Home
Next Event
Past Events
Resources
Volunteers
What's New
ACM SIGGRAPH

Presented by Rick Sayre, Ronen Barzel,
Rich Quade, and Hal Hickel
Pixar

Tuesday, November 28, 1995

Notes by John Peterson

After showing a trailer for Toy Story, the Pixar crew showed an early clip from John Lasseter's Disney days. A test for a project based on "Where the Wild Things Are", it featured Lasseter's hand animated characters on 3D computer backgrounds done by MAGI Synthevision. This was an early Disney foray into computers, but "they didn't get it." Pixar (then Lucasfilm) saw the test and did get it; they hired Lasseter.

Next an early test for Toy Story was shown. This piece was done just as the film got a "green light" for production. In it, Woody was a large ventriloquist's dummy with a hinged jaw and Buzz was a small GI Joe sized action figure. These early characters had some problems, for example ventriloquist's dummies were sort of an obscure toy (with evil connotations in other films) and the jaw was hard to animate. The size difference between the two also made staging very difficult. The plastic doll of version Woody proved much easier to animate. Over all, most of the models changed significantly from the initial tests to the final picture.

The Story Reel

The Story Reel is the initial skeleton that the film is built around. It starts off with the animators reading script lines while the corresponding story board sketches are displayed. This provides the rough timing for the film. Later, sound track is re-recorded with the "expensive guys". As portions are completed, storyboard sketches in the reel are replaced with:
  • Still shots with rough models in position
  • Rough animation (models just sliding into place, some stick figure stand-ins, etc.)
  • Refined animation (perhaps several passes) with finer tuned models
  • Animation with the real models
  • Renderman shot, with simple, flat shading and lighting
  • Final shading and rendering.
These refinements are cut into the story reel as they're available.

Storyboard sketches are just a suggestion, and don't always correspond to the final film. For example, in the story sketches, "Ham" (the pink piggy bank) was depicted as a "stand up" character. When the model was finished, it was clearly a horizontal, standing-on-all-fours design, so changes had to be made from the storyboard sketches.

The Art Department

The Art Department designs everything appearing in the film. Door knobs, Andy's shoes, crumpled paper on floor, everything is designed for how they're made, how they look, and overall look of the final frame. Many pastel sketches are made to give overall guide to colors and lighting for each scene though they're not literal sketches of what actually appears in the final film. Everything appearing in the film was art directed - the goal was to create a stylization of reality, not duplicate of it.

Modeling

A detailed "model packet" is created for each character and model. It includes blue-print style plans for constructing the model, and sometimes detailed color paintings of the finished piece. The heads of the main characters were made from clay sculptures that were digitized with a Polhemus 3D wand. All modeling is done strictly in spline patches and NURBS, polygonal models won't work because a model may be viewed at any level of detail and the flat edges of a polygonal model can't be tolerated.

Traditional computer graphics models are geometry stored in a database, but at Pixar, models are procedural - that is, they exist as programs that are executed to generate the actual shapes. Some of the background and rigid pieces are done in a traditional modeling system (Alias), and these are incorporated into the "program". The procedural modeling system (custom developed by Pixar) is the basic environment that everything developed for the film goes into.

The model procedures include various "avars" (animation variables). These are the "puppet strings" the animators control to move and change the model. Mr. Potato head's mouth used 8 avars for the basic shape, some others to control its position on the characters face, etc. The "holes" in Mr. Potato Head's body could be turned on and off as needed with an avar. Avars can be high-level, e.g., controls for walking, bending over, etc. The controls are usually given appropriate names (e.g., "move elbow"). A complex character like Woody has over 700 avar controls.

For animation, stick figure stand-in models are used because they can be moved in real time on the animator's workstation. Later, the data recorded by the animator is "executed" again with the real model to generate the data for the film.

Animation

First, a layout crew assembles the models, positions them, adds the set dressing, and may do some basic animation, such as sliding characters into place. Then the actual animation process begins.

The ideal work flow would be for an animator to do all the characters in a series of shots for a scene, but this often didn't work out. Lasseter gave animators substantial freedom for working out the details of the action. Animators often made small thumbnail sketches to block out action, work out timing and dialog, check camera views, etc. Rough takes are done before refinement sets in. Lasseter constantly reviewed the animators' work in progress (a bad sign was silence from Lasseter followed by "OK...has anybody talked to you about this?").

