Motion Retargetting and Transition in Articulated Figures
Date Issued
2005
Date
2005
Author(s)
Hsieh, Ming-Kai
DOI
en-US
Abstract
In movie or game industries, there is a need for creating motions of 3D characters all the times. Animated characters can make the scene more realistic and vivid. However, the process of animation creation is really time-consuming, and the resulting motion data is diffcult to reuse. With more and more motion libraries are available on the World Wide Web
or commercial companies, the problem of how to reuse motions become more important.
But we cannot transfer the motion data directly if two skeletons have different number of bones. Same problem can occur even we are dealing with same type of characters. Different motion capture companies may define different human skeletons; different artists tend to define different animal skeletons according to their experiences. If we can transfer motions between two different characters, or transit motions between different skeletons, we can shorten the animation creation time.
In our system, we develop an interactive interface for users to find the corresponding bones of two skeletons. This system first construct a meta-skeleton that contains both bone structures, then aligns the initial poses and local coordinates of these two skeletons. With the initial poses aligned, motion data can be transfer between them. In addition, by blending the similar pose of different skeletons, we can produce a seamless transition sequence.
The main contribution of this thesis is we present a framework that can not only transfer 3D skeletal animations between different skeletons, but also produce a smooth transition between motion clips of distinct characters. This method can be applied to animation editing software, or movie and game industries.
Subjects
三為動畫編輯
動作重複利用
動作轉移
動作串接
骨幹對應關係
3D animation editing
motion reuse
motion retargetting
motion transfer
motion transition
skeleton correspondence
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-94-R92922055-1.pdf
Size
23.31 KB
Format
Adobe PDF
Checksum
(MD5):632dc4ac4bd5a37d327a54831d757567