Interactive control of avatars animated with human motion data

1
Interactive control of avatars animated with
human motion data

• Jehee Lee
• Jinxiang Chai
• Paul S. A. Reitsma
• Jessica K. Houdgins
• Nancy S. Pollard

Presented by YoungSang
2
Why Motion Capture?

• Keyframe animation
• Artist-driven
• Kinematics
• Difficult to create a realistic animation
• Physically-based animation
• Dynamics
• Not reflect human behavior
• Not real time

3
Motion capture

• Capture real human movement
• Realistic
• Efficiency
• But, not controllable not flexible

4
Motion representation

• Tree structure
• Motion clip
• Xx1,x2,,xn
• At a frame i
• xipi,0,qi,0,qi,1,,qi,m

2
3
1
1
0
5
Motion database

• In video games
• Many short, carefully planned, labeled motion
  clips
• Manual processing

6
Walk Cycle
Stop
Start
Left Turn
Right Turn
7
Motion database

• This approach
• Extended, unlabeled sequences
• Automatic processing
• Connectivity
• Search
• Real-time, controllability

8
Motion data acquisition
9
Maze sketch interface
10
Re-sequence
Motion capture region
Virtual environment
Sketched path
Obstacles
11
Re-sequence
Motion capture region
Virtual environment
12
Overview
13
Overview
14
Unstructured input data

• A number of motion clips
• Each clip contains many frames
• Each frame represents a pose

15
Unstructured input data

• Connecting transition
• Between similar frames

16
Graph construction
17
Distance between frames

• Possibility frame i -gt frame j
• Distance

18
Pruning transitions

• Reduce storage space
• O(n2)
• Better quality
• Pruning bad transitions
• Efficient search
• Sparse graph

19
Pruning transition

• Contact state
• Avoid transition to dissimilar contact state
• Threshold
• User-defined threshold

20
Pruning transition

• Local maxima

j
i
Probability Pij
21
Pruning transition

• Avoid dead-ends strongly connected components
• Every node is connected to each other directly or
  indirectly

22
Overview
23
Clustering
A
C
A
B
C
A
D
E
D
E
24
Clustering
A
C
A
B
C
A
D
E
D
E
25
Transitions
A
C
A
B
C
A
D
E
D
E
26
Cluster tree

• Three possible actions ABA, ABC, ABD

A
A
B
C
D
A
C
A
B
C
A
D
E
D
E
27
Cluster forest
B
A
A
C
A
B
C
C
C
E
D
A
A
C
B
C
A
D
E
D
E
A
D
B
D
28
Cluster paths and most probable motions

• Given a cluster path p anda
  first frame of motion sequence
• Calculate motion sequence

Find motion sequence among all possible motion
sequence by maximizing
29
Avatar control

• Choice
• Cluster tree
• Restriction

30
Avatar control

• Path sketching
• Find a cluster path

31
Avatar control

• Vision interface

32
search
3 sec
Video buffer
A
A
C
B
C
A
D
E
D
E
33
search
3 sec
A
A
B
D
A
A
C
B
C
A
D
E
D
E
34
Conclusion

• Motion graph
• Double layered graph
• Various user interface

35
Appendix I

• Distance

36
Appendix II

• Clustering (EM algorithm)

37
Appendix III

• Number of cluster