Physically Based Motion Transformation

Zoran Popovic, Andrew Witkin

Proceedings of the 26th Annual Conference on Computer Graphics and Interactive Techniques (SIGGRAPH)

Pages 11-20, 1999

Motivation

It is difficult to create a system in which an animator may automatically create character motion that is both realistic and controllable.

The goal is to use motion transformation instead of motion synthesis: that is, the authors will start with a real human motion sequence and transform this motion into a different kind of motion One example is transforming a real human running sequence into a running sequence with a limp that appears realistic.

Goal of This Paper

This paper presents a motion transformation algorithm for modifying character animation sequences that preserves essential physical properties of the motion. The algorithm consists of four main stages:

Character simplification: Create an abstract character model containing the minimal number of degrees of freedom necessary to capture the essence of the input motion. Map the input motion onto the simplified model.
Spacetime motion fitting: Find the spacetime optimization problem whose solution closely matches the simplified character motion.
Spacetime edit: Change spacetime motion parameters, introduce new pose constraints, change the character kinematics, objective function, etc.
Motion reconstruction: Remap the change in motion introduced by the spacetime edit onto the original motion to produce the final animation.

The first two stages require substantial human intervention, while the last two stages are fully automated and much faster than the first two stages.

Related Work

Forward dynamics methods work well for animating realistic motion of rigid objects [5] [4] [21] and secondary motion of cloth [12] [6] by accurately modeling the physics of each system. However, character motion is controlled by complex mechanisms through an intricate musculoskeletal structure, so forward dynamics cannot be easily used to generate controllable, realistic character animations. Spacetime constraints provide a useful framework however [33] [9] [20] [28] for generating realistic controllable character animations, but do not work well on very complex models and are extremely sensitive to the starting position of the character specified by the animator.

Robot controller design has also been applied to animation [26] [32] [31] [17]. After fine tuning, a wide range of animations can be generated [26] [19]. Motion can be generated from footsteps using some simple physical properties [30]. A controller transformation algorithm has been reported [18].

Motion capture editing techniques have been proposed [34] [8] [15] [14] [16] [27], but they ignore the dynamics of the motion. For large changes in motion, ignoring the dynamics creates undesirable unrealism in animations.

Biomechanists have studied similarities between vastly different locomotion models [7], indicating ways in which locomotion can be modeled with much fewer degrees of freedom than a full human model. The optimality of motion has been studied both in nature [1] [2] [24] and in sports [22].

Results

A variety of animations of human running and broad jumping were created using motion transformation from a real human running sequence and a real human broad jumping sequence. The results appeared realistic for variations of human running. An extra component was needed for the limp run animations to prevent the legs from colliding. A very simple hopper model was used to create the broad jump animations, which still produced results that appear realistic.

What This Paper Contributes

This paper is the first to solve the problem of editing captured motion that takes dynamics into consideration. A wide variety of character models can be handled by the algorithm described in this paper. A general way of controlling motion of complex models with simpler ones is part of the algorithm.

What This Paper Does Not Contribute

The methods of this paper work best for high-energy movement, but not for slower motions like picking up an object or getting up from a chair. Also, the motion fitting stage of the algorithm is largely manual. Model simplifications performed during this stage restrict the types of motions that can be generated, e.g. a waving gesture cannot be added to a human run sequence if the waving arm were simplified into a single rigid body.

No comparison of the results to experiment or physics-based simulations are made, so all judgments of results are based mainly on the overall appearance of each animation. The computed animations are not physically realistic in the sense that no dynamics computations are performed on the final animation.

