On account of the technology involved, and the toil and time required, animators are almost always working well within the outer limits of their theoretical ceiling. But if visual effects are supposed to be a seamless superimposition, then its animators ought to be at the boundary between an impossibly accurate portrayal and the constraints of whatever method they employ. “Ultimately no matter what tool you’re using,” explains Phil Tippett, “whether it’s stop-motion, go-motion, computer graphics – you’re learning how to work with the limitations of the tools that you have. You have a mental image in your mind of what the thing should be moving like, and you’re constantly moving – no it shouldn’t be like this, it should be like that, it should be like this – and just try to keep all this stuff moving towards what, you know, see in your mind’s eye.” While such a painstaking process always revolves around minor frame to frame developments, this incremental way of working is frequently furthered by major leaps in science and technology.
In the digital age, in which effects-driven studio tentpoles dominate multiplexes twelve months of the year, it has become popular for fans and critics alike to bemoan the omnipresence of computer graphics. While the public at large flock to these films in ever-greater numbers, some claim that advancing technology increasingly detracts from the cinema experience. As their familiarity with computer animation breeds contempt, they wax nostalgic for the days of practical effects. While Ray Harryhausen would not count himself among these detractors, he notes “I still get letters from dyed-in-the-wool fans saying they prefer the old hands on technique” (Cotta Vaz, 33). These fans acknowledge its undeniable charm, but neglect the similarity between the use of stop-motion and the implementation of computer graphics, which is that the techniques themselves have often determined the form of a film, and that the effects therein are its reason for being.
As it turns out, the contemporary cinema of effects was born out of an altogether different approach. Namely, effects should emerge as a function of what the film must accomplish, rather than shape its superstructure from beginning of the production onward. As these demands might have little basis in what was deemed practical or thought possible, the focus of the modern effects house would be to accommodate a range of ideas through new and often radical approaches. Building on the mythology of effects pioneers Willis O’Brien and Ray Harryhausen, the story of contemporary effects begins in a refurbished warehouse in Van Nuys, with the foundation of Industrial Light & Magic and the creation of effects of Star Wars. The chief innovation of this film was the motion control camera, developed by John Dykstra and used to depict fluid flight through bluescreen photography and computer programming. In following features, the same technology would take traditional stop-motion animation to a whole new level.
As their first project shown only short snippets of stop-motion, it was not clear what role the effects technique would play at ILM. While budget and time constraints prohibited its appearance in anything besides a single scene, these are not the only reasons its use was limited. “George Lucas wasn’t too fond of stop-motion before the chess sequence in Star Wars,” Dennis Muren revealed after the completion of The Empire Strikes Back. “He felt that in the films he’d seen the results didn’t look real, and the technique wouldn’t work for some of the ideas he had in mind” (Mandell, 5). As the scene featured virtual pieces in the form of holograms, realism was of little consequence, but the puppets, animated by Phil Tippett and Jon Berg out of their garage, performed admirably. Compared to the vector graphics shown during a mission before the climactic battle, this holographic display was light years ahead. Even by the standard of living animals, these creatures were not only interesting but believable. The success of this vignette prompted Lucas to form a dedicated stop motion unit, which would have its own studio space in the newly constructed San Rafael facility.
Even after establishing this new unit, the supervisors at Industrial Light & Magic considered alternative approaches to key effects, such as the lumbering mechanical walkers. Meant to stand more than twenty meters high, miniature versions were the only option, but even then, stop-motion was not the first choice. “There was some talk about doing them motorized,” Muren recalls, “but I think we’d still be working on that if we’d actually tried to make real walking machines” (Empire). Even thirty years later, robotics lack the subtle performance quality of dimensional animation, which enabled the war machine to dynamically shift its weight, and fire on a passing speeder as though it were swatting a fly, an exciting shot emphasizing the power of the Empire and its unstoppable assault.
