Virtual Humans for Visual Effects: Machine Learning Methods for Face, Head and Body Performances

Today’s TV shows and films are supported by a large industry of visual effects studios for post-processing. To enhance actor performances, virtual humans are created and used to provide hyper-realistic 3D renderings of actors’ performances. The artists working on these high-quality virtual humans currently perform a heavy workflow of scanning actors in 4D to obtain hi-fidelity textures, rigging body models and capturing motion capture performances to apply adjustments to virtual humans and creatures. In this project, we will apply machine learning algorithms to improve the artists’ workflow and push forward research in performance enhancement for 2D and 3D performances.

Faculty Supervisor:

Angelica Lim

Student:

Partner:

Scanline VFX ULC

Discipline:

Computer science

Sector:

Information and cultural industries; Professional, scientific and technical services

University:

Simon Fraser University

Program: