3D Model Anything skill
The scan-to-threejs skill refines a 3D scene in a loop, checking renders against room photos from matching camera angles.
- 2,993 ★ on the repository
- Last commit
Extended access
Why this level
- Optional depth reconstruction runs local image-processing code with external libraries (Open3D, Pillow).
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/3d-model-anythingDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Optional depth reconstruction runs local image-processing code with external libraries (Open3D, Pillow).
This is third-party code. Review the repository files before installing.
What it does
3D Model Anything combines three parts: an iPhone app with LiDAR that captures a room's geometry and photographs, a Three.js viewer for reviewing the result (dollhouse, walking, overhead and wireframe modes), and the `scan-to-threejs` skill, which guides an agent through measurement, modeling and checking the visual against photos from matching camera positions. A scan exported through the system share sheet is extracted into an ignored `data/captures-in/<room>` folder, and a new room is registered in `demos/index.ts`. This is an iterative reconstruction workflow, not a one-click converter: the agent compares renders against photos and refines the model in a loop.
Who it is for. LiDAR-equipped iPhone owners who want an editable 3D model of their room in code, not just a raw point cloud.
Good fit when
- A room scan needs to become an editable Three.js scene, not a static model.
- The agent should check the generated scene against real photos from matching camera angles.
- The result needs to open in a browser in several modes: dollhouse, walking, overhead.
- Several rooms are planned for one project through `demos/index.ts`.
Not a fit when
- An iPhone or iPad without LiDAR: scanning is unavailable, only importing and reviewing someone else's scans works.
- A finished App Store release is required: the app is currently source-available, and a store release may not exist yet.
- The subject is not a room or interior: the current version is specifically built for that.
Example request
Use scan-to-threejs to rebuild the room in data/captures-in/my-room and compare renders with the original photographs.Limitations
Scanning needs a LiDAR-equipped iPhone or iPad; devices without it can only import and review existing scans. The app is built from source via Xcode, and the README does not promise a finished App Store release. Optional depth reconstruction needs NumPy, Open3D and Pillow; photo comparison needs Pillow; capture scripts use the `agent-browser` CLI. The viewer needs Node.js 22.12 or newer. The author's own office scans and photo collections are not distributed with the repository, so Scan and Overlay are disabled in the published example and only work with the user's own captured data.
How to disable. Remove the skill folder from the agent's skills directory and the local project clone, and stop the viewer (`npm run dev` or `npm run preview`) if it is running.
FAQ
Is a separate App Store release required?
No, the app is source-available for building with Xcode; the README does not promise a separate App Store release.
What if my device has no LiDAR?
Scanning needs a LiDAR device, but importing and reviewing already-captured scans works without one.
How does the agent check model accuracy?
The skill compares scene renders against photographs from matching camera positions and refines the model in a loop; it is an iterative process, not a single pass.
Can I open the author's office example with my own scan?
The demo includes the reconstructed office and selected textures, but its Scan and Overlay modes are disabled: those only work with your own exported capture.
How do I add a new room to the project?
Place the export in `demos/<id>/` and register the room in `demos/index.ts`; the folder format keeps the `house-capture/5` contract.
Discussion
No questions or reviews yet.
Related
SkillsA skill rebuilds the object in a reference image as a procedural Three.js model written in code, not downloaded art
3dviz Pro Max
3dviz-pro-max
A skill for 3D visualizations (Three.js, Blender, glTF): an idea becomes a scene. Screen capture is optional through Playwright. Some demos are historical
One line naming a machine, and Claude Code builds an interactive physics-based 3D model of it in a single HTML file
A plugin that builds a 3D film, app, lesson, book or game on any subject, with the agent checking its own frames.
A Codex skill that builds a walkable Unreal Engine project from photos of a room or house, one question at a time.
A Claude Code workflow that builds a dressed, lit Blender scene from reference images or one paragraph, with graded renders.
README in short
The author keeps a public example of his own office rebuilt through this workflow, with several saved reconstruction versions and a playground page for experimenting with viewing modes.