Why Jurassic Park Used Less CGI Than You Remember

Six minutes of digital dinosaurs, a mechanical tyrannosaur in the rain, and the work of making them belong to the same film.

The tyrannosaur attacking the children’s car in Jurassic Park has several bodies. Stan Winston’s crew built a machine that could bring its head close to the actors and the vehicle. Industrial Light & Magic created a digital animal that could move through the wider scene. The sequence passes between them while keeping our attention on the children trapped inside. The practical creature’s contribution is especially visible in the close encounters with the car. [4]

That continuous animal occupies much more space in the imagination than the digital footage occupies in the film. ILM’s own retrospective gives the familiar figures: approximately six minutes of computer-generated dinosaurs, intercut with nine minutes of practical puppets. Its project page describes fewer than sixty shots of fully CG dinosaurs. [1], [2]

Those numbers describe the creatures, rather than every use of digital technology in the production. Compositing, the process of combining separate image elements into a finished shot, also involved substantial digital work. Counting minutes of rendered dinosaurs gives us one measure of the film’s effects, with clear limits. [2]

The useful question is how such a small amount of footage could sustain the impression of an island full of animals. The answer runs through the design of individual shots: what the creature needed to do, what could be photographed on set, and what the next cut would ask us to believe.

DINOSAUR SCREEN TIME

6 minCG dinosaurs

9 minPractical puppets

Approximate screen time reported by ILM. Not a breakdown of the whole film.

A production plan that changed after a test

The original approach combined full-size mechanical dinosaurs with animation of miniature models. Phil Tippett’s team brought expertise in stop-motion and go-motion, methods in which a physical model supplies the performance. Go-motion added movement during exposure to produce the blur associated with a photographed moving subject. [12]

At ILM, Steve Williams and Mark Dippé developed digital dinosaur tests that helped persuade Steven Spielberg to change that plan. With Dennis Muren supervising ILM’s visual effects, the studio had to turn a promising demonstration into finished creatures under production deadlines. Tippett and his colleagues remained involved as the work moved into computer animation. [1]

A full-body digital animal could travel through a landscape without carrying the support machinery of a large puppet. Dippé later described CGI’s role in terms of those practical demands: moving herds and wide views of standing dinosaurs were among its principal assignments. Close interaction could draw on the physical creatures already being built. [5]

This division helps explain the limited running time. Digital production was taking on selected problems inside a film whose effects involved several departments. It also explains why a rule such as “close-up equals puppet” will eventually mislead you. Framing was one consideration. The movement, contact and particular demands of the shot could change the choice.

For the wider history of the studio behind those tests, see Industrial Light & Magic: The Company That Built the Impossible.

A dinosaur close enough to touch

A physical creature brings useful evidence into the photograph. Its skin catches the same light as the set. It blocks whatever stands behind it. When an actor approaches it, both occupy a space whose distances can be judged on the spot.

Those advantages require a machine capable of giving a performance. John Rosengrant, who worked on Winston’s dinosaurs, recalled the difficulty of moving such large constructions naturally. Hydraulic power alone was insufficient; the controls had to keep the movement from looking mechanical. Building the enormous skins and molds was another substantial part of the job. [6]

The tyrannosaur’s construction also limited what could be filmed. Artist James Gurney visited the set in 1992 and later described upper bodies mounted on motion bases, with separate feet for close shots. During the visit, he saw the crew drying foam-rubber skin that had absorbed water from the rain machines. [7]

A wet dinosaur gave the camera a richly textured surface, while giving its operators an additional problem. The benefits and difficulties came from the same material. Physical effects demanded decisions about where the apparatus could move, what the camera could reveal and how the crew would keep it working between takes.

In the car attack, the close physical encounters establish the creature’s scale against objects we understand. A head beside a vehicle gives us a more immediate comparison than a dinosaur isolated against a landscape. The wider digital views can then extend its movement through the scene. Read this way, the cutting lets each version supply evidence that supports the other.

That interpretation depends on continuity. An unexplained change in the animal’s appearance or rhythm of movement could draw attention to the substitution. Matching a creature across techniques was part of creating it.

Even a digital animal could cast a practical shadow

The lighting problem extended beyond making dinosaur skin look plausible. A creature had to affect the scene around it.

In an interview published during the film’s original release, cinematographer Dean Cundey explained that some shadows were produced on set by placing a flag or shaped cutout in front of a light. Others were generated and combined with the image digitally. The choice depended on what the lighting arrangement allowed. [3]

This is where a simple split between practical and digital starts to lose detail. The visible animal and the evidence of its presence could be made by different means. Their timing still had to agree. A shadow arriving before its supposed cause, or moving in the wrong direction, would describe a different event from the creature above it.

