Title: Creating Special Effects for "2001: A Space Odyssey" by Douglas Trumbull
From: American Cinematographer
Do it right - then do it better - then do it all over again.
"2001: A Space Odyssey" was an extremely complex and difficult film to make, and naturally there are many interesting stories connected with the production. Probably the most important aspect of the film is its special effects, and in this article I shall try to relate some of the specific problems encountered in a production of this type, some of the techniques we used to create the effects, and a few other interesting points about the production as a whole.
One of the most serious problems that plagued us throughout the production was simply keeping track of all ideas, shots, and changes and constantly re-evaluating and updating designs, storyboards, and the script itself. To handle all of this information, a "control room," constantly manned by several people and with walls covered by pert charts, flow diagrams, progress reports, jog sheets, punch cards, and every conceivable kind of filing system, was used to keep track of all progress on the film.
With a half-dozen cameras shooting simultaneously, some on 24-hour shifts, and different aspects of many sequences being executed at once, the problem of keeping apprised of each shot's progress was difficult at best. For the purpose of being able to discuss a shot without referring to a storyboard picture, each scene had a name as well as a number. For example, all scenes in the Jupiter sequence were named after football plays - "deep pass,' "kickoff," "punt return," etc. Each of these terms called to mind a certain scene which related in some way to the name.
Early in production we began to realize that storyboards were useful only to suggest the basic scene idea, and as soon as a particular model or effect would come before the camera, something new would suggest itself and the scene would be changed. This change would often influence subsequent scenes. As each element of a shot was completed, a frame clip of the 35mm rush print would be unsqueezed and blown up to storyboard size with prints distributed to all of the people concerned. It was necessary to keep such an accurate record so that work could begn on other elements of the same shot. For example, each scene of the Discovery spacecraft required a different angle and speed of star movement, and a different positioning and action of the miniature rear-projected image in the cockpit.
All moving images in the windows of the various spacecraft were rear projected either at the time of photography of the model, although as a separate exposure, or later after the model image had been duped using Technicolor Yellow-C y a n-Magenta Masters, or "YCM's.''
A few scenes show a miniature rear-projected image in the window of a spacecraft as the spacecraft is matted over an image of the moon. For this effect the foreground spacecraft was a still photograph mounted on glass and, using a hi-pack camera, the masters of the background image could be printed with a white backing behind the still photo - the photo silhouette producing its own matte. Then the photo and the rear-projected image could be shot as separate exposures onto the same negative. To produce exactly the same movement on each successive exposure, all movement drives and film advances were selsyn synchronized. The mammoth device designed to produce this effect we nicknamed the "Sausage Factory," because we expected the machine to crank out shots at a very fast rate. This turned out to be wishful thinking, however, and shooting became very painstaking and laborious work. Another drawback to printing masters in this way was the fact that lens flaring caused by the white backing would partially print the image within the silhouette. Therefore only very dark backgrounds could be used for these shots.
One of our first serious special effects problems presented itself during the live action shooting. The interior set of the Orion spacecraft (which flew from the earth to the space station) and the interior set of the Aries spacecraft (which flew from the space station to the moon) were both equipped with pinhole star backgrounds outside the windows. These backgrounds were made of thin sheet metal with each star individually drilled, and were mounted on tracks to produce an apparent motion from inside. As shooting began it became apparent that when the stars had the correct intensity in the 35mm print-down, they were much too bright in a 70mm print. And, when the stars looked correct in the 70mm version, they would disappear altogether in 35mm. So star brightness became a compromise, and after all the problems encountered in trying to accurately control star intensity on the set, almost all stars shot subsequent to those interiors were photographed on the animation stand.
The 0xberry animation stand equipped with a 65mm Mitchell camera was used for shooting backgrounds of stars, Earth, Jupiter, the Moon, as well as for rotascoping and shooting high contrast mattes. All stars shot on the animation stand were spatter-airbrushed onto glossy black paper backing and were shot at field sizes of from six to twenty-four inches wide. Extensive tests were made to find the optimum star speed for each shot and great care was taken to control the action so that the stars wouldn't strobe. In almost all shots it was necessary for the stars to be duped, but this became a simpler problem because they required only one record instead of the usual three YCM's.
