sequentially exposed stereoscopic photographs. Although most of the other image points in the scene fuse together well, the brain is unable to resolve the contradiction of the opposing views of the limb which has moved laterally from left to right. This provides a good example image disparity, which all the books on stereophotography warn against, and why all subject motion is taboo in stereo-photography.

The figure continues to rotate his leg through a further 90o, in the same direction (clockwise), yet this time the obvious visual disparity of the previous sequence does not break up the image .

One of the most interesting visual effects to become apparent in the hologram is that of the merging of horizontal temporal parallax with stereoscopic parallax, which proves that they are two different attributes of the scene which are summed together in our perception of the image.

What is occurring with the lateral subject motion in frames #7 to #12 is that the temporal parallax is being constructively combined with the stereoscopic parallax, to result not in a visual conflict, but rather a false fusion, which gives rise to a three dimensional image which has hyper parallax distortions.
 


To view frames #8 and #9 in 3D try to fuse the two yellow dots into one.

If one examines frames #8 and #9 as a stereoscopic pair, one can see the hyper parallax distortion on the outstretched leg. Resting one's gaze on the extended foot, and comparing its position relative to the grid backdrop, one can discern that the limb is stretched out towards the observer at an impossible distance : it seems like a tiny foot at the end of a insect-like leg.

Although the motion of the leg is constant, in a clockwise direction, its path across the visual field changes from left-to-right to right-to-left. The brain misinterprets the horizontal temporal parallax as being stereoscopic parallax only when the local subject motion goes counter to the direction the of stereoscopic parallax shift.

In conventional stereo-photography one does not usually think of stereoscopic parallax as being "directional", because normally only two photographs are involved and they are either in the correct positions or they are transposed, however in a sequence of stereoscopic images this relationship is extended, and there is a correct order of frames, which if reversed yields a pseudoscopic image. Similarly, the temporal parallax in the scene has directionality, which can be "with" or "against" the direction of stereoscopic parallax.

The first part of the figure's rotation works against the stereoscopic parallax and is seen as disparity, whilst the second part of the rotation adds to the stereoscopic parallax, leading to a heightened sense of depth.

The fusion or disruption caused by the combining of stereoscopic with temporal parallax is determined by the extent of the movement as well as the direction. For example, if the subject motion is exactly equal to the progress of the camera, and going in the same direction, it will be as if both were standing still. The stereoscopic parallax will be cancelled out by the temporal parallax and the image will appear flat in that part of the scene. However, if the subject's motion from left-to-right exceeds the progress of the camera from left-to-right, it will be as though the subject were still and the camera were going in the wrong direction: the image will actually appear locally pseudoscopic.


The brain very rarely encounters pseudoscopic images in the natural world and thus tends to reject them or re-interpret them. Some subjects are so familiar that it is very difficult to see them as pseudoscopic. The clearest example is the human face: we are so accustomed to seeing faces the right way round that when we see a pseudoscopic face we are quite likely to mentally invert the image and believe it to be orthoscopic.

Throughout the latter part of the sequence there is "positive" temporal parallax in the right leg, and "negative" temporal parallax in the whole of the upper torso, so that it is actually inside-out: the grid which forms the backdrop actually dissects the upper half of the body, showing that it is pseudoscopic. Yet this is not how it appears: our perception is of a solid three-dimensional body. Only by reversing the anaglyph glasses can we see the arm is in front of the grid and therefore truly orthoscopic.

Each of the different Muybridge holograms was chosen because it illustrated a different aspect of stereoscopic and temporal parallax, however they were also chosen because of the images. The appropriated photographs are transformed in the holograms not simply because they are now three-dimensional animations, but because we view them with 21st Century eyes and see afresh just how extraordinary some of the images are, such as the nudes on horseback and naked children - images that would be difficult to make today, yet were permitted in the Victorian era, one we used to think of as prudish. When they were originally made the sobriety of the scientific undertaking exculpated the nudity, as works of art they are similarly absolved, but in this context we are enticed to pause and ponder what they signify. The holograms are just a channel for the force which is dormant in the photographs, just as the parallax was latent, waiting to be discovered.