170 likes | 317 Views
Animating Human Athletes. By J.K. Hodgkins and W.L. Wooten Arjun Rihan CS 99K: Digital Actors. Introduction. People are skilled at perceiving the subtle details of human motion
E N D
Animating Human Athletes By J.K. Hodgkins and W.L. Wooten Arjun Rihan CS 99K: Digital Actors
Introduction • People are skilled at perceiving the subtle details of human motion • If synthesized human motion is to be compelling, it is necessary to create actors for computer animations and virtual environments that are realistic
Algorithmic Approach • Control algorithms that allow a rigid-body human model to run, vault and ride a bicycle in various environments • Built from a common toolbox: state machines, inverse kinematics, etc.
Advantages • Can be easily modified to compute similar but different motions • Greater interaction possible, especially in virtual environments • Possible to generate secondary or composite motion
Disadvantages • Difficult to design • Range of parameter variation allowed is relatively narrow • Workstation speed limits their use. For example, the simulations show earlier run from 6 to 20 times slower than real time in virtual environments
Compare the actor to the human diver Examples
Notice the how the actor mimics reality by trying to regain balance after the jump Example
Dynamic Behaviors • Each simulation contains: • - the equations of motion for model and environment • - control algorithms for balancing, running, etc. • - graphical user interface for simple top-level parameter control
Dynamic Behaviors (contd.) • In the case of a gymnast performing a vault, the control system obtains the following parameters and then computes the required forces and positioning of the limbs from control algorithms
Rigid links connected by rotary joints with varying degrees of freedom Examples of possible constraints on the degrees of freedom of the foot The Human Models
The distinct phases and corresponding changes in control actions are generated by a state machine This is a schematic of the state machine for the runner Transitions refer to the active leg Active leg and idle leg keep alternating Modeling Phases with State Machines
The state machine determines the control laws that are in effect for each phase of the vault Another State Machine - Gymnast
What This Means • The algorithms presented here enable an animator to generate motion for several dynamic behaviors
Evaluation • Through side-by-side comparison of video footage and animation, results appear to be good (What do you think?) • Comparison with biomechanical data is favorable
Evaluation (contd.) • A final form of evaluation would be a Turing test • Would involve direct comparison between simulated and human data on the same graphical model • Which motion would viewers prefer, given that that they wanted a more “natural” motion?
Next Steps • How can we make it easier to generate control algorithms for a new behavior? • By using a toolbox of general techniques to construct them as demonstrated here • How can we make the motion more natural? • Combine this approach with motion-capture and keyframing • Fine-tune the algorithms with human aesthetics