Technical Papers

Faces

Monday, 11 August 3:45 PM - 5:00 PM | Vancouver Convention Centre, East Building, Ballroom B-C Session Chair: Mark Pauly, École polytechnique fédérale de Lausanne

Facial Performance Enhancement Using Dynamic Shape-Space Analysis

The facial performance of an individual is inherently rich in subtle deformation and timing details. This technique adds fine-scale spatio-temporal details and expressiveness to low-resolution art-directed facial performances, demonstrated on various forms of input.

Amit Bermano
Disney research Zürich, ETH Zürich

Derek Bradley
Disney Research Zürich

Thabo Beeler
Disney Research Zürich

Fabio Zund
Disney Research Zürich, ETH Zürich

Derek Nowrouzezahra
Université de Montréal

Ilya Baran
Disney Research Zürich

Olga Sorkine-Hornung
ETH Zürich

Hanspeter Pfister
Harvard University

Robert W. Sumner
Disney Research Zürich

Bernd Bickel
Disney Research Zürich

Markus Gross
Disney Research Zürich, ETH Zürich

Controllable High-Fidelity Facial Performance Transfer

This paper introduces a novel facial expression transfer and editing technique for high-fidelity facial animation data. The key idea is to decompose high-fidelity facial performances into large-scale facial deformation and fine-scale facial details, and transfer and edit them appropriately to reconstruct the desired retargeted animation.

Feng Xu
Microsoft Research Asia

Yilong Liu
Tsinghua University

Jinxiang Chai
Texas A&M University

Xin Tong
Microsoft Research Asia

Displaced Dynamic Expression Regression for Real-Time Facial Tracking and Animation

Introducing a calibration-free approach to real-time facial tracking and animation with a single video camera. The approach can robustly handle fast motions, large head rotations, exaggerated expressions, and large lighting changes.

Chen Cao
Zhejiang University

Qiming Hou
Zhejiang University

Kun Zhou
Zhejiang University

Rigid Stabilization of Facial Expressions

This paper presents the first automatic method to factor out rigid transformation from facial scans, which achieves professional-quality results on par with manual stabilization. The method stabilizes the facial expressions by explicitly aligning them to an estimate of the underlying skull using anatomically motivated constraints.

Thabo Beeler
Disney Research Zürich

Derek Bradley
Disney Research Zürich