Face Off

Monday, 25 July, 3:45 pm - 5:35 pm, Anaheim Convention Center, Room 304 AB

ILM Facial Performance Capture

ILM's facial performance capture system records an actor's on-set facial performance and reproduces it with cinematic fidelity onto a realistic digital character. It has brought to life over two dozen characters on over 1200 performances for films including "Warcraft", "Teenage Mutant Ninja Turtles", and "Star Wars: The Force Awakens".

Brian Cantwell
Industrial Light & Magic

Paige Warner
Industrial Light & Magic

Michael Koperwas
Industrial Light & Magic

Kiran Bhat

"Kung Fu Panda 3": Mandarin Lip-Sync Reanimation Process and Pipeline

Producing an alternate version of "Kung Fu Panda 3", reanimated with Mandarin dialog, presented unique production challenges for both DreamWorks Animation and Oriental DreamWorks, as each group has its own pipeline and toolset. This talk discusses the workflows and tools involved in creating this first-of-its-kind film.

Mark Edwards
DreamWorks Animation

Robert Armstrong

Alvin Ho Tsun Chuen
Oriental DreamWorks

Shape-Analysis-Driven Surface Correction

This talk introduces a new generalized way of storing and restituting surface corrections and automating their blending by using live surface-shape analysis combined with interpolators based Radial Basis Function (RBF).

Claude Levastre
ON Animation Studio

As-Rigid-As-Possible Deformation Transfer for Facial Animation

This novel tool for transferring facial motion capture performance combines the shape deformation techniques of Deformation Transfer and As-Rigid-As-Possible, and allows users to control the balance between animation fidelity and physically plausible deformation.

Lucio Moser
Digital Domain 3.0, Inc.

David Corral
Digital Domain 3.0, Inc.

Doug Roble
Digital Domain 3.0, Inc.

Warping With Accumulated Motion Vectors

From facial lacerations that follow slight stretching of the skin to deformations of hard-bodied objects or subtle nuances of water simulations, this method automatically warps an image across a sequence from any arbitrary frame to any other using accumulated motion vectors, minimizing the keyframes needed to produce high-quality results.

Stephen Willey
Double Negative

Gregory Keech
Double Negative