Backgrounds of the Earth, Jupiter, Jupiter's moons, and others were back-lit Ektachrome transparencies ranging in size from 35mm to eight by ten inches, and these were shot from much larger painted artwork. The Moon was a series of actual astronomical glass plates produced by the Lick Observatory. These plates were used only after nearly a year of effort at the studio to build a moon model - several attempts, in fact, by different artists, and all were unsuccessful.
It may be noted that in only a few effects shots in space does one object overlap another. The reason for this is that normal matting techniques were either difficult or impossible to use. The rigging to suspend the models was so bulky and complex that the use of the blue screen technique would have been very awkward. Also, the blue screen would have tended to reflect fill light into the subtle shadow side of the white models. It became a monumental task merely to matte the spacecraft over the stars, and the final solution to this was meticulously rotascoped, hand-painted mattes.
Since we couldn't afford to tie up the animation stand, or any camera, for very laborious and time-consuming rotascope jobs on so many shots, a unique rotascoping system was devised. Using ordinary darkroom enlargers, equipped with carriers for rolls of 70mm film, each frame-by-frame image was projected onto specially marked animation peg boards, to which the projected image of the perforations had to be visually aligned.
We found that a star could not be allowed to penetrate the edge of any spacecraft image even to the very slightest degree, although it was unnoticeable if the star was extinguished several frames before reaching that image. So to account for the poor tolerances in our visually registered system, each rotascoped cel was painted with a slightly oversized image.
All special effects work involves the standard problems of film steadiness, color correction, and matting, and "2001" was no exception. Since every effects shot necessitated the combining of multiple separate images onto one negative, absolute film steadiness was essential. After trying for months to find some rhyme or reason as to why some shots were steady and some weren't, we began the tedious task of comprehensive steady-tests on every roll of raw stock, every set of YCM's, and every roll of 35mm print-downs.
Another problem that gave us many headaches was the loss of black density due to multiple duped images being exposed onto one negative, and in a space film like "2001" the retention of blacks was very important. Part of this problem could be solved by ordering very dense sets of YCM masters to retain maximum contrast. Most original negatives were shot slightly over-exposed so that a higher printer light would be required to reproduce the image. This helped a little, but if carried too far would take the brilliance out of the whites. These precautions were only partially helpful and any shot involving more than two or three sets of masters would suffer a noticeable greying of the blacks.
The solution was to make at least one element in the scene an original unduped image. Aside from helping to retain the blacks, an original image is naturally preferable to a duped one, and in many cases great pains were taken to keep all elements of an entire scene on the original negative.
Interior view of the Orion commuter ship approaching the Space Station, with all of its scopes showing readouts. Preparation of these readouts for all the spacecraft ate up tons of technical material and required many months to animate.
The first live action shooting on "2001" took place in the giant moon excavation set built on Stage H at Shepperton Studios. The set itself only included a small portion of terrain at one end for the astronauts to walk on, so shots that included the complete Moon terrain, stars, and Earth, were held undeveloped for nearly a year until these other elements could be completed, tested, and then exposed onto the held original negative.
The "held take" shots at the Moon excavation were relatively simple compared to the held takes of moving live action miniature projections. Many shots required that a weightless, gyrating astronaut be moving through space, matted over the stars. For this effect, a 65mm shot of the astronaut was projected onto a small white card, and the camera moved relative to that card to produce the apparent motion. Since this miniature projection was already a form of duping, although it remained sharp and brilliant due to the extreme reduction in size, it was important that this image not go through a further dupe generation. In order to retain the miniature projection as a held take, four separate but identical takes would be shot, using the "Sausage Factory" selsyn system to retain absolute synchronization. Only one of these takes would be sent to the lab for processing, where a 35mm rush print would be made to check color and movement, and a 70mm print would be made so that the rotascoping process could begin. Later, the other duped elements of the shot and the matted stars would be exposed onto one of the held takes, still leaving two more takes to iron out any problems which might have arisen in the first "marry-up."
