Reality based virtual humans and animals

人类和动物运动的精确建模和动画是困难的，因为它需要解剖学，生物力学和神经控制的知识。 动作捕捉已经成为流行的计算机动画的人类和动物，因为它避免了这个困难，直接捕捉运动的微妙之处。 然而，修改运动捕捉以满足新动画的约束可能很困难，因为运动捕捉并不能揭示它是如何制作的。 在这项研究计划中，我们打算通过结合基于物理的模拟，生物力学模型，神经控制的近似，以及运动和接触力的捕获来改善虚拟人类和动物的动画。 现有的方法估计被动行为的捕获科目将得到验证，估计控制刚度变化的新方法将进行调查。 我们将探索自愿的运动，如达到，抓，和有节奏的运动，如抓挠和运动。

Accurate modeling and animation of human and animal motion is difficult because it requires knowledge of anatomy, biomechanics, and neural control. Motion capture has become popular for computer animation of humans and animals because it avoids this difficulty by directly capturing the subtleties of movement. However, it can be difficult to modify motion capture to satisfy the constraints of a new animation because motion capture does not reveal how it was produced. In this research program we intend to improve the animation of virtual humans and animals through a combination of physically based simulation, biomechanical models, approximations of neural control, and the capture of motion and contact forces. Existing methods of estimating the passive behaviour of capture subjects will be validated, and new methods of estimating controlled stiffness changes will be investigated. We will explore voluntary motions such as reaching, grasping, and rhythmic motions such as scratching and locomotion.