An agent skill that guides AI coding agents through creative 3D visualization, turning an idea into a Three.js or Blender scene with recipes, kits and frame capture.
81099JavaScriptMITUpdated last month
img2threejs/img2interior
An agent-driven skill that turns one interior room photo into an editable Blender scene, a Three.js viewer and a portable GLB model.
- Stars
- 46
- Forks
- 6
- Open issues
- 0
- Last commit
- 18 hours ago
What is it?
img2 Interior is an agent-operated skill for Claude Code, Codex and OpenCode, built on the img2 harness. Given one interior reference image, the agent inspects the picture, authors a Blender scene with geometry and physically based materials, renders it, compares the renders against the reference, and then builds a Three.js viewer.
The workflow produces three deliverables: a packed editable Blender master file, a runtime GLB with its lightmap and probe sidecars plus the viewer code, and a separate static portable GLB with baked appearance. Dimensions, hidden surfaces and camera values are labeled as estimates, since a single image is incomplete evidence. The project does not claim exact recovery or photorealism, and it is not a pretrained image-to-mesh model.
Installation uses npm to fetch the package into a dedicated prefix, then registers it with the img2 harness. Native authoring requires a separate Python 3.11 environment with bpy and numpy, and the source is available under a source-available license with restricted commercialization.
Key features
- 01Builds an editable Blender scene from one interior reference image.
- 02Renders and compares results against the reference, then repairs geometry and materials.
- 03Exports a packed .blend master, a runtime GLB with a Three.js viewer, and a portable GLB.
- 04Records resource provenance, assumptions and verification evidence in the project.
- 05Installs through npm and registers with the img2 harness for Claude Code, Codex and OpenCode.
Use cases
- Turning an interior design concept image into an explorable 3D room for client review.
- Creating an editable Blender starting point for an interior scene from a single reference photo.
- Publishing a browser-based walkthrough of a room with a Three.js viewer.
- Producing a standalone GLB model of a room for other glTF-compatible tools.
Quick start
- 1
Install the package from npm into a dedicated prefix
npm install --prefix "$HOME/.local/share/img2interior" @img2threejs/img2interior@0.3.1 - 2
Register the installed copy with the img2 harness
npx --yes @img2threejs/img2@0.4.0 add --link "$HOME/.local/share/img2interior/node_modules/@img2threejs/img2interior" --yes - 3
Restart the agent session, open your project and use the copy-and-use prompt with one room image
Things to know
- Not a pretrained or one-click model; results depend on scene-specific authoring and visual review.
- Requires Claude Code, Codex or OpenCode plus a separate Python 3.11 environment with bpy 4.3.0.
- Requires the img2 harness (0.4.0 or newer) and a WebGL-capable browser.
- License is listed as NOASSERTION; commercializing the tool itself requires prior written permission.
Topics
FAQ
- What is img2interior?
- An agent-driven skill that turns one interior room photo into an editable Blender scene, a Three.js viewer and a portable GLB model. img2 Interior is an agent-operated skill for Claude Code, Codex and OpenCode, built on the img2 harness. Given one interior reference image, the agent inspects the picture, authors a Blender scene with geometry and physically based materials, renders it, compares the renders against the reference, and then builds a Three.js viewer.
- Is img2interior free to use? What license does it have?
- img2interior is open source under the Other license. Check the license terms before using it in commercial products.
- Is img2interior still maintained?
- Actively maintained. The last commit to img2interior was on Oct 19, 2026.
- What can I use img2interior for?
- Turning an interior design concept image into an explorable 3D room for client review. Creating an editable Blender starting point for an interior scene from a single reference photo. Publishing a browser-based walkthrough of a room with a Three.js viewer. Producing a standalone GLB model of a room for other glTF-compatible tools.
- How do I install img2interior?
- Install the package from npm into a dedicated prefix: npm install --prefix "$HOME/.local/share/img2interior" @img2threejs/img2interior@0.3.1 Register the installed copy with the img2 harness: npx --yes @img2threejs/img2@0.4.0 add --link "$HOME/.local/share/img2interior/node_modules/@img2threejs/img2interior" --yes Restart the agent session, open your project and use the copy-and-use prompt with one room image
- Where can I get img2interior?
- The source code and installation instructions are on GitHub: https://github.com/img2threejs/img2interior
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Rebuilds an object from one reference image as procedural, animation-ready Three.js TypeScript code, not a mesh file.
18.4K1.7KPythonApache-2.0Updated 2 days ago
Plain-language agent skills for Claude Code, Codex CLI, Cursor CLI and Gemini CLI that turn session reports into a one-screen verdict anyone can follow.
1.4K119ShellMITUpdated 15 days ago
A Claude Code skill that turns a website URL into a 60 fps product launch video built from HTML, with music-synced sound effects and motion blur.
32247JavaScriptMITUpdated 5 days ago
An Agent Skill that turns a topic into a narrated explainer video with voiceover, hard subtitles and cover images, rendering HTML frames locally to MP4.
17921PythonMITUpdated 6 days ago
An agent skill for Claude Code and Codex that turns your app's real UI into launch films and short videos, cut to music and rendered locally.
1057JavaScriptMITUpdated 19 days ago