Motion Blending - PowerPoint PPT Presentation

1 / 23
About This Presentation
Title:

Motion Blending

Description:

Real-time motion retargeting. Real-time Dynamic Control. Incremental time warping ... Walking, jogging, and running motions. Gyratory samples for each motion ... – PowerPoint PPT presentation

Number of Views:40
Avg rating:3.0/5.0
Slides: 24
Provided by: tcl2
Category:

less

Transcript and Presenter's Notes

Title: Motion Blending


1
Motion Blending
  • Hyun Joon Shin

2
Contents
  • Introduction
  • Related work
  • Basic ideas
  • Representation
  • Analysis
  • Motion selection
  • Time warping
  • Posture blending
  • Retargeting
  • Experiment
  • Conclusion
  • Future Work

3
Introduction
  • Motion Blending

Happy walking
Sad walking
4
Introduction
  • Motion Attributes
  • Qualitative
  • Happy ? Sad
  • Clueless ? Knowledgeable
  • Energetic ? Disheartened
  • ......
  • Quantitative
  • Speed
  • Degree of slope
  • Degree of gyration
  • ......

5
Related Work
  • Motion Interpolation
  • Verbs Adverbs Multidimensional Motion
    Interpolation Rose et. al. 1998
  • Everything by Example Cohen et. al. 2000

Happy
Clueless
Knowledgeable
Sad
6
Related Work
  • Multidimensional Motion Blending
  • Scattered data interpolation
  • Linear approximation
  • Radial basis
  • Time warping
  • Key time
  • Generic time
  • Motion transition
  • Verb graph
  • Sinusoidal function

7
Problems
  • Precise Motion
  • Qualitative attributes
  • Manual attribute specification
  • Euler angles representation
  • Inaccurate IK solving
  • Real-time Dynamic Control
  • Time warping problem

8
Basic Ideas
  • Precise Qualitative Control
  • Motion analysis
  • Layered representation
  • Quaternion interpolation
  • Real-time motion retargeting
  • Real-time Dynamic Control
  • Incremental time warping
  • Real-time motion retargeting

9
Overview
attributes
Motion Selection
Target Character
weights
Motion Analysis
Motion Retargeting
Posture Interpolation
Motion Samples
Target Motion
Attributes Layered Representation
10
Currently, Problem is
  • Motion Samples
  • Walking, jogging, and running motions
  • Gyratory samples for each motion
  • Almost regular distribution
  • Attributes
  • Speed of walking
  • Degree of gyration
  • ......

11
Motion Representation
  • Layered Representation
  • 1st layer abstracted body trajectory
  • Speed
  • Degree of gyration
  • Key-times

tk
t1
tn
12
Motion Representation
  • 2nd layer relative to abstracted trajectory
  • Body posture
  • Footholds
  • 3rd layer relative to foothold
  • Foot posture

13
Motion Analysis
  • Assumption
  • Circular trajectory
  • Constant velocity
  • The same number of steps
  • Speed
  • arc length/time
  • Degree of Gyration
  • gyration/time
  • Key-times Hill-Strike Toe-Off
  • Manual specification

?
l
14
Motion Selection
  • Four Neighbors
  • Weight Based on Bilinear Interpolation

speed
right turn
left turn
15
Time Warping
  • Generic time Rose et. al. 1998
  • Actual time interpolation

Backward !!
W0.5
W0.5
tk
t1
W0.5 ?1
16
Time Warping
  • Incremental Time Warping
  • Blending derivatives
  • Finite integration

W0.5
W0.5
tk
t1
W0.5 ?1
17
Posture Blending
  • 1st Layer
  • Exact computation
  • 2nd Layer
  • Posture blending
  • quaternion blending
  • Foothold blending
  • hill-strike position
  • preserve until toe-off
  • 3rd Layer
  • Quaternion blending

18
Quaternion Interpolation
  • Bilinear Interpolation

Spherical Linear Interpolation
19
Motion Retargeting
  • On-line Motion Retargeting
  • Importance-based approach Shin et. al

20
Spatial Constraints
  • Foothold interpolation
  • Ankle position
  • Fixed position during stance
  • Ankle displacement
  • Displacement during stance
  • 3rd layer of motion
  • Importance
  • Distance between
  • Ankle position Floor

21
Experiment
22
Conclusion
  • Real-time Motion Interpolation
  • Qualitative attribute
  • Speed gyration
  • Dynamic attribute control
  • Application
  • Game
  • Sport replay
  • Animation prototyping

23
Future Work
  • Motion Set Attributes
  • Non-regular samples
  • More quantitative Qualitative attributes
  • High dimensionality
  • Motion Analysis
  • Multidimensional interpolation extrapolation !!
Write a Comment
User Comments (0)
About PowerShow.com