Furthermore, the major problems of stop-motion are of less consequence when animating robotic figures. The stroboscopic motion produced by the technique would not only be consistent with the subject, but these large models were moved on average between on sixteenth and one thirty-second of an inch between frames. The same could not be said of the tauntauns, bipedal beasts of burdens the heroes ride early in the film. As the creature would be required to run, its great strides meant moving the puppet as much as an inch each frame, exacerbating the stroboscopic appearance of stop-motion. “Pushing a puppet that far per frame was asking an awful lot. Many of the basic running and body moves were almost an inch-and a-half, which would have resulted in a very stroby image,” Tippett explained (Mandell, 37). As such movement would further degrade the persistence of vision upon which the technique depends, they began to search for a better approach.
A chief limitation of stop-motion is that, when photographed, the model remains completely motionless. As a result, each frame of film is perfectly clear and crisp, in contrast to the footage with which such effects are composited. The average shutter speed of 1/48th of a second is relatively slow for moving objects, effectively smooths out the movement. Although the eye adjusts to such peculiarity when fixated on stop-motion, the difference at best noticeable whenever animated puppets are intercut with live footage (Smith, 92). While the only way to achieve this motion blur would be to move the puppet while the film is exposed, a solution presented itself in the form of the computer control mechanism that made possible the fluid flight of space vessels in the first film. “It was just a matter of having an open mind and access to equipment,” explained Tippett (Mandell, 37). A hybrid of traditional a stop motion armatures and computer control mechanisms, the tauntaun would be animated by hand, but could be moved during exposure to produce motion blur, albeit only along its X and Y axes (Wilson, 42).
Even as The Empire Strikes Back featured the most advanced stop motion to date, the technique remained subject to criticism. In spite of the improvements they had made, a small number of audience members still noticed that which some preview audiences thought to be slightly off. “About two percent of the preview cards from Empire Strikes Back had mentioned the stop motion-motion tauntauns didn’t look quite right,” Dennis Muren remembers (Duncan, 69). Confronting this criticism, ILM developed the technique to the point that they could photograph individually moving parts, instead of a posed puppet pushed along a single path. “What had kept the idea from being done before,” Muren explained, “was that everyone had thought that it had to be done internally (Wilson, 42). In the case of Vermithrax Pejorative, the winged wyrm of Dragonslayer, such an approach would be impossible, as the puppet was too small to house motors to drive its movement. Inspired by traditional Japanese puppetry, they reasoned that the puppet could be attached to motors via rods extending from the motion control mechanism and attached at key joints (Smith, 93).
Even with this crucial conceptual hurdle behind them, the crew still had numerous artistic and technical challenges to overcome. The original design of the dragon would not have worked as a stop-motion puppet, and Phil Tippett and Ken Ralston proposed many alternatives, even as they were hard at work on Empire. (Years later, Tippett, no longer at ILM, reworked the design for the key creature on Dragonheart. The most complicated computer character to date, and an enormous challenge for the effects house to model and animate, the dragon was dubbed “Tippett’s Revenge” for the introduction of computer graphics, which replaced traditional effects animation). “One feature of the original design,” explained David Bunnett, the artist responsible for the dragon concept, “which is different from the final one is that the original was much more gaunt and ragged. But when you have a model that moves, you have to limit yourself; because if you have a design that exposes the musculature to a substantial degree, it’ll look fine when its stable, but when you see it moving you’re going to expect to see those muscles doing things,” (Wilson, 45). It would take twelve years, and the increased sophistication of computer technology, before animators gained control of these anatomical details.
As the dragon would be connected to a motion control rig, the project required improvements to this system. To date, the most complicated motion control maneuver had needed nine motors, while new rig would demand sixteen channels. According to Stuart Ziff, who engineered the system, this upgrade was evolutionary rather than revolutionary. “It’s no different from standard motion control except that we connect it to the puppet,” explained Ziff. “The significance is that we got it working” (Wilson, 43). One other important difference is that the new mover operated with one-to-one gearing, which enabled rotations as small as a tenth of a degree. The low gearing with which standard motion control movers are equipped required minute changes to be made through many revolutions. This improved system would have allowed the dragon to be moved in real time, although it was never practical to do so, as repeatedly performing rapid movements would cause damage to the puppet. Furthermore, the device could only drive gross body movements, with rods attached to the wings, head, and torso of the dragon. This meant that small details, such as its claws, or the wizard-in-training who leaps onto the dragon, would be animated in traditional fashion after the primary moves were programmed.
Although results surpassed expectations, the process had further complicated the already painstaking process of stop-motion animation. The motion control rig had to be configured differently for every single shot. Even when blue screen photography, which was required on twenty more shots than originally planned, freed the creature from its miniature cavern, the animators need to ensure that the rig and its rods never passed between the puppet and the lens. Camera placement was often so precise that shifting it a single inch would expose rods or another revealing mistake (Wilson, 46). While fraught with challenges, this new technique, deemed ‘go-motion,’ was clearly the next generation of dimensional animation. In light of its extreme difficulty, however, they created another coinage. “It was so painful,” Muren recalled, “we came up with an expression on Dragonslayer: ‘No mo go mo’” (Duncan, 69).
The technique continued to be used in a reduced capacity. When the children in ET: The Extra-Terrestrial ride their bikes across the sky, motion control was used to pedal the wheels and their legs, thus producing the privileged blur (Smith, 96). A decade would pass before ILM began another full-fledged go-motion effort, when Amblin Entertainment began the unusually long pre-production phase for Jurassic Park. While full-size animatronics would be featured in the majority of shots, it appeared that the cumbersome go-motion technique would be the only way to achieve the believable dinosaurs that were crucial its success. For Phil Tippett, who had left ILM to form his own studio and produce the stop-motion short Prehistoric Beast, this appeared to be the perfect opportunity. He would now be able to bring these animals to life using the most sophisticated techniques to date.
Even discounting the extreme difficulty of go-motion, the technique was not without its flaws, as its most perfect application would inevitably be plagued by the problems of a more traditional approach. After reviewing the tyrannosaurus and velociraptor animation tests, director Steven Speilberg was impressed but unsatisfied. “Tippett had perfected the motion blur that gave go-motion a closer resemblance to real life,” Speilberg offered, “but it wasn’t one hundred percent; it was still jerky” (Jurassic). Effects supervisor Dennis Muren agreed. “It seemed like if you did that work as good as you could possibly do,” mused Muren, “it wouldn’t be up to the quality of what audiences today expect and deserve and what we’re capable of giving them.” Encouraged by the success of the liquid metal T-1000 in Terminator 2, artists and engineers at ILM continued to experiment with computer graphics.
As go-motion was a complicated, specialized process, it would also be expensive. These high estimates inspired an ongoing search for any alternative that could recreate or improve upon the blur seen in go-motion. One promising idea was to add motion blur to conventional stop-motion using a digital morphing program. With no more than a few days work, they were able to achieve a promising result. “We got close to seventy percent reality,” Muren explained. “We did this test in a chaotic three months when we were approaching the project from all sorts of different angles” (Duncan, 52). The most important of these experiments were conducted independently by a handful of dedicated individuals, who planned to prove that lifelike animals could be created on the computer.
Even among traditional stop-motion animators, there was increasing interest in the potential of computer graphics. This enthusiasm had often been met with skepticism, as when Craig Hayes demonstrated the infant software to Phil Tippett. “I showed it to him,” Hayes remembered, “and he was like, ‘Yeah, that’s cute.’ He wasn’t that interested. Back then, all you could do with computer graphics was render a sphere; but you knew the next year you would be able to render something more complex; and the year after that you would be able to render a monster” (Duncan, 77).
