I’ve been building Posecode, a human-readable language for describing movement.
Instead of storing motion only as keyframes, movements can be written as joint actions, timing and constraints:
knees: flex 95
hips: flex 80
repeat 10
The browser renderer is built with Three.js and includes range-of-motion validation, forward kinematics, basic IK and ground-lock behavior.
Playground: Posecode Playground: Write and Preview 3D Human Movement as Text
Source: GitHub - posecode-dev/posecode: Posecode is a tiny language an LLM can write, designed for physiotherapy, mobility, posture, yoga, and training, that renders to an animated 3D figure in the browser. Every joint is clamped to a safe range of motion, so the result is always anatomically plausible. · GitHub
I’m currently improving foot contact and transitions between movement phases. Feedback on the motion, renderer architecture or presentation would be very useful.