Physically Based Rigging for Artist-Directed Facial Animation

How to read this

Category

Physics-based facial rigging method (hybrid blendshape and simulation rig)

Contributions

• Couples a traditional blendshape facial rig with a finite-element soft-tissue simulation, letting artists pose the face with familiar blendshape sliders while the underlying flesh deforms with physically plausible volume and secondary motion.
• Derives muscle and tissue parameters so the physical simulation stays anatomically plausible while remaining directly artist-controllable rather than purely simulation-driven.
• Targets the specific production complaint about pure blendshape rigs, that in-between poses can lose volume or look mechanically interpolated, by letting simulation fill in nonlinear detail between artist-authored key poses.

Context

This paper builds on Sifakis et al.'s 2005 work on automatically determining facial muscle activations from sparse marker data, moving that anatomical muscle simulation machinery from a passive capture-driven tool into an artist-directed rig. It sits earlier in the same physically based facial animation lineage as Cong, Bhat and Fedkiw's 2016 Art-Directed Muscle Simulation, which pushed the same blendshape-meets-simulation idea further toward production scale at ILM.Builds on: Automatic Determination of Facial Muscle Activations from Sparse Motion Capture Marker Data

Correctness

The paper's claims rest on physically motivated finite-element tissue modeling validated against artist-directed poses rather than against ground truth performance capture, so its correctness is really a question of whether the physics produces results artists find plausible and controllable, which the paper demonstrates through example poses and comparisons to pure blendshape interpolation. The exact abstract text could not be retrieved for this guide, there was no local PDF and web search surfaced related citing work rather than the paper itself, so specifics here are drawn from the archive's note plus the paper's well documented position in the physically based facial rigging literature, including its direct citation by Cong et al. 2016.

Clarity

As an SCA paper it assumes familiarity with finite-element simulation and blendshape rig fundamentals, it is written for facial rigging and physical simulation researchers rather than as a tutorial. The artist-directability framing keeps the motivation clear even for a reader without a finite-element background.

How to read it

First pass, read the abstract and any figures comparing pure blendshape interpolation against the physically simulated result, since that comparison is the paper's whole argument. Second pass, read how the finite-element tissue model is parameterized and driven by the blendshape controls, this is the mechanism that lets artists keep a familiar workflow on top of the simulation. Third pass, read Cong et al.'s 2016 Art-Directed Muscle Simulation paper alongside it to see how the same blendshape-drives-simulation idea matured into a production pipeline at ILM four years later.