With Jurassic, the challenge was not merely to model a movie monster, but to create lifelike animals. Working alone, animation supervisor Steve Williams began work on a computer model of a tyrannosaurus, starting with its skeleton. “I showed it to Dennis,” Williams recalled, “and his interest went up a notch” (Duncan, 55). After Kathleen Kennedy and Frank Marshall saw this skeletal form walking and running, they authorized ILM to construct a complete computer model of the tyrannosaurus. In order to create a lifelike animal on the computer, the artists and engineers at ILM would need to make massive improvements to their technical systems.
One of the primary challenges was to match these computer-generated creatures to the full-size animatronics already under construction at Stan Winston Studios. Although maquettes produced as reference could be scanned into the computer, this process posed a number of problems, even before the model had been digitized. The 1/5th scale tyrannosaurus “was too big for the Cyberware scanner,” explained effects supervisor Mark Dippé, “so we cut it up into pieces, scanned each piece separately, then put the pieces together on the computer” (Duncan, 55). There were other complications that developed as a result of this approach. “We took the highest resolution that the Cyberware data could give us, which was extremely dense – tens of thousands of points – but that wasn’t manageable because we would have to animate all of those points.” These complications meant that is was only useful in the case of computer creatures that needed to match perfectly their physical counterparts. As the ostrich like gallimimus would be portrayed exclusively through computer generated imagery, it could be modeled manually, resulting in a model that was much less complicated. “Another advantage to having fewer points was that we got a surface that was much smoother and rounder; so the galli had a more fluid, muscular feeling” (Duncan, 87).
Although dimensional data could be translated directly to the computer, pigment and other texture needed to be painted by hand. While people have been painting by hand for many centuries, they difficulty of mapping a texture to a 3D mesh stems from the fact that the image itself is necessarily flat. (Those unacquainted with this problem should imagine the difficulty of pasting wallpaper to a tyrannosaurus). To minimize distortion, the texture can be unwrapped as if skinning a polygonal pelt, but this flattened version inevitably yields an imperfect result. John Schlag and Zoran Kacic Alesic solved this problem by authoring Viewpaint, an application that would allow computer models to be painted in three dimensions (Duncan, 66). With this newly developed software it became possible to create textures that were not only highly realistic, but a precise match to the maquettes scanned into the computer.
The remarkable realism of these computer creatures meant that the cumbersome go-motion technique would no longer be needed. Go-motion, as it turned out, was not only less believable and less flexible than computer graphics, the traditional approach would also have been more expensive. As the pioneers of Industrial Light & Magic heralded a new age of effects, others mourned the death of the traditional, hands-on approach. “The change was devastating,” Phil Tippett recalls. “That was an excruciating moment,” remembers Jules Roman, his wife and business partner. “We knew that ILM was exploring computer graphics of course, and we knew it was coming along; but we had no idea that it was moving along to the point that they could use it for the dinosaurs in Jurassic Park” (Duncan, 77).
Computer animation offered an unprecedented degree of flexibility, but at this developmental stage, there was no one particularly suited to bringing these shots to life. On one hand were stop-motion animators accustomed to directly manipulating physical puppets, and uncomfortable performing similar tasks on the computer. Those working with the software had backgrounds in 2D animation, but not in representing lifelike animal behavior. Even as the production would forego stop-motion animation altogether, many believed the film would benefit from the influence of that traditional approach. “I didn’t want to leave out all that animation talent,” insisted Dennis Muren. “I wanted to be able to use the skills of Phil and Randy Dutra and Tom St. Amand. It would have been too scary, and downright stupid, to go into that show with the talent of those animators who had studied and knew animal behavior,” (Duncan, 78).
The process of representing how these animals might have moved began during preproduction, as Tippet and his crew created moving versions of key effects sequences. These animatics employed their traditional stop-motion techniques in order to better represent the look and feel of the final film. “Storyboards don’t show you any of the temporal cadence, the sequence, the timing of it,” explained Tippett. “And the animatics allowed us to block out the entire sequence,” (Jurassic). Not only did these provide a template that would make the filming of complicated sequences more straightforward, it also served as a reference for how the dinosaurs would move.
