Today’s digital movie characters are more realistic, interactive, and endearing than ever before. But creating them is still expensive and labor-intensive. Most are either hand-drawn frame by frame with digital pen and ink (and computers that fill in some gaps), or they are based on “motion capture” techniques, which record and digitally mimic the movements of live actors to create on-screen characters. But as movies and video games pack more and more digital creatures onto the screen, animators are turning to physics and new artificial-intelligence methods for faster, more efficient ways to bring their stories to life.
Physics equations guide NaturalMotion’s characters. (Image courtesy of NaturalMotion)
NaturalMotion, a spinoff of the University of Oxford in the United Kingdom, has developed artificial-intelligence software that gives digital characters the power to animate themselves. The approach: a programmer specifies a character’s physical shape and properties and adds equations to govern the movement of its body parts. When animators apply a simulated force like gravity or a push from behind, the character responds realistically without further programming. “What you see on the screen is not a computer graphic of the character” dumbly mimicking recorded motions, says NaturalMotion’s chief executive, Torsten Reil. “It is the actual character.” The company’s goal is to generate interactive animations in real time, allowing animators to use the technology in video games as well as films, Reil says. At the Computer Graphics Laboratory of Stanford University, researchers Katherine Pullen and Christoph Bregler have combined automation techniques with the traditional motion-capture approach. Using their system, animators can manually draw the most active parts of a character-its legs, say, during a walking sequence-and use archival motion-capture data to automatically complete the body and give it extra texture. “Within five years, all high-end games will use methods like these,” says Casey Muratori, lead developer at RAD Game Tools in Kirkland, WA. Making characters not only move but also emote convincingly requires a different level of technology-software that generates “believable facial expressions and body language,” says Ken Perlin, director of the Media Research Laboratory at New York University. Perlin’s group is developing tools that will make it easy for animators to add layers of subtle gestures, for instance, a raised eyebrow or drooping shoulders. Such effects must be drawn by hand today. Though farther from commercialization, Perlin’s software is a step closer to animation’s larger goal-to dazzle our hearts as well as our eyes.
These weird virtual creatures evolve their bodies to solve problems
They show how intelligence and body plans are closely linked—and could unlock AI for robots.
A horrifying new AI app swaps women into porn videos with a click
Deepfake researchers have long feared the day this would arrive.
Chinese hackers disguised themselves as Iran to target Israel
But they left a few clues that gave them away.
DeepMind says it will release the structure of every protein known to science
The company has already used its protein-folding AI, AlphaFold, to generate structures for the human proteome, as well as yeast, fruit flies, mice, and more.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.