Filmmakers have strip-mined the story for decades, and yet Walt Disney’s animated Tarzan — the swinger’s 48th trip to the big screen — is astounding audiences as well as studio bean counters. Nearing the $80 million box office mark in its first 10 days (closing in on the $99 million total domestic gross of summer 1997’s Hercules), Tarzan has also made history by virtue of its thoroughly convincing junglescapes. Live-action directors take this kind of scenery for granted, but for a largely handcrafted animated feature, whipping up a vision of tropical paradise requires exhaustive tending.
The studio might have had to settle for an underfoliated look if not for a technical breakthrough called Deep Canvas. Engineered specifically for Tarzan by a brain trust of Disney art directors and computer programmers (the studio holds the patent), it’s a computer-based tool for enhancing the apparent depth of backgrounds.
Tech talk make reader scratch head? Relax — the basics are as easy as See Jane Run. Just watch the screen. From a breathless roller-coaster-style sequence in which the loincloth-clad hero, surfing along the passing boughs like an extreme-sports skateboarder, rescues Jane from a pack of baboons, to a closing shot where the couple sails through a perfectly three-dimensional tree canopy, the movie’s Deep Canvas sequences — which constitute only about 10 minutes of the 88-minute running time — provide an instantly recognizable visual jolt. ”This finally lets us do what live-action can do with a Steadicam,” says Kevin Lima, who codirected the film with Chris Buck. ”For the first time, we can move our camera in and around a background, instead of just over flat layers of two-dimensional backdrops.”
How’d the Disney folks do that? Essentially, by animating the backgrounds before they animated the characters. For each Deep Canvas shot, the directors first decided on their camera move. A cadre of technician-artists would then create a moving background environment made up of simplified, largely cylindrical computer-generated shapes, with tree trunks and branches represented by simple grayish geometrical abstractions.
Once this three-dimensional setting was ready, the scene went to character animators, who drew in the figures in shifting perspectives by hand in each frame. Simultaneously, it went to background painters, who’d nurture those bland gray undulations into convincing organic objects. Except that instead of using physical palettes and paintbrushes, they painted digitally, using a stylus, a digitizing tablet, and software that simulates many styles of brush strokes and paint mediums.
”Deep Canvas is something like a player piano,” says Eric Daniels, who spent several years creating the technology under art director Daniel St. Pierre, with help from programmers George Katanics and Tasso Lappas. ”Except it goes further and learns how the painter paints. Then it replicates one or two paintings, or fragments of paintings, into dozens or hundreds of paintings, all from slightly different angles. It can do in hours what would take a painter months to execute by hand.”