# nv-tlabs/kimodo

Official implementation of Kimodo, a kinematic motion diffusion model for high-quality human(oid) motion generation.

Repository: https://github.com/nv-tlabs/kimodo
Canonical: https://ross.abutalabs.com/products/kimodo
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-07-13T17:31:48+00:00

## Health v2 (maintenance only)
Score: 56/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 92, release rhythm 35, longevity 12
- inputs: {"age_days": 170, "days_push": 51, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3365, forks 369 (observed 2026-08-28T04:07:58.146172+00:00)

## What it is
Kimodo is NVIDIA's official implementation of a kinematic motion diffusion model trained on 700 hours of motion capture data to generate high-quality 3D human and robot motions. It supports text-prompt control plus kinematic constraints (pose keyframes, end-effector targets, 2D paths), and ships with a CLI, an interactive timeline demo, and a motion generation benchmark.

## Use cases
- generate 3D human motion animations from text prompts
- create humanoid robot motions controlled by end-effector constraints
- author character animations with a timeline of prompts and keyframe poses
- benchmark and evaluate motion generation models on text and constraint following
- generate motion sequences from 2D paths and waypoints
- produce training or simulation data for humanoid robotics

## When to choose
- you need high-quality text- or constraint-controlled 3D human or humanoid motion generation
- you want an interactive tool to author motions with prompts and kinematic controls
- you need a standardized benchmark for comparing motion generation models
- you work with human or robot skeletons like SOMA, G1, or SMPL-X

## When to avoid
- you need real-time motion generation rather than offline diffusion inference
- you lack a GPU or work with very limited VRAM
- you need physics-based or physically simulated motion rather than kinematic motion
- you need a lightweight production animation tool without ML dependencies

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-inference, cli, benchmarking, data-generation
- domain: machine-learning, deep-learning, artificial-intelligence, robotics, simulation, gaming-tools
- platform: python, cli
- tags: motion-diffusion, motion-generation, human-motion, text-to-motion, motion-capture, humanoid, diffusion-model, animation, nvidia, linux, gpu

## Member repositories
- nv-tlabs/kimodo (main) score 56

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:58.146172+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:40:18.227309+00:00, confidence not recorded.
  - readme: https://github.com/nv-tlabs/kimodo (fetched 2026-08-28T04:07:58.146172+00:00, sha 75f31dd83a08)
- Data as of 2026-08-30T08:39:29.467469+00:00.