Publication:

POSE2POSE2: POSE SELECTION AND TRANSFER FOR FULL-BODY CHARACTER ANIMATION

Loading...
Thumbnail Image

Files

NG-SEAN-WYN-THESIS.pdf (3.26 MB)

Date

2021-08-17

Journal Title

Journal ISSN

Volume Title

Publisher

Research Projects

Organizational Units

Journal Issue

Access Restrictions

Abstract

It is difficult for artists to accurately incorporate the persona of real-life human subjects into character animation, as many actors, speakers, and dancers have their own set of unique movements and gestures. In order to convey a subject’s personality through an animation, artists must determine representative poses of the subject. Artists must also determine optimal points of transition between poses in an animation. This process is often very manually intensive, so certain programs have tried to rectify this issue by automatically generating animations from video. However, this leaves less room for artistic freedom on the part of the animator.

Pose2Pose2 extends the Pose2Pose [22] animation tracking and transfer system designed to address the inefficiencies that come with manually creating an animation. Pose2Pose automatically extracted upper-body poses from a subject within a video, displaying them on a user interface for artists to choose from in order to draw representative poses. However, Pose2Pose did not have the functionality for tracking full-body or three-dimensional poses. Pose2Pose2 develops these functionalities, as well as additional user interface and animation features. Pose2Pose2 preserves artistic freedom while introducing a higher level of automation into the animation process.

Description

Type of resource

Princeton University Senior Theses

Keywords

Location

Citation