1. Advanced Computer Graphics (Fall 2010) CS 283, Lecture 24: Motion Capture Ravi Ramamoorthi http://inst.eecs.berkeley.edu/~cs283/fa10 Most slides courtesy James O’Brien from CS294-13 Fall 2009

2. Motion Capture  Record motion from physical actors  Use motion to animate virtual objects  Retarget and optimize as needed  General trend of data-driven appearance (BRDF)

3. Captures “Signature” of Actor

4. Types of Objects and Motions  Human, whole body  Portions of body  Facial animation  Animals  Puppets and cartoons  …

5. Capture Equipment

6. Types of capture equipment

7. Active Optical

8. Facial MoCap

9. Auto Calibration

10. Parameter Estimation

11. Manipulating Motion Data  WYSIWYG vs WYSIAYG  Basic Tasks  Adjusting  Blending  Transitioning  Retargeting  Building graphs

12. Nature of Motion Data

13. Adjusting

14. Adjustment  Define desired motion function in parts  Select adjustment function from nice space, such as C2 B-splines  Spread modification over reasonable time period  User selects support radius

15. Adjusting

16. Blending  Given two motions make a motion that combines qualities of both  Assume same DOFs  Assume same parameter mappings

17. Blending  Blending slow walk and fast walk

18. Time Warping  Define timewarp functions to align features

19. Blending in Time  Blend in normalized time  Blend playback rate

20. Blending and Contacts  Blending may still break features in original motion

21. Blending  Add explicit constraints to key points  Enforce with IK over time

22. Blending / Adjustments

23. Multivariate Blending  Extend blending to multivariate interpolation  Scattered data interpolation methods as needed

24. Transitions  Transition from one motion to another

25. Cyclification  Special case of transitioning  Both motions are the same  Need to modify beginning and end simultaneously

26. Motion Graphs  Hand built motion graphs often used in games  Significant amount of work required  Limited number of transitions by design  Motion graphs can also be built automatically

27. Motion Graphs  Similarity Metric  Measurement of how similar two frames of motion are  Based on joint angles or point positions  Must include some measure of velocity  Ideally independent of capture setup and skeleton  Capture a “large” database of motions  Compute similarity between all pairs of frames  Can be expensive, but preprocessing step  May be many good edges

28. Motion Graphs  Random Walks  Start in some part of the graph, randomly make transitions  Avoid dead ends  Useful for “idling” behaviors  Transitions  Use blending algorithm we discussed

29. Motion Graphs  Can have requirements  Start at particular location, End at particular  Pass through some points  Can be solved using dynamic programming  Efficiency may require approximate solution  Notion of goodness of a solution

30. Reordering

31. Content Tags

32. Integrating Physics  Simulation added to base motion  Inverse dynamics for matching  Oracle to assess results

33. Integrating Physics

35. Suggested Reading 1

36. Suggested Reading 2