1. This paper presents a method for generating whole-body skeletal animations of non-humanoid characters from human motion capture data. (Intro, paragraph 1)
2. Often, characters that are inspired by animals or inanimate objects are expected to be anthropomorphic, but most data-driven techniques for animation, such as keyframing, do not work for these characters due to large skeletal and motion style differences. (Intro, paragraph 2)
3. Key idea: First, capture motions of a human subject acting in the style of the target character. A few selected key poses are then used to build a statistical model for mapping a human pose to a character pose.
4a. Every frame of the human motion capture can be mapped to the character model using the mapping function, with fine adjustments for physical realism and contact constraints; also significantly reduces the time and cost compared to keyframing. Demonstrated for mapping a jumping motion to a lamp, and dancing motions to a penguin and squirrel.
4b. Can be used to provide an alternate way of animating characters aside from 3D animation software, such as Maya.
5. Author’s web page contains no videos or links unfortunately.