Blazing a New Trail of Curls

How to read this

Category

Pixar SIGGRAPH Talk on a production grooming and simulation pipeline for tightly coiled hair

Contributions

• Develops a new grooming structure for Toy Story 5's character Blaze that keeps render hairs and simulation guide hairs coincident, so curl shape carries through from groom to simulation without the warping loss of older pipelines
• Builds an intermediate high resolution guide representation using procedural operators, scraggle, loft, curl width and count, that supports fast interactive resculpting during look development
• Introduces Fizt Strands, a curve simulation formulation built on Pixar's Fizt cloth solver, adapting cloth dihedral bending for twist and bending forces on curls
• Fixes a specific instability in tightly curled hair, where degenerate or flipped virtual triangles at an edge caused sudden violent twist forces, by tying twist stiffness to bend stiffness
• Reports reducing look development iteration cycles from weeks to days by moving curl exploration into review sessions instead of offline simulation passes

Context

This replaces Pixar's earlier hair simulator Taz, which debuted on the curly haired character Merida in Artistic Simulation of Curly Hair (2013) and had stability problems with Blaze's much tighter curls. Fizt Strands keeps the same production goal, believable curl dynamics matched to a specific character's ethnicity and personality, but solves it on top of the newer Fizt cloth solver instead of Taz.Builds on: Artistic Simulation of Curly Hair

Correctness

The claims are grounded in shipped production use on Toy Story 5 rather than a benchmark suite or user study. This is a 2 page SIGGRAPH Talk, so expect qualitative before and after descriptions of the collision and stability problems solved, not measured error numbers or a comparison table.

Clarity

Very readable, written by and for production groom and cfx artists and TDs, it assumes familiarity with hair grooming and cloth solvers but explains the specific triangle degeneracy bug in plain terms.

How to read it

First pass, read the abstract and the groom process section to see the coincident guide and render hair idea. Second pass, read the simulation section closely for the Fizt Strands formulation and the twist stiffness fix, that is the reusable technical idea. Third pass is worth it for anyone maintaining a cloth based hair or fur solver who hits similar curl instability, otherwise the talk is short enough that a single full read covers it.