Massive Software's AI-driven animation system was utilized by Animal Logic for Happy Feet, which recently won the Academy Award for best animated feature. Using Massive, Animal Logic filled the screen with enough happy-footed "extras" to populate several Busby Berkeley showstoppers. The Warner Bros. Pictures feature represented the first 3D-animated project by both director George Miller and the Australia-based Animal Logic.
Animal Logic acquired Massive to complete hundreds of HAPPY FEET shots that worked seamlessly with the Animal Logic animation pipeline, which included proprietary crowd software Horde. Development of Massive agents focused on the production of two agent types: the adult and toddler versions of the Emperor Penguin, in which social behaviors alone comprised more than half of the adult agent brain.
Though the number of agent types was low compared to other film projects, performance requirements were set very high, due in part to the environment and terrain demands. "With the sunny lighting of the environment, and the black feet on ice, it was obvious at the beginning that our terrain adaptation would need to be spot on, "commented Tim Rowlandson, crowd supervisor, Animal Logic. "We had to create a system that could handle pitches up to 60 degrees with animation that wasn't designed for this extreme angle. Kerry Simpson, Massive td, worked on putting together a very creative system within Massive that used limited animation data to hit the 60-degree target."
Massive also allowed Animal Logic to create stable performance simulations throughout 1,000-plus frames. "Collision avoidance was also a big concern for us. Just like the terrain adaptation, we had a similar problem with the characters' strong black-and-white coloring highlighting interpenetration," Rowlandson added. "Using some interesting features in Massive, we were able to come up with a dynamic avoidance system that scaled based on proximity. This gave our characters a very accurate avoidance system when tightly packed together or spread apart and moving at a quick pace. In one test, an agent did various side step/walk blends in order to get through a very tight crack in a huddle so that he could leave the group. The director was blown away."
Miller wanted to ensure that the Massive crowds weren't just filling in the background, but were able to deliver complete character performances. "One example is the work we put into commuter lines. On a two-way commuter line our agents could receive a trigger to chat based on a probability variable," Rowlandson reported. "Depending on how strong this urge to chat was, some chatter agents would just say 'hi' as they walked by each other, while others would stop for a bit and have a jaw waggle before continuing on their way to avoid causing major pile-ups, while non-chatting commuters would adjust their collision avoidance allowing them to see far enough ahead." The agents made these adjustments dynamically and could join the commuter line at any time during the simulation.
"We had penguin agents that would sing as they walked along, and toddlers that would play and run around in groups," Rowlandson continued. "There were agents that would get bored and fall asleep, agents that would huddle together to keep warm, or that would break away from a huddle to start a flocking group. These behaviors, plus many more, were applied onto only two main agents. We directed them using probability variables and color maps. When placed in a scene and with minimal user direction, the results were amazing."
One standout Massive shot in the film shows a cluster of teen penguins jumping off a cliff into the ocean. The original brief was to have the penguins pushing at the edge of the cliff, then a few would jump setting off a cascade of penguins building as the shot progressed. Rowlandson set up a virtual barrier that stopped the penguins at the edge of the cliff. Using a flow field directed toward the cliff edge kept the motion of the crowd alive and constantly pushing forward. Color maps were used to alter the agent's behaviors so that the ones on or close to the edge of the cliff would behave differently than those at the back. After that, a moving jump trigger was set up to replace the barrier, moving each penguin agent through to the water as they moved closer toward the edge.
Academy Award-winning Massive is the innovative 3D animation system that incorporates procedural animation and artificial intelligence. Massive is used by animation and visual effects artists to explore the new world of creative opportunities that AI-enabled characters make possible.
Agents are 3D characters that have a fuzzy logic AI "brain" and the natural senses of sight, sound and touch, which enables them to interpret and react autonomously to the world around them. Agents can be just about any character you can imagine from humanoids, birds, animals and insects to inanimate objects such as cars.
The intuitive interface of Massive allows artists to interactively control and direct agents providing highly realistic, directable and emotive performances.
New Zealand-based Massive Software (www.massivesoftware.com) is the leading creator of artificial intelligence-based 3D animation systems. Massive was founded when Stephen Regelous programmed a unique piece of software for director Peter Jackson to make creation of complicated visual effects scenes involving hundreds of thousands of digital characters a practical reality. Regelous garnered a Scientific and Engineering Award from the Academy of Motion Picture Arts and Sciences in 2004.
One of the world's most highly respected digital production companies, Animal Logic (www.animallogic.com) produces award-winning design, animation and visual effects solutions for the film, television and advertising industries. The company has a studio in Sydney, Australia, and a production office in Los Angeles.
Bill Desowitz, former editor of VFXWorld, is currently the Crafts Editor of IndieWire.