(Left) Astronaut hangs vertically by wires with camera pointing directly upward. He is shown here at the top of his fall toward camera lens. (Right) Astronaut on wire at bottom, after falling toward camera. (Middle) The scene as it appears in the film, after he had blasted his way through the vacuum of space into an air-lock of the spacecraft.
The models in "2001" are probably the most precisely detailed ever constructed for a film. As soon as the overall design was completed on each model, construction was begun to produce the basic form of that spacecraft, and this process often took several months. Then the arduous task of detailing and painting the model would begin. Massive crews of model detailers worked around the clock for several more months to produce the finished results. Basic construction was of wood, fiberglass, plexiglass, steel, brass, and aluminum. The fine detailing was made up of specially heat-formed plastic cladding, flexible metal foils of different textures and thicknesses, wire, tubing, and thousands of tiny parts carefully selected from hundreds of every conceivable kind of plastic model kit, from boxcars and battleships to airplanes and Gemini spacecraft. A delegation from the production was sent to an international model exhibition in Germany to select the best kits available.
Every minute facet of each model had to be perfect, so that photography would not be restricted in any way, and during shooting the cameras came relentlessly close with no loss of detail or believability.
Each spacecraft was built to a scale which best suited that particular model, without any particular regard to scale relationship between models. Only the Discovery spacecraft and the pod were on the same scale, since they had to work so closely together. Very tricky calculating had to be done for the approach of the Orion spacecraft to the space station because both models couldn't be built to the same scale. Roughly, the Orion was three feet long, the space station eight feet in diameter, the Aries two feet in diameter, the Moon rocket-bus two feet long, and the Discovery fifty-four feet long with a thirteen-inch diameter pod. The main "command module" ball of Discovery was six feet in diameter, and for long shots another complete model of Discovery was built to a length of fifteen feet. All moving parts on the models were motor driven and extremely geared-down since most shooting was at a very show rate due to the necessity for stopping down to small lens apertures to obtain maximum depth-of-field.
"Discovery" spacecraft under construction, showing the beginning of detailing using parts from thousands of model kits
The Moon terrain models required considerable depth-of-field also, and in order to keep the distance from foreground to infinity within a focusable range, they were built with extremely forced perspective. Detail was graduated from very large foreground rocks and rubble to tiny mountain peaks and plains on the horizon in a total actual depth of about five feet. To reproduce in model form exactly what a drawing required, the drawing would be photographed as a 70mm-size transparency and projected onto the work area from the exact point at which the Super Panavision camera would be shooting, and with the same focal-length lens. In some cases we still couldn't hold the depth-of-field even with forced perspective, so the model would be shot as two four-by-five black-and-white stills, one focused on the foreground and one focused on the background. Large prints were made of each, cut out, retouched, pasted together, and then shot on the animation stand.
During the filming of what are probably best termed the "psychedelic" effects for the end sequence, we all joked that "2001" would probably attract a great number of "Hippies" out to get the trip of their lives. It seems now that what was once a joke is fast becoming reality, and as of this writing, I understand that each showing draws an increasing number of these people, who would probably prefer to just see the last two reels over and over again.
Stanley Kubrick strongly emphasized to all members of the production crew that he wished the specific techniques used in the last sequence to remain as unpublicized as possible, so out of respect for his wishes and in appreciation of the wonderful opportunity he gave me and others to experiment and produce these very costly effects, I will describe them only briefly without specific details.
As the black monolith vanishes into a strangely symmetrical alignment of Jupiter and its moons, the camera pans up and the "Stargate" engulfs the screen. For this infinite corridor of lights, shapes, and enormous speed and scale, I designed what I called the Slit-Scan machine. Using a technique of image scanning as used in scientific and industrial photography, this device could produce two seemingly infinite planes of exposure while holding depth-of-field from a distance of fifteen feet to one and one-half inches from the lens at an aperture of F/1.8 with exposures of approximately one minute per frame using a standard 65mm Mitchell camera.
