Dykstra, Imageworks key software decision overcomes obstacles
You’ve probably seen the just-add-water demonstrations: a quick shot of a wire frame on a computer screen, then 10 seconds of digital ink and paint and there you go, a 3-D digital character is complete.
At least that’s the “Access Hollywood”-“Entertainment Tonight”- version of the kind of CG animation on display in 2002 box office champ “Spider-Man.”
The fact is creating photo-real feature-length computer animation that can achieve verisimilitude and pass scrutiny on a 50-foot-wide cinema screen is labor-intensive work.
“There’s nothing we can’t do, even if we have to proceed pixel by pixel,” says multi-Oscar’d veteran f/x man John Dykstra. “But it’s often time-consuming, hard work.”
Dykstra should know what he’s talking about. Since making a splash in 1971 as a 22-year-old wunderkind on the sci-fi cult classic “Silent Running,” Dykstra has amassed an impressive resume: the original “Star Wars,” the first “Star Trek” chapter, “Contact,” both “Stuart Little” films, “Spider-Man” and its sequel, among others.
One of the first “Spider-Man” sequences to get the full attention of Dykstra and the CG boffins at Sony Pictures Imageworks a couple of years ago was wall crawl. In that sequence, the soon-to-be no-longer-geeky Peter Parker tries out his newfound powers for the first time climbing up a multistory brick wall.
“We filmed a test run with a stunt man in an outfit rigged with a safety harness doing the stunt,” recalls visual f/x supervisor Scott Stokdyk. But when they digitally removed the wires and played the footage back: “It just looked like a stunt man crawling up a wall.”
No magic there. So it was back to the digital drawing board. Though they weren’t sure how they were going to do it, director Sam Raimi and his f/x team decided to make the wall-climbing Spider-Man completely digital.
And therein lay the rub (or at least the volume of work). In as simple an act as walking across a room, a human being goes through thousands of moment-by-moment, minute physical adjustments.
Re-creating that digitally is hard enough. Add to that the person is scaling a multistory building clad in a skintight outfit (where almost every muscle movement is visible) and one begins to understand the sheer amount of information needed to create a digital version of Tobey Maguire’s Lycra-clad form.
At the subatomic level of the work done on the wall crawl (and other sequences), was a basic decision between using software either “nurb-” or polygon-based.
A MIT-sounding choice to be sure, but an important one. In brief, nurb-based software is good at creating curved shapes and surfaces such as musculature. It’s not so good at creating sharp edges. Polygonal-based software is good with the edges but not so good with surfaces. Animating cloth, for example, where there are edges and surfaces, requires using bits of both.
At a third level, CGI artists can use software that uses data known as subdivision surfaces (pioneered by Pixar) — essentially enhanced polygonal software with nurblike capabilities.
“They’re like different mediums for sculpture,” says Stokdyk. “Some people prefer to work in clay, others in stone or metal.”
For its part, SPI tends to model and animate using nurb-based programs.
As Spider-Man advances up the wall his body geometry is constantly changing and his costume (which at this point in the film is homemade and not so form-fitting) creases and flexes.
“What we use in any sequence is determined by the object we are animating. It’s at this nexus of decision-making that the character comes to life,” Stokdyk continues. “It’s where the art of CGI and the vision of the director enters into the process.”