BodyVista: a free 3D human anatomy site with 2,900 structures

Flat atlases cannot teach spatial relationships — which nerve runs under which muscle, which membrane wraps which organ. BodyVista, an open-source anatomy and physiology learning site from FreedomAI, puts 2,900 structures, 12 body systems, 140 dynamic demos and 6 microscopic scenes into the browser. Core interactions run locally: no signup, no backend, and the live site opens on phone or desktop alike.

Core features

• Anatomy explorer. 2,900 structures organized by system and 7 body regions; rotate, zoom, progressively hide overlying structures, fade outer layers translucent, or focus a single one.
• Life in Motion. 140 demos (133 system-level, 5 whole-body, 2 oral and dental) with playback, progress and speed controls, connecting form to function.
• Microscopic Exploration. Six topics — alveolar gas exchange, the nephron, sarcomere contraction, neural synapses, intestinal villi and the vessel wall — running from organs down to cellular mechanisms.
• Anatomy Challenge. Five-question rounds of structure identification, finding structures in 3D, and microscopic knowledge, with feedback, results and replay.
• Drawing, measuring and sectioning. Adjustable brush, eraser, labels and measurement tools, plus perspective/orthographic switching and transverse, sagittal and coronal section planes.
• Structure knowledge and search. Look structures up by name for explanations, related structures and references; dental entries carry FDI tooth numbers. One click opens ChatGPT or DeepSeek with a prompt about the current structure.

Typical use cases

• Students. Search an unfamiliar term, find it on the model, then self-test with the challenge before an exam.
• Teachers. Hide outer layers, switch to a coronal section, and project the layered relationships during class.
• Science communicators. Screenshots and recordings for teaching material, with three interface languages to work in.

Quick start

Open freedomintelligence.github.io/BodyVista; a Hugging Face Space mirrors it. Self-hosting needs Node.js 22.12 or later:

git clone https://github.com/FreedomIntelligence/BodyVista.git
cd BodyVista && npm ci && npm run dev

Models and thumbnails ship inside the repository, so the first clone and first load take a while.

Summary

BodyVista’s code is MIT, but the models and content are not uniformly licensed: the README flags some anatomy records with “NC exceptions; research only” pending per-record verification, so don’t assume commercial use is fine. Animations and microscopic scenes are schematic, meant for teaching rather than clinical decisions. The repo was created on October 29, 2026; as of November 4 it counts 133 stars, built on JavaScript and Three.js, with VR and AR versions on the team’s stated roadmap. Next to LearnHarness, a desktop tutor that adapts problems to your understanding, BodyVista owns the spatial side of learning — the two pair well.