October 30, 2025
Cloth sim, real drama
ZOZO's Contact Solver (for physics-based simulations)
ZOZO’s no-clip physics flex lands with emoji storm and AI side-eye
TLDR: ZOZO released a physics tool that stops objects from clipping and scales to huge scenes, but it’s slow per frame. Comments erupted over real-time limits, emoji-heavy docs, and a mysterious “Claude” commit, splitting fans between awe at the research and skepticism about practical use.
ZOZO dropped a shiny physics tool that promises zero “objects passing through each other,” even with jaw‑dropping mega scenes—think 150 million collisions at once. It runs on graphics cards and keeps cloth from stretching more than 1%, all wrapped in Docker and one‑click Jupyter demos. The crowd? Equal parts wow and eyebrow. One fan pointed to a quick explainer via Two Minute Papers, but the party paused when a blunt comment declared: not realtime. Minutes per frame means this is movie‑quality, not game‑night.
Then came the drama: a commit log shout‑out—“claude 19 commits, +21,000 lines”—ignited speculation and jokes about who (or what) wrote the code. The resident engineers chimed in with serious questions: contact physics is notoriously hard, and they want proof it handles soft cloth slamming into stiff objects without faking the forces. Meanwhile, the emoji parade in ZOZO’s docs sparked a vibe war: cute branding or distracting fluff? The community split into camps—research flex vs. real‑world practicality—while snarky comments turned the release notes into a mini‑meme fest. Verdict: impressive tech, big promises, and a thread that’s just as entertaining as the demo videos.
Key Points
- •The solver handles shells, solids, and rods with penetration-free contact resolution.
- •Both contact and elasticity solvers run massively in parallel on the GPU using single precision.
- •Cloth inextensibility is enforced with strict upper bounds (e.g., ≤1% stretch).
- •Scales to beyond 150 million contacts and is validated via GitHub Actions stress tests (10 consecutive runs).
- •Includes Docker packaging, JupyterLab examples, documented Python APIs, and cloud deployment guides with an AWS budget table.