(Left) Designer assembling a bank of photo-electric control devices for slit-scan mechanism. (Center) 65mm Todd-A0 camera with follow-focus device mounted for operation of slit-scan set-up. (Right) Todd-A0 camera with Acme stop-motion motor and selsyn-driven follow-focus mechanism. The slit-scan techique, designed by the author, created the spectacular "psychedelic" sequence at the end of the film, in which the camera seems to be rushing through endless geometrical corridors of color.
After the Stargate, there follows a series of fantastically delicate, apparently astronomical cataclysms. The images implied exploding stars, vast galaxies, and immense clouds of interstellar dust and gas. Without revealing too much detail, I'll merely say that these effects involved the interactions of certain chemicals within a camera field of a size no larger than a pack of cigarettes.
The final series of shots before Keir Dullea ends up in his unusual predicament were done by shooting some fairly unusual aerial scenes, and then juggling the color filters in the YCM duping process. It took months of experimentation to find the key to this technique.
The live action sequences in "2001" involved so many different trick sets, rear projections, and stunts, that the only approach to writing about them is to handle each in the order in which it occurs in the film.
Filming of the 'Dawn of Man" sequence took place entirely on only one stage at the studio. Distant backgrounds for all the action were front-projected eight-by-ten Ektachrome transparencies, using probably the largest front-projection device ever made, and constructed specially for "2001". ' The projector consisted of a specially intensified arc source with water-cooled jaws to hold the oversized carbons, special heat-absorbing g l a s s, giant condensing lenses which would occasionally shatter under the intense heat, special eight-by-ten glass plate holders and positioning mounts, an extremely delicate semi-silvered mirror, and a specially built nodal point head so that the camera could pan, tilt, and zoom without fringing of the image.
To camouflage the varying light transmission rates between rolls of the front projection screen material on the giant 40- by 90-foot screen, the material was cut up into small, irregular pieces and pasted up at random so that slight variations in the transmission rates would merge with cloud shapes or be lost altogether in brilliant sunlight effects. Since the screen occupied an entire wall of the stage, and the front-projection rig was delicate and cumbersome, the sets were built on a giant rotating platform which covered most of the stage floor. Widely varying camera angles could then be obtained with no movement of the screen, and little movement of the projection rig.
During the testing of this front-projection system, it was found that the intense light and heat being poured through the transparency would burn off layers of emulsion in a matter of minutes. Additional heat filters were installed but the only real solution was to expose the plate only during the critical moments that the camera was running. Duplicate plates were used for various line-ups, tests, and rehearsals. Even with such an intense light source, the long throw from projector to screen required lens apertures of around F/2.
The first live action shots in the space sequence took place on board the Orion spacecraft during its journey to the space station. For long shots of the apparently weightless floating in mid-air, the pen was simply suspended on thin monofilament nylon strands. For the close-up reverse angle shots the entire end of the set was floated away, and an eight-foot diameter rotating glass was moved into position with the pen lightly glued to it. The stewardess merely had to pluck it off.
The movie being shown on the TV set in front of the sleeping passenger was a little more complicated. Kubrick wanted shots of a futuristic car, and close-ups of a love scene taking place inside. A crew was dispatched to Detroit to shoot a sleek car of the future which was provided by, I believe, the Ford Motor Company. The exteriors were shot in 35mm, but the interiors were shot without seats or passengers, as four-by-five Ektachrome transparencies. Using these as background plates for a normal rear-projection set-up, on actor and actress were seated in dummy seats and Kubrick directed the love scene. Shot on 35mm, this was cut together with the previous exterior shots, and projected onto the TV screen using a first-surface mirror.
In the cockpit of the Orion spacecraft, during its approach to the space station we begin to see a few of the 35mm animated, rear-projected computer displays on multiple screens. Throughout the space sequences these displays depict the activities of computers on board the Orion, Aries, Moon Rocket Bus, Discovery, and Pod spacecraft.