Bibliography

1: R. M. Alexander.
Optimum walking techniques for quadrupeds and bipeds.
Journal of Zoology, London, 192:97-117, 1980.
2: R. M. Alexander.
Optimum take-off techniques for high and long jumps.
Philosophical Transactions of the Royal Society of London, 329:3-10, 1990.
3: K. N. An, B. M. Kwak, E. Y. Chao, and B. F. Morrey.
Determination of muscle and joint forces: A new technique to solve the indeterminate problem.
Journal of Biomechanical Engineering, 106:663-673, 1984.
4: D. Baraff.
Curved surfaces and coherence for non-penetrating rigid body simulation.
Computer Graphics (SIGGRAPH), 24(4):19-28, 1990.
5: D. Baraff.
Fast contact force computation for nonpenetrating rigid bodies.
Computer Graphics (SIGGRAPH), 28:23-34, 1994.
6: D. Baraff and A. Witkin.
Large steps in cloth simulation.
Computer Graphics (SIGGRAPH), pages 43-54, 1998.
7: R. Blickhan and R. J. Full.
Similarity in multilegged locomotion: Bouncing like a monopode.
Journal of Comparative Physiology A, 173:509-517, 1993.
8: A. Bruderlin and L. Williams.
Motion signal processing.
Computer Graphics (SIGGRAPH), pages 97-104, 1995.
9: M. F. Cohen.
Interactive spacetime control for animation.
Computer Graphics (SIGGRAPH), 26:293-302, 1992.
10: R. D. Crowninshield and R. A. Brand.
A physiologically based criterion of muscle force prediction in locomotion.
Journal of Biomechanics, 14(11):793-801, 1981.
11: P. de Leva.
Adjustments to zatsiorsky-seluyanov's segment inertia parameters.
Journal of Biomechanics, 29(9):1223-1230, 1996.
12: T. DeRose, M. Kass, and T. Truong.
Subdivision surfaces in character animation.
Computer Graphics (SIGGRAPH), pages 85-94, 1998.
13: P. E. Gill, M. A. Saunders, and W. Murray.
Snopt: An sqp algorithm for large-scale constrained optimization.
Technical Report NA 96-2, University of California, San Diego, 1996.
14: M. Gleicher.
Motion editing with spacetime constraints.
In M. Cohen and D. Zeltzer, editors, 1997 Symposium on Interactive 3D Graphics, pages 139-148. ACM SIGGRAPH, 1997.
15: M. Gleicher.
Retargetting motion to new characters.
In Proceedings of SIGGRAPH 1998, pages 33-42, 1998.
16: M. Gleicher and P. Litwinowicz.
Constraint-based motion adaptation.
Journal of Visualization and Computer Animation, 9(2):65-94, 1998.
17: J. Hodgins, W. Wooten, D. Brogan, and J. O'Brien.
Animating human athletics.
In Proceedings of SIGGRAPH 1995, pages 71-78, 1995.
18: J. K. Hodgins and N. S. Pollard.
Adapting simulated behaviours for new characters.
Computer Graphics (SIGGRAPH), pages 153-162, 1997.
19: J. K. Hodgins, P. K. Sweeney, and D. G. Lawrence.
Generating natural-looking motion for computer animation.
In Proceedings of Graphics Interface 92, pages 265-272, 1992.
20: Z. Liu, S. J. Gortler, and M. F. Cohen.
Hierarchical spacetime control.
Computer Graphics (SIGGRAPH), pages 35-42, 1994.
21: M. Moore and J. Wilhelms.
Collision detection and response for computer animation.
Computer Graphics (SIGGRAPH), 22(4):289-298, 1988.
22: M. G. Pandy, F. E. Zajac, E. Sim, and W. S. Levine.
An optimal control model of maximum-height human jumping.
Journal of Biomechanics, 23:1185-1198, 1990.
23: D. J. Pearsall, J. G. Reid, and R. Ross.
Inertial properties of the human trunk of males determined from magnetic resonance imaging.
Annals of Biomedical Engineering, 22:692-706, 1994.
24: A. Pedotti, V. V. Krishnan, and L. Stark.
Optimization of muscle-force sequencing in human locomotion.
Mathematical Biosciences, 38:57-76, 1978.
25: L. S. Pontryagin, V. G. Boltyanskii, R. V. Gamkrelidze, and E. F. Mishchenko.
The Mathematical Theory of Optimal Processes.
John Wiley and Sons, 1962.
26: M. H. Raibert and J. K. Hodgins.
Animation of dynamic legged locomotion.
Computer Graphics (SIGGRAPH), 25:349-358, 1991.
27: C. Rose, M. F. Cohen, and B. Bodenheimer.
Verbs and adverbs: Multidimensional motion interpolation.
IEEE Computer Graphics & Applications, 18(5):32-40, 1998.
28: C. Rose, B. Guenter, B. Bodenheimer, and M. Cohen.
Efficient generation of motion transitions using spacetime constraints.
Computer Graphics, pages 147-154, 1996.
29: A. Seireg and R. J. Arvikar.
The prediction of muscular load sharing and joint forces in the lower extremities during walking.
Journal of Biomechanics, 8:89-102, 1975.
30: M. van de Panne.
From footprints to animation.
Computer Graphics Forum, 16(4):211-224, 1997.
31: M. van de Panne and E. Fiume.
Sensor-actuator networks.
In Proceedings of SIGGRAPH 1993, pages 335-342, 1993.
32: M. van de Panne, R. Kim, and E. Fiume.
Virtual wind-up toys for animation.
In Proceedings of Graphics Interface 94, pages 208-215, 1994.
33: A. Witkin and M. Kass.
Spacetime constraints.
Computer Graphics (SIGGRAPH), 22:159-168, 1988.
34: A. Witkin and Z. Popovic.
Motion warping.
Computer Graphics (SIGGRAPH), pages 105-108, 1995.

ctj