One animator previously worked in clay animation, a medium that demanded linear start-to-finish work with no tweaking and adjustment along the way. He said the computer allows for much more refinement and control of his work.

For source material, the animators found some of the footage of Tom Hanks (Woody) reading his lines was a useful source of facial expressions. By contrast Don Rickles (Mr. Potato Head) had a great voice but little facial response while reading lines. To help visualize how the green toy soldiers would walk, an animator nailed a pair of sneakers to a board and figured out how to walk around in them. No motion capture or rotoscoping was used for the film.

Typical progression for animating a character is to slide the body into position; then add up & down motion as character walks, move the feet, set up eye movement. Face and expressions are done last. The general pattern is to work from the center of the character outwards.

Scenes were reviewed by the crew together in a big screening room, always on a large screen so the crew would have a better feel for how it would look in the theater. Comments were solicited during these screenings, and Lasseter was great at encouraging teamwork on the film.

Rendering

The color scheme was created to match the mood and setting of each scene in the film. Again very careful design was used. Painting and line art from Illustrator used to decorate surfaces. This included adding lots of small details like scuff marks, dirt, etc.

Procedural shaders (based on Renderman's programmable shading feature) were used everywhere in the film ("perhaps more than we should") to provide special controls. They often used geometry deforming shaders for special tasks. For example to get the character's feet to sink into the bed, a special shader was used that "talked" to the models. This was much easier than trying to create a special 3D model of the bed with animated footprints.

Much like real movie set design, many detail elements were painted "flats" with decorations created from other models. For example the outdoor trees were flat patches with tree designs painted on them, distant houses could be flats as well. Sometimes background matte paintings were used, much like live action work.

Lighting

"Lighting" was the generic name given to everything required for creating the final image - lights, shaders, colors, special effects (e.g., smoke), shadows, etc.

No radiosity techniques were used. Light that reflected off the floor was modeled as another light where the floor was. Many aspects of lighting and shading were faked to both save computation time and provide more control over result. A feature of Renderman is shadows are independent of light sources, so shadows can be moved to provide the best results.

Lights could also be set up to light some things in a scene but not others. For example, lights for the background could be turned off for the characters; and characters can carry their own lights around with them independent of the background to get best results. Like real films, lots of cheats were employed to get the right result. The philosophy is that photo realism is a measure of complexity, not an end unto itself.

Many little problems and "errors" did crop up, but if people didn't notice, they didn't bother to fix them. E.g., Mr. Potato Head's ears have a strange glow when they should have been in shadow, but it doesn't distract from overall scene. Lasseter had a great sense of what's important and what could be left as is (e.g., Bo Peep's hook doesn't quite follow her hand - did you notice?)

The Out Takes

The speakers showed a truly goofy out-takes reel of all sorts of computer screw-ups and animators having fun.
  • Buzz & Woody arguing, then hugging each other.
  • Buzz talking with his eyes sunk down at the bottom of his helmet.
  • Mom's head floating dis-embodied through the room
  • Mom's torso rotated 90 degrees and sticking out of the side of the car.
  • Wrong model parts stuck on the models; pieces missing, etc.

Q&A

Was motion capture used?
They investigated it, but decided against it. They wanted "human" characters to have a certain look. Motion capture requires multiple takes of a performer, getting performer to get it right, etc. The range of motion from animation is more to the style of the film.

How much computer power?
A big wall with rack after rack of Sun Sparkstations to was used to render the frames. The typical rendering time is 4 hours a frame, but this could stretch to 13 hours for some pathological cases. Four hours is a basic workable rendering time, using bigger computers would just allow higher scene complexity, but not speed production much. The biggest bottleneck for production is skilled human time, not computers.

How long did it take?
1.5 years of story development, green light, 2.5 years of animation.

Is computer animation better than other styles?
Computer animation is a style unto itself; not better than other forms, just different. It's unique in its ability to combine rich, motion blurred scenery with wild camera work.

[Back to Silicon Valley Siggraph home page]


This page created with Adobe SiteMill

Send comments to the webmaster