To produce thousands of feet of continually changing graphic readouts to cover the multitude of screens in "2001" would have been an impossibly long job using ordinary animation techniques. We terminated work with the local animation camera service, set up our own 35mm Mitchell camera with stop-motion motor, and with the help of a very talented and artistically oriented cameraman, we began the job of pasting up and juggling around artwork under the camera as we were shooting. In this way sometimes as much as a thousand feet of active, colorful, diagram animation could be produced in one day. Specific readouts showing docking alignments taking place, testing procedures under way, and other specific story points were not as fast and easy to shoot, however, and the job of producing all of the read-outs for "2001" took nearly a year.
The interior of the space station was a giant curved set over three hundred feet long, and sloping up at one end to nearly forty feet. It may be noticeable that in the long shot of two men approaching the camera from the far end, their pace as slightly awkward, and this was due to the very steep slope at that end of the set. Most action took place in the more comfortable area at the bottom. The Earth image seen through the window of the space station was a rear-projected four-by-five transparency in a special rotating mount.
Aboard the Aries spacecraft on its trip to the moon, in the passenger compartment a stewardess is watching another TV screen, and again the action was directed and edited by Stanley Kubrick. The galley scene of this spacecraft where the stewardess comes in, picks up a tray, and then walks up the wall to exit upside down, was filmed using a rotating set with all lights and the camera secured to the rotating structure. The stewardess merely remained upright as the set and camera rotated around her.
The Discovery spacecraft included the most exciting sets of the production, and the most spectacular of these was the giant centrifuge. At a cost of over $750,000 the massive forty-foot diameter structure could rotate like a ferris-wheel. With the actors either standing, walking, or even running at the bottom of the set, cleverly thought-out camera angles made it appear that the actors could stand upright at any angle around the circular set.
In one of the most difficult shots Gary Lockwood was strapped into his seat and had to hang upside-down pretending to eat glued-down food while Keir Dullea climbed down the ladder at an angIe 180 degrees opposed to Gary. As Keir began to walk around the centrifuge toward Gary, the centrifuge was slowly rotated until Keir and Gary were together at the bottom. The camera, which was locked down to the centrifuge floor, was then at the top. For other shots the camera was mounted on a specially made 360-degree tilting platform which was bolted to the floor of the centrifuge, and the camera operator sat in a ferris-wheel type seat which kept him upright at all times. Other shots were done with the camera mounted on a small rubber-tired dolly, which would be pulled by grips frantically clambering up the inside of the centrifuge as it rotated, trying to keep ahead of an actor shadow boxing at the bottom.
All lights and large banks of 16mm projectors also rotated with the set, so that exploding bulbs, loose junk, and reels of film constituted a serious hazard to people nearby. Hard hats had to be worn by everyone involved, and the control area from where the centrifuge was driven, and action directed by closed-circuit television, was netted over with chicken wire and heavy plastic.
The cylindrical corridor which linked the hub of the centrifuge to the rest of the ship, was constructed of two separately rotating sections, with the camera mounted securely to the corridor end. With the hub end rotating, the actors could walk down the static corridor and then step onto the hub as the port came to a position at the bottom. As soon as they stepped across onto the hub, it would stop and the corridor would begin to rotate in the opposite direction. From the camera's point of view the apparent rotation remained constant, but the actors seemed to be completely defying the law of gravity.
Other apparently weightless effects, which took place during the excursions outside the spacecraft, and in the "Brain Room," were created by suspending the astronaut on wires and then shooting from directly below so that he would cover his own means of support.
Several versions of the full-sized pod were used during the Discovery sequence. Three dummy pods were used in the pod-bay, two of which had operational doors, but only roughly mocked-up interiors. A separate interior pod set was built which included all the instrumentation, controls, and readout displays. Finally, a full-sized pod was built with completely motorized, articulated arms. It took ten or twelve men at long control consoles to simultaneously control the finger, wrist, forearm, elbow, and shoulder actions of the two pod arms, and the interior of that pod was a maze of servos, actuators, and cables.
Possibly one of the most unusual aspects of the live action photography on the interior sets of "2001" is that almost all lighting was an actual integral part of the set itself, and additional lighting was used only for critical close-ups.