Digital Humans R&D Developing a Scan-Based, FACS-Driven Facial Workflow - CASE STUDY
Research & Development | 2017
Overview
IIn 2017, I had the opportunity to lead the R&D approach for a digital-human project at CyberAgent. Having already explored FACS-based facial rigging through a personal R&D project, I built on that initial research to develop a workflow for creating a realistic digital human.
I designed the workflow, from defining facial capture requirements and blendshape specifications to developing a FACS-based facial rig and exploring a reusable base-mesh pipeline. Following the initial experiments, the approach was applied to facial scans of an actress, taking the research from a preliminary test towards a scan-driven workflow.
Building on this earlier research, I am now revisiting aspects of digital-human creation through my own short-film project using Unreal Engine and MetaHuman. This ongoing exploration looks at how facial rigging, character setup, and performance workflows can be adapted to a modern cinematic pipeline, connecting the technical foundations established in 2017 with my current work.
Personal R&D: FACS-Based Facial Rigging and Digital human
F.A.C.S BASED RIG
Digital Human R&D : Concept Pitch & Workflow Planning
R & D Project | CyberAgent
Working with CyberAgent's scanning partner, I defined the facial expressions and blendshape requirements for the actress's capture session, including a target set of approximately 40–60 facial shapes. The capture process was also supervised to help ensure the required expressions were recorded, with feedback provided to the team during the subsequent facial-data cleanup.
Building a FACS-Based Facial Rig
Once the cleaned facial blendshapes were received, they were extracted and prepared for integration into the FACS-based facial rig, which was originally developed around individual facial action units during my earlier personal R&D tests.
This structure allowed facial movements to be controlled and combined, providing a flexible foundation for exploring a wider range of expressions.
The proposed automation workflow aimed to generate an actor-specific facial rig from a generic base rig, adapting the setup to the individual characteristics of each scanned actor. The goal was to streamline the process of transferring facial data and rig configurations to new digital humans while maintaining consistent topology and a controllable facial system.
This approach was intended to reduce repetitive manual work, improve reusability, and establish a more scalable pipeline for creating animation-ready digital humans from scanned facial data.
Revisiting the Approach with MetaHuman
Building on the earlier FACS-based facial rigging research, I am revisiting aspects of the digital-human workflow through my current short-film project using Unreal Engine and MetaHuman.
This exploration looks at how facial rigging, character setup, and performance workflows have evolved, and how these tools can be incorporated into a modern cinematic pipeline.
Further breakdowns of the process and ongoing experiments will be added as the project develops.
Comments