From 2004's hits to 'King Kong,' CG turns to academics
Having perfected the right look for skin in the last tentpole he worked on, Joe Letteri is now concentrating on hair.
No makeup artist, Letteri is a visual-effects supervisor for Weta Digital — and also part scientist and part computer programmer.
It’s these latter unofficial titles that come in handy when trying to rise to CG’s ever-climbing bar of challenges, as digital f/x and animation pros work on the leading edge of academic research and program their own software tools to solve CG’s toughest problems.
Back in 2001, for example, Letteri and a number of his colleagues were particularly inspired by a white paper presented at Siggraph by UC San Diego professor Henrik Wann Jensen. It suggested that all non-metallic materials, notably skin, are translucent, and light scatters inside them before getting absorbed or leaving in another location.
Previously, skin had been dealt with in CG primarily in a two-dimensional context. But Letteri and his team incorporated Jensen’s ideas on “subsurface scattering” for the creation of the creepy Gollum in the second and third installments of “The Lord of the Rings,” and that resulted in perhaps the most life-like CG character ever rendered — and contributed to “LOTR’s” last two f/x Oscars.
Weta is hardly alone in its use of subsurface scattering. Almost everyone involved with digital f/x and animation employs the technique these days, including Industrial Light & Magic, which used it to create the Dobby character in the “Harry Potter” movies, and Digital Domain, which used it for “I, Robot’s” killer androids.
“We use subsurface on every character,” says Janet Healy, a producer on DreamWorks’ upcoming animated feature “Shark Tale.” “It’s becoming the norm.”
Still, improvements in the way CG treats its skin are overshadowed by an entire body of innovation yet to come.
“Since we’re doing ‘King Kong’ (for Universal and helmer Peter Jackson), obviously, we’re preoccupied with hair,” Letteri says. “With skin, you have an organized surface. But with hair, every time it brushes up against something, or gets muddy or wet, it looks entirely different.”
Letteri and his team are about midway through Kong’s developmental stage, mish-mashing off-the-shelf software programs themselves to create custom tools — a process that consumes significant time and resources at nearly every big, envelope-pushing CG shop.
Letteri says his team’s not through monkeying around with Kong’s hair, but he’s confident that, in about four months, it will have all the big ape’s parameters worked out. At that point, Letteri’s crew will concentrate on Kong’s acting performance, and hopefully not have to worry about rendering times and other technical mumbo-jumbo.
While good hair days will certainly advance the art of character rendering, water continues to vex digital f/x and animation pros. And so far, there’s been no unifying theory, similar to subsurface scattering, with every house handling this challenging, ever-changing element differently.
“Over the last two years, everybody — from (those on) animated films like ‘Finding Nemo’ to live-action films like ‘Master & Commander’ — has been working on fluid simulation, and there are a myriad of ways to go about it,” says Digital Domain’s David Prescott, who supervised f/x on Fox’s “The Day After Tomorrow.”
Regardless of the approach, Prescott says digital water has come a long way since “Titanic” — yet another film he’s worked on — became one of the first to dip its toes into it.
Complex fluid simulation models are now used to help create scenes such as those in “Day After Tomorrow,” one set of which features a tidal wave swallowing Manhattan.
“Water has a terminal velocity, and it can’t move beyond a certain speed without turning into gas. If we do shots, and try to make it move too fast, it will just look wrong. And somewhere in the back of the (audience’s) mind, they’ll know it looks fake, too.”
Consultations with doctorates — mathematicians, physicists and the like — are essential,
Prescott says. “We’re trying to be the scientist and the artist at the same time.”
“We have a lot of smart people here,” adds Sony Pictures Imageworks prexy Tim Sarnoff, “but we hire a lot of consultants.” On the company’s groundbreaking “Stuart Little,” for example, Imageworks employed an animal expert to ensure correct mouse movements; on “Hollow Man,” a physiologist was brought in to show how the human body is put together; and on “Castaway,” an oceanographer was hired to oversee the waves.
While outside experts ponder the water issue, another CG conundrum just beginning to emerge is how to go about creating legitimate computer models of fire.
“I don’t think any natural phenomena will ever be completely conquered (in CG),” Prescott says. Still, he adds that improvements in “volumetrics” have enabled digital f/x and animation artists to get reasonably close to mastering such amorphous entities as smoke and clouds.
Ultimately, Prescott says, the rendering of all things — from characters to natural elements — will be vastly improved when CG pros understand light a little better.
“We have to learn to write software that emulates the exact lighting components of specific types of light,” he explains. “For example, we say, ‘You are a tungsten light,’ and the program will give off that kind of light. We’ve made huge steps in the last four years, but we need to make the next jump.”
Lighting is still a “trial and error” process,” agrees Cliff Plumer, chief technology officer for ILM, with CG pros introducing characters into frames, then testing various lighting parameters, then waiting laboriously for the computer to render it all.
A more efficient, automated approach to light is becoming increasingly important, CG pros say, as more blue-screen is used, and entire environments that are computer generated, as well as “digital doubles” for actors, are becoming as ubiquitous.
Of course, some Ph.D. could present a breakthrough approach to lighting at Siggraph this year. But Plumer says that wouldn’t make his job any easier.
“Just when we reach a milestone to do one thing,” he notes, “some director will come along and raise the bar.”