CAREER: Reusable, Realistic Motion Libraries for Computer Animation

University Of Washington, Seattle WA

Investigators

Abstract

Animation is a tremendously powerful medium of expression. Ideas that would be complex or even impossible to express with still images or words can easily be conveyed using animated sequences. Unfortunately, even when the concept to be conveyed via animation is simple, the process of producing the animation is inordinately difficult. Indeed, this is one of the reasons computer animation has become so dominant - it holds the promise of automating, and thus significantly simplifying, the production process. Sadly, the process of transferring ideas into computer animations is still far from simple. For example, the animated feature film Toy Story 2 required extremely hard work by as many as 200 highly skilled computer and animation experts for more than three years. My career goal is to develop the concepts and algorithms that will enable effortless, rapid development of realistic animations that effectively convey the creator's purpose. This research hopes to create the theoretical and experimental foundation for building much more powerful animation tools than exist today. Such tools will sharply reduce the production time needed by skilled animators and, more importantly, enable average computer users to express their ideas through animation. Such tools will also help educators teach more effectively and allow each one of us to become creators and directors of personal and fictional stories. Only when animation production is easier will animation be as commonplace on the web as images and text are today. Moreover, intuitive methods for creating realistic human motion will further enable the creation of human avatars in tele-presence applications, new anthropomorphic human-computer interfaces, and realistic digital actors in feature films and video games. Computer animation will not achieve its full potential until 1) animation tools require little or no skill, 2) an animator's creativity is not stifled by the limitations of the animation tools, and 3) the animation systems are available to every person with a computer and a story to tell. I hope to significantly contribute towards these goals. At the heart of this research is the observation that the animator's creativity will not be fully realized until it becomes easier to produce realistic motion sequences. This problem is tremendously challenging because the underlying physical models of motion are difficult to create and control. I believe that the solution lies in the creation of highly-flexible realistic motion libraries, together with tools to modify them intuitively and extensively. The power of libraries stems from the ability to reuse already existing high-quality motion, reducing the need for animation skills. In this paradigm, the process of animation turns into selecting the specific motion library, and modifying a set of motion properties that transform the original motion into a final animation. In my research, I propose to develop a methodology for creating and using realistic motion libraries flexible enough to be used in a wide range of applications. In 1995, I proposed the use of motion transformation as new way to create animations - a method fundamentally different from the traditional way of creating animations from scratch. This transformation approach is particularly useful for modifying realistic motion data captured from real-world actors. Unfortunately, during the transformation process, much of the realism tends to be lost. Recently, I published a novel transformation approach which demonstrates that animations can be intuitively transformed into a wide range of new sequences without violating the fundamental dynamic properties of motion. The realism is preserved by maintaining a model of the dynamic and biomechanic properties of the animated character. In the future, I will further develop mathematical models that can be combined with real-world data to create reusable motion libraries. Aside from preserving the realism of motion, the most important requirement for effective motion libraries is the flexibility with which the libraries can be adjusted to meet the needs of an animator. I plan to achieve this flexibility by decomposing a character's motion into a fundamental component and a style component in a way that allows us to independently transfer these components to new characters. For example, the animator can produce a child's cheerful run sequence by starting with a running motion library extracted from the captured human run. This motion can be modified by applying a happy, exuberant motion style, which is then transferred to an animated character of a small child. Such decoupling of the fundamental motion, style, and performing character provides an extremely powerful and flexible animation paradigm that can produce a wide range of animations from a very small motion dataset. The main goal of the educational component of this proposal is to demonstrate the utility of advanced animation tools in education and story-telling. I am particularly interested in introducing these new animation tools to Seattle high-school students to verify that our research does indeed empower every story-teller with an accessible medium of expression. I am also planning a project-oriented animation course intended for undergraduate and graduate students involved in animation research, as well as in art and music. Together, the students will work on using our tools to create an artistic short film project. This synergistic effort will expose the artists to the latest animation technology and provide direct user feedback to students involved in research. In addition, the created motion libraries as well as the animation tools, will be freely distributed through the web and CDs with the hope that they will help animation become an effective communication medium across educational, economic and cultural boundaries.