AI 3D
three.ws
Turn text or a photo into a living 3D avatar with an LLM brain and wallet.
★ 221 スターApache-2.0
ガイドを見る →img2threejsは、img2threejsが公開しているオープンソースのAI 3Dプロジェクトです（GitHub: img2threejs/img2threejs）。GitHubスターは17,720、フォークは1,486で、Apache-2.0ライセンスの下で公開されています。GitHubでの最終更新は2026年10月8日です。
Turn a reference image into a procedural, animation-ready Three.js model - as code
Choose a clear image of the single object you want rebuilt in 3D.
Run the npx installer to add the base skill to your agent; add plugin harnesses only if you need them.
Run the doctor command to confirm the setup is healthy.
Ask your agent to rebuild the image and review the generated Three.js scene.
The output is readable Three.js code you can edit, animate and version - not a baked mesh.
Reconstructions pass through quality gates before being accepted.
Designed to minimize the agent tokens needed per reconstruction (the README explains the approach).
Installs as a skill your existing agent uses; a legacy skill can also be cloned straight into a skills folder.
Optional plugins extend capability - the README advises installing only what you need.
Optional reference-fidelity tooling and a GLB-baseline character pipeline are available for harder subjects.
Get the source code from GitHub and enter the project directory.
git clone https://github.com/img2threejs/img2threejs.git
cd img2threejsThe installer sets up the skill; the doctor command verifies it.
npx --yes img2threejs@latest install
npx --yes img2threejs@latest doctorA dry run shows exactly what the installer would change.
npx --yes img2threejs@latest install --dry-runReaders working through the AI 3D shelf also stop at metashape-mcp, AI-driven photogrammetry: control Metashape over MCP. The AI 3D shelf also holds ARIS-in-AI-Offer — Bilingual AI interview cheat sheets plus a fact-checked academic homepage builder. Readers working through the AI 3D shelf also stop at OpenPointClass, Free, fast semantic segmentation of 3D point clouds.
It is maintained by the img2threejs project on GitHub.
Yes - img2threejs is Apache-2.0 licensed and free to use.
The tooling is free, but the reconstruction is performed by your AI agent, so you need an agent setup (for example Claude Code) with its own model access.
Compared with neural image-to-3D services, img2threejs outputs procedural, editable Three.js code optimized for web use and animation, rather than a static mesh file.
Run the npx installer, verify with the doctor command, and ask your agent to reconstruct a simple object first.
Turn text or a photo into a living 3D avatar with an LLM brain and wallet.
Turn phone video into 3D scenes your AI agent can query.
Shap-E — OpenAI's text-to-3D and image-to-3D model: generate 3D objects as implicit functions and render them as NeRFs or textured meshes, MIT
Nerfstudio — the plug-and-play studio for NeRFs and Gaussian splatting: turn photos of a real scene into explorable 3D, with a modular API and viewer, Apache-2.0
Baking Gaussian Splatting into Diffusion Denoiser for Fast and Scalable Single-stage Image-to-3D Generation and Reconstruction (ICCV 2025)
Stream ComfyUI output live into TouchDesigner.
LibreAIカタログに掲載されたAI 3Dプロジェクトの中で最もスターの多いもの — img2threejsが合わなかった場合に検討したい、現実的な候補リストです。
Control Blender with natural language: an MCP plugin that lets any LLM model, light and animate in 3D
Pascal Editor: an open-source 3D architectural editor with MCP tools for humans and AI agents
Give your AI agent CAD superpowers - generate real STEP/STL models from text
High-Resolution 3D Assets Generation with Large Scale Hunyuan3D Diffusion Models.
Official repo for paper "Structured 3D Latents for Scalable and Versatile 3D Generation" (CVPR'25 Spotlight).
カタログの中でimg2threejsとスター数が近いAI 3Dプロジェクト — 前後の棚もそのまま見て回れます。
Give your AI agent CAD superpowers - generate real STEP/STL models from text
High-Resolution 3D Assets Generation with Large Scale Hunyuan3D Diffusion Models.
Pascal Editor: an open-source 3D architectural editor with MCP tools for humans and AI agents
Official repo for paper "Structured 3D Latents for Scalable and Versatile 3D Generation" (CVPR'25 Spotlight).
Control Blender with natural language: an MCP plugin that lets any LLM model, light and animate in 3D
Shap-E — OpenAI's text-to-3D and image-to-3D model: generate 3D objects as implicit functions and render them as NeRFs or textured meshes, MIT