While this coincided with his supervisory role, and allowed the animators at Tippett Studios to exercise their expertise, their work would be limited by the context of a computer interface. “We are hands on guys,” explained Tom St. Amand. “We’re used to actually walking up to the puppet, making each of these moves by hand. I’m not used to sitting down at a keyboard and having to hit buttons. It’s kind of like animating with boxing gloves on” (Jurassic). In order to accommodate a kinesthetic approach, they began work on a device Dennis Muren had conceived of while at work on another project. “During Terminator 2,” he explained, “I had discussed with Phil an idea for building such a device. Jurassic Park presented the ideal opportunity to finally do it” (Duncan, 56). This invention, which came to be known as the dinosaur input device, or DID, resembled a stop-motion armature but could be used to translate movements directly to the computer.
St. Amand, who had sixteen years experience building traditional armatures, machined the mechanical framework for the input device. “In a sense, the DID armatures were not much different from the usual kind we would make,” explained St. Amand, “except that they were all hinges and swivels, without any ball joints. They had to be one-axis joints – either up-and-down or side-to-side – because we had to mount encoders onto each of them” (Duncan, 58). From an engineering standpoint, this was the only major difference between the DID and a conventional armature. The other important consequence of accommodating the encoders is that the framework needed to be much larger than that of a typical puppet. “At every joint on the armature,” explained Craig Hayes, “were these little black boxes about the size of four sugar cubes put together. Each box consisted of a bundle of wires which came off at that navel of the armature, a big umbilicus.” These cables transferred all of the data regarding the physical configuration of the armature through an electronic interface and into the digital realm.
With the device connected to the computer, it could be mounted to model movers, and manipulated as though the techniques developed for go-motion. The animators could program the gross movements of the body, then return to the beginning of the timeline to animate the arms, legs, head, and tail on a frame-by-frame basis. In this regard, the DID functioned in much the same way as the mother dragon puppet used in Dragonslayer. While the device employed similar technology, this computerized approach offered many advantages over its photographic counterpart. For example, the rubber skin covering a conventional armature might prevent an animator from making minute adjustments. When an armature is freed of this material restriction, an animator can easily tighten or loosen joints as necessary for certain maneuvers, and judge precisely the range of movement available a particular joint (Knep, 2). This device proved intuitive relative to contemporary computer interfaces, and even easier to manipulate than a traditional puppet. Best of all, the fifteen shots animated with the device featured all the fluidity of computer generated images, while retaining the kinetic character of traditional stop-motion.
While the dinosaur input device, as it continues to be called regardless of application, functioned admirably, an indispensable intermediary at this transitional stage, such tools are rare today. Fully digital rigs are constantly being improved to better approximate the intuitive process of manipulating physical armatures. In its simplest form, a digital skeleton consists of nothing more than a series of joints linked together in a hierarchical chain. In the case of bipeds, all joints are linked to the pelvis. When this parent joint is moved, the rest of the joints in the skeleton move with it. When it is rotated, the entire model will change direction. Such a setup is difficult to work with, and would be especially cumbersome in the case of dinosaurs, which have huge heads, long necks, and powerful tails that move independently of their bodies. Softimage solved many of these problems through the introduction of inverse kinematics, a method of organizing movement that was derived from robotics. As described by the software developer, “you define the goal of the action by positioning the effector of a chain and Softimage calculates how to position the rest of the chain.” Like the rods attached to a traditional puppet or a go-motion rig, the movement of this key joint affects adjacent ones. This more closely resembles instinctive motion, wherein wanting to reach out and touch something you simply move your hand towards it, rather than worry about how to rotate your shoulder or bend your elbow.
Not only could this new technology be made to mimic familiar ways of working, but it opened the doors to brand new possibilities, enabling lifelike details that would have been impossible to achieve through stop-motion. The fluid movement of computer graphics could be accompanied by realistic skin and bulging muscles. By adding simple, spherical influences to the dense network of points composing a computer character, it was possible to simulate internal organs and musculature, and fit their motions to the primary animation. These subtle enhancements are best illustrated by the gallimimus stampede, during which the tyrannosaurus waits in ambush at the edge of a forest. When the tyrannosaurus breaks her cover, she makes long, powerful strides, the muscles in her legs rippling with the impact of her six ton body. As she violently shakes her prey, her muscles slide back and forth under her skin, as if attached to bone. While Willis O’Brien was able to simulate breathing by inserting football bladders in his puppets, animators could now create a convincing illusion of mass and tissue, a union of flesh and bone underneath the skin of a lifelike animal.
More importantly, and indeed more spectacularly, the scene could never have been accomplished with traditional techniques, as it featured almost two dozen dinosaurs. “Those kinds of things would take forever to do in stop-motion,” explained Ray Harryhausen. “Each one of those skeletons in Jason and the Argonauts had to be animated separately. You couldn’t just take one or two and then keep multiplying the same thing. The fact that you can make herds of creatures now is incredible” (Cotta Vaz, 33). Animating herd of twenty animals using stop-motion would require the configuration of 480 unique poses every single frame. On the computer, it was simply a matter of perfecting a single run cycle and duplicating the dinosaurs to create a full herd. Of course, the process is still labor intensive and dependent on practiced, studied, and inspired human input. “I still think CGI is basically a tool,” Harryhausen clarified. “It can certainly help animation, but there still has to be somebody to push the buttons properly.”
Indeed, the increasing specialization associated with every facet of computer graphics, along with the ever greater volume of work available, means that there are more and more people pressing more and more buttons. While the The Valley of Gwangi features 400 or 500 effects shots, the most Ray Harryhausen ever animated by himself, such a undertaking would be inconceivable today. Speaking with Harryhausen, Phil Tippett could not help but be astonished. “My facility, with 150 people couldn’t do 500 shots right now,” he suggested. “How many people worked on The Phantom Menace? A thousand, 900 people? It’s a world of technology, where everybody has a part of the whole picture” (Cotta Vaz, 155). At the same time, computer graphics have evolved such that a single artist could render sophisticated effects from his personal computer. Just as digital video has become available to a wide range of practitioners, so has the omnipresence of computer graphics continued to open new pathways through animation. “As a result of this huge technological breakthrough, you can do anything that you want,” Tippett mused after the release of Jurassic Park. “It’s opened everything up. You’re no longer constrained by materiality, you can do what you imagine.”
Cotta Vaz, Mark “Harryhausen and Tippett: The Motion in Our Minds,” Cinefex 80 (2000): 141-156.
Duncan, Jody. “The Beauty and the Beasts,” Cinefex 55 (1993): 42-101.
Duncan, Jody. “The Tippett Touch,” Cinefex 121 (2010): 58-97.
Empire Strikes Back, The. Screenplay by Lawrence Kasdan and Leigh Brackett. Dir. Irvin Kershner. Perf. Mark Hamill and Harrison Ford. Fox, 1980.
Harryhausen, Ray, and Tony Dalton. Ray Harryhausen: An Animated Life. London: Aurum Press, 2004.
Knep, Brian, Craig Hayes, Rick Sayre, and Tony Williams. “Dinosaur Input Device.” 1993.
Making of Jurassic Park, The. Written by John Schultz. Dir. John Schultz. Perf. James Earl Jones. Universal, 1995.
Mandell, Paul. “Tautauns, Walkers, and Probots,” Cinefex 3 (1980): 4-41.
Rickitt, Richard. Special Effects: The History and the Technique. London: Aurum Press, 2007.
Shay, Don, and Jody Duncan. Making of Jurassic Park. New York: Ballantine Books, 1993.
Smith, Thomas G. Industrial Light & Magic: The Art of Special Effects. New York: Del Rey, 1986.
Wilson, S.S. “Dragonslayer,” Cinefex 6 (1981): 30-61.