Cartwheel's new paper: compute-optimal scaling laws for human motion generation
tw_killian · x · 2026-08-26
Cartwheel has published its flagship paper, Compute-Optimal Scaling Laws for Human Motion Generation, the first demonstration that motion generation scales with compute.
Author Andrew Carr explains the key enabler: the field has been severely data-constrained — for years the largest dataset was 27 hours of mocap, which most papers trained and evaluated on, far too small for scale sweeps. Cartwheel's founding thesis was fixing this bottleneck: using video as the data source, they built a high-quality video-to-3D-motion estimator (validated with top entertainment companies), constructed the world's largest high-quality human-motion dataset, and trained hundreds of models to derive compute-optimal scaling laws.
More from Multimodal
- HyperFrames launches Reference Any Video and auto-applied Brand Kits for AI video — aftahi_ai · 2026-10-16
- Krea 2 + LoRA still beats Flux 3 Klein in pixel-art style test — Angaisb_ · 2026-10-16
- World Labs starts rolling out Atlas beta with pixel-perfect camera path control — theworldlabs · 2026-10-16
- Krea 3 image editing model lands on fal: high-quality edits in under 6 seconds — gorkem · 2026-10-16
- Tripo P2 vs Meshy 7.1: benchmarking today's best image-to-3D models for game assets — noahsolomon · 2026-10-16
- BytePlus-backed AI Film Award offers $1M grand prize for films 70% made with Seedance — DavidmComfort · 2026-10-16