Vellum 2 - Dress Sim
A brief two part vellum project inspired by metaverse applications in the fashion and retail spaces.
I used one vellum solver to create the resting form for a simple “dress”. I then merged this data into a separate solver to test different cloth variables that might imitate different real-world fabrics. This test used a mixamo animation that’s been retargeted to the UE mannequin skeleton before exporting to Houdini.
Part 1 - Retarget Mixamo animation to UE skeleton.
Before retargeting the animations, it helps to adjust the UE skeleton’s default A pose to match the Mixamo T pose as closely as possible. This part is tedious but the UE pose can be saved for future projects to quickly convert Mixamo animations into something usable by UE skeleton or Metahumans skeletons.
Part 2 - Generate the dress’ resting shape.
The resting form of the “dress” was created using a simple vellum solver and a flat revolved surface. two rows of vertices near the center are fitted to the waist and assigned the group “belt”. These points are vellum constrained to the reduced UE mannequin geometry for the first sim.
Part 3 - Apply cloth traits and simulate!
Dress and animation interactions are separated into their own Vellum Solver. The same reduced (Blast node) mannequin and “belt” point group are reused as the collision and constraints. In this isolated geo node, I am tailoring (pun intended) properties like mass and thickness for this cloth. The dress is now cached in this geo and can be used as collision/contraint data for other clothing like a top or flowing belt.
Camera Move - Focal point translation, rotation, and camera properties are parented and keyed using 3 separate nodes.
