3D Model Anything

Overview

3D Model Anything is a room-capture and reconstