← ArchivePaper2020
Dynamic Facial Asset and Rig Generation from a Single Scan
Jiaman Li, Zheng-Fei Kuang, Yajie Zhao, Mingming He, Karl Bladin, Hao Li
Generates personalized blendshapes, physically-based textures, and full facial rig from a single face scan automatically.
Abstract
The creation of high-fidelity computer-generated (CG) characters for films and games is tied with intensive manual labor, which involves the creation of comprehensive facial assets that are often captured using complex hardware. To simplify and accelerate this digitization process, we propose a framework for the automatic generation of high-quality dynamic facial models, including rigs which can be readily deployed for artists to polish. Our framework takes a single scan as input to generate a set of personalized blendshapes, dynamic textures, as well as secondary facial components (e.g., teeth and eyeballs). Based on a facial database with over 4, 000 scans with pore-level details, varying expressions and identities, we adopt a self-supervised neural network to learn personalized blendshapes from a set of template expressions. We also model the joint distribution between identities and expressions, enabling the inference of a full set of personalized blendshapes with dynamic appearances from a single neutral input scan. Our generated personalized face rig assets are seamlessly compatible with professional production pipelines for facial animation and rendering. We demonstrate a highly robust and effective framework on a wide range of subjects, and showcase high-fidelity facial animations with automatically generated personalized dynamic textures.
How to read this
- Category
- Method: automatic facial asset and rig generation
- Contributions
- Generates a full facial rig (personalized blendshapes, dynamic physically-based textures, and secondary components like teeth and eyeballs) from a single neutral scan
- Uses a self-supervised network trained on a 4,000+ scan pore-level database to learn personalized blendshapes from template expressions
- Models the joint distribution of identities and expressions, and outputs assets compatible with professional production pipelines
- Context
- Builds on automatic anatomical face modeling, related to Cong et al.'s 'Fully Automatic Generation of Anatomical Face Simulation Models', shifting from simulation-model fitting to learned single-scan rig generation.Builds on: Fully Automatic Generation of Anatomical Face Simulation Models
- Correctness
- Assumes a single neutral scan plus a large curated scan database suffice to infer plausible personalized expression rigs; demonstrated as a high-fidelity pipeline, but output identity/expression range is bounded by the training database's demographics and the assets are meant as an artist-polishable starting point rather than final.
- Clarity
- Readable systems-style write-up; a first pass conveys the single-scan-to-rig pipeline, a second pass is needed for the self-supervised blendshape learning and joint identity-expression model.
- How to read it
- Focus on the input/output of each stage and where the 4,000-scan database constrains results; second pass for the network if you build face-rig automation.
Related work
- Creating an Actor-Specific Facial Rig from Performance Capture 2016 / DigiPro
- Neural Face Rigging for Animating and Retargeting Facial Meshes in the Wild 2023 / SIGGRAPH
- Position Manipulation Techniques for Facial Animation 2016 / PhD Thesis
- RigAnyFace: Scaling Neural Facial Mesh Auto-Rigging with Unlabeled Data 2025 / arXiv
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 →