CPSC : Data-driven Character Animation Jinxiang Chai
Data-driven Character Animation Given motion capture data, how to create desired animation Motion capture Animation Control
Applications Entertainment (video games, movies, broadcast) Virtual environments Trainings Human-computer interactions Education Etc.
Applications: Entertainment Performance-based facial animation for home use Tiger Woods PGA Tour 2005 Mike Tyson Heavyweight Boxing Friday Night 3D Bowling Xbox Outlaw Tennis ShrekThe Lord of the Rings Polar ExpressFinal Fantasy
Applications: Virtual Environments Performance-based facial animation for home use Virtual Teleconferencing (from BT) Multi-user Virtual Worlds Virtual Poker Room
Applications: Training Performance-based facial animation for home use Military Training (BDI) Tai Chi Training (CMU)
Applications: Human-computer Interactions Performance-based facial animation for home use Tessa (Text->British Sign Language)
Applications: Education Performance-based facial animation for home use Fish animation for zoology class
Robotics and control Computer vision AI Animal science Biomechanics, statistical learning, applied math etc. Interdisciplinary area Performance-based facial animation for home use
Short Bio New CS faculty member at Texas A&M Ph. D thesis in computer animation Undergraduate degree in EE
Research Interested in animation, graphics, and vision Methods for creating and manipulating high-dimensional visual media (animation, models, images, and videos) Data-driven approach Video-based data capture Thesis: exploiting spatial-temporal constraints for interactive animation control
Thesis Research Goal: everyone can generate and control human animation easily and quickly Online animation control
Thesis Research Goal: everyone can generate and control human animation easily and quickly Online animation control
Thesis Research Goal: everyone can generate and control human animation easily and quickly Offline animation control User input Output animation
Thesis Research Goal: everyone can generate and control human animation easily and quickly Offline animation control User input Output animation
Why Do I Teach This Course? Provide an in-depth study of character animation techniques with an emphasis on data-driven approach Teach you how to find and formulate research problem Refine your presentation skill Inspire some of you to do research with me
Prerequisites A good working knowledge of C/C++ or Matlab A good understand of linear algebra Willing to learn (optimization, statistical learning, etc.)
Grading Schemes Paper presentation (20%) Class participation/discussion (20%) Homework (20%) Final project (40%)
Paper Presentation (20%) Choose a paper from the list Read/understand the paper well Paper presentation Come to my office hours if u need help
Class Participation/Discussion (20%) Everybody reads Participate in paper discussion Come to my office hours if you still have any questions
Homework (20%) Homework 1 (10%): Key-frame interpolation and forward kinematics Homework 2 (10%): Inverse kinematics or mocap Students work individually Late policy: 20% reduction per day if you do not have good reasons
Final Project (40%) Implement a project approved by the professor Student can work in a group of two Get extra 20 points if you do an excellent job Talk to me if you need any helps
Grading Schemes Paper presentation (20%) Class participation/discussion (20%) Homework (20%) Final project (40%) There are no exams!!
Class Overview: Background Introduction Motion capture and motion capture data format Polar express CMU mocap database
Class Overview: Background Introduction Background introduction for character animation Motion representation Forward kinematics Inverse kinematics Motion capture
Class Overview: Paper Presentation Motion capture data processing Motion warping/edit Motion retargeting Motion splicing Motion segmentation Motion compression Motion synopsis
Class Overview: Paper Presentation Data-driven motion synthesis Motion graphs/patches Motion interpolation Statistical motion synthesis
Class Overview: Paper Presentation Combine mocap data with other techniques Motion planning Physically based animation Key-framing
Class Overview: Paper Presentation Animation control Online animation control Offline Animation control
Class Overview: Paper Presentation Motion perception
Class Overview: Paper Presentation Data-driven approach for Facial animation Hand animation Skin deformation Animal animation
Class Overview: Paper Presentation Data Capture
Other Information My My homepage: My office: Rm 527D Bright Office hours: MW 4:00-5:00 Pm Course webpage:
Me Today Your background Graphics? Math? Coding? Your research Interest? Master/Ph.D. (year)? Why do you take this course?