NVIDIA Open-Sources Kimodo for Text-to-3D Human Motion Generation

NVIDIA's Toronto lab released Kimodo, the first open-source diffusion model that generates full-body 3D human and robot motions from natural language with kinematic constraints. With about 282M parameters, it runs on a single RTX 3090 and ships with an Apache-2.0 codebase plus a Hugging Face motion-generation benchmark.

2026-08-31 ~ 2026-08-31 · 4 related posts

• NVIDIA's Kimodo Turns Text Into Full-Body 3D Human and Robot Motion — maier_ak · 2026-08-31
• NVIDIA Kimodo includes HF benchmark, code is Apache-2.0 — maier_ak · 2026-08-31
• Deep Dive: NVIDIA's Kimodo, a full open-source pipeline for text-to-3D-motion — maier_ak · 2026-08-31

1 near-duplicate retellings: maier_ak