Cundey also described using physical dinosaur stand-ins to establish lighting that the digital team could reproduce. That gave the computer artists a photographed target for the creature’s appearance. [10]

For a 3D artist, this is a useful distinction between completing a model and completing a shot. The model can look convincing on its own and still sit awkwardly in the photographed environment. The light needs a plausible direction; contact needs weight; objects passing in front must interrupt the creature correctly. Each relationship contributes to the impression that the camera recorded one event.

Night and rain change those relationships. They can leave parts of a body difficult to see, while highlights make other parts unusually prominent. Darkness therefore changes what an effects shot must solve. It offers no exemption from matching the visible evidence.

MAKING THE METHODS AGREE

Practical performanceA creature photographed on set.

Digital animationA creature added to the image.

---

One consistent sceneLight and shadows agree. Feet and objects line up. Reflections describe the same motion.

Conceptual explanation, not a reconstruction of a production pipeline.

The kitchen had to reflect the same raptor

The kitchen sequence places the dinosaurs among work surfaces, cabinets and narrow routes through which the children try to escape. It gives the viewer a room whose scale and obstacles remain understandable while the creatures move through it.

Cundey recalled the particular difficulty of its brushed stainless steel. The crew had to hide unwanted reflections of equipment while preserving surfaces into which the effects artists could add the dinosaurs’ reflections. He also credited Spielberg’s willingness to adjust shots when a technical difficulty emerged. [8]

Practical performance remained part of that scene: Rosengrant himself performed as a raptor. The published production account also identifies the sequence’s combination of animatronic and CG creatures. [6], [3]

The steel makes the integration problem especially easy to understand. A moving body can occupy one part of the picture while its reflection occupies another. Both must describe compatible motion. The set supplies familiar visual cues against which the added creature will be judged.

This puts pressure on planning. A cabinet edge can conceal part of a dinosaur, but the animal must still emerge at the right height and speed. An eyeline can direct our attention toward empty space, but the finished raptor must occupy the position the performance implies. The work begins with the photographed scene and continues through animation and compositing.

Some of the kitchen raptors’ computer-generated movement began with an animator handling a physical object. Their performances carried a direct connection to Tippett’s earlier methods.

The hands behind the digital movement

The Dinosaur Input Device, or DID, was an articulated model equipped with sensors. An animator could pose it, save the position and move on to another pose. That information could then drive a digital dinosaur.

In Ian Failes’s oral history of the device, Randal Dutra describes using it for the kitchen raptors. Adam Valdez explains that the resulting animation needed varying amounts of computer adjustment, including work on foot contact and alignment with objects in the filmed scene. Dutra also recalls that smoothing the movement could reduce accents he had put into the performance. [9]

The device preserved access to skills developed through physical animation, while introducing its own problems. A planted foot on the model did not guarantee a planted foot in the final image. The animator’s intention still had to survive the transfer.

Observation supplied another connection between the methods. ILM art director TyRuben Ellingson recalled the team studying animal movement, including elephant reference footage that prompted discussion of mass and weight. The creatures needed behavior that made their bodies comprehensible. [4]

A viewer has no memory of watching a living tyrannosaur. There are still familiar expectations against which to judge one: a foot supporting weight, a head turning toward a stimulus, a body recovering from a change of direction. Animation organizes those clues over time. Texture alone cannot supply them.

Six minutes can occupy an entire scene

The dinosaur remains part of a scene while the camera shows a frightened person or the space through which an attack might come. Sound helps maintain that presence. Gary Rydstrom’s work on the film included sound design and mixing, another part of the construction excluded from a count of visible digital creatures. [11]

Cundey’s contemporary account describes a gradual reveal, with lighting and framing controlling how much the audience could see. He also ties the effectiveness of the threat to whether viewers care about the people facing it. [3]

That offers a practical explanation for the mismatch between running time and memory. A creature’s appearance changes the meaning of the surrounding shots. Once a predator has entered the space, a view of someone hiding can sustain the encounter without showing the predator again. Its visible seconds are part of a longer dramatic event.

The famous six-minute figure cannot tell us how much CGI another film ought to contain. It records one production’s choices and constraints. Reproducing the proportion would leave the difficult decisions untouched: where the camera belongs, what the animal must do, how it affects the environment and when the scene needs to return to a human face.

In the kitchen, that work reaches down to a foot meeting the floor and a reflection moving across steel. Decisions made while building and lighting the set become conditions the animator must satisfy later. A production that settles those relationships early gives every department something concrete to match, long before the finished creature reaches the screen.