← ArchivePaper2020
FaceBaker: Baking Character Facial Rigs with Machine Learning
Machine learning method that approximates complex facial rig deformations, reducing evaluation cost and enabling portability of proprietary Pixar rigs.
Abstract
Character rigs are procedural systems that deform a character’s shape driven by a set of rig-control variables. Film quality character rigs are highly complex and therefore computationally expensive and slow to evaluate. We present a machine learning method for approximating facial mesh deformations which reduces rig computations, increases longevity of characters without rig upkeep, and enables portability of proprietary rigs into a variety of external platforms. We perform qualitative and quantitative evaluations on hero characters across several feature films, exhibiting the speed and generality of our approach and demonstrating that our method out performs existing state-of-the-art work on deformation approximations for character faces.
How to read this
- Category
- Method: machine-learning rig approximation
- Contributions
- A learning method that approximates film-quality facial mesh deformations to cut rig evaluation cost
- Increases character longevity without rig upkeep and enables porting proprietary rigs to external platforms
- Qualitative and quantitative evaluation on hero characters across several feature films
- Context
- Continues the line of neural deformation approximation for production faces, explicitly building on Bailey et al.'s Fast and Deep Deformation Approximations (FDDA).Builds on: Fast and Deep Deformation Approximations
- Correctness
- Validated on Pixar hero characters and reported to outperform prior deformation-approximation work, but it is an approximation of a specific class of proprietary rigs, so fidelity and generality outside the tested characters should not be assumed.
- Clarity
- Accessible; a first pass conveys the motivation and pipeline, with a second pass for the network and training details.
- How to read it
- First pass for why baking a rig helps (speed, portability, longevity); second pass on the network design and the qualitative/quantitative comparisons versus FDDA if you need to judge accuracy.
Builds on
Built upon by
Nothing yet.
Related work
- Accurate Face Rig Approximation with Deep Differential Subspace Reconstruction 2020 / SIGGRAPH
- Position Manipulation Techniques for Facial Animation 2016 / PhD Thesis
- FaceLab: Scalable Facial Performance Capture for Visual Effects 2020 / DigiPro
- Fast and Deep Facial Deformations 2020 / SIGGRAPH
Keywords
This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →