A Generative Model for Cyclic Motion Synthesis And Blending Using Probability Density Estimation

Abstract
The main focus of this paper is to present a method of reusing motion captured data by learning a generative model of motion. The model allows synthesis and blending of cyclic motion, whilst providing it with the style and realism present in the original data. This is achieved by projecting the data into a lower dimensional space and learning a multivariate probability distribution of the motion sequences. Functioning as a generative model, the probability density estimation is used to produce novel motions from the model and gradient based optimisation used to generate the final animation. Results show plausible motion generation and lifelike blends between dierent actions.

synth_blends3

[Image showing animation of blended walks using PDF. (a) Blend from a female walk to a female run. (b) Blend from a male walk to female skipping. (c) Blend from a female skipping to female run]

Authors
Dumebi Okwechime
Richard Bowden

Resources

OkwechimeBowdenAMDO08_pic
A Generative Model for Cyclic Motion Synthesis And Blending Using Probability Density Estimation
Okwechime, D. Bowden, R.
Fifth Conference on Articulated Motion and Deformable Objects pp218-277 (AMDO 2008).
[
pdf 752KB] [bibtex]




ch3
Demo (2:30 mins)
(Requires Quicktime to view)
[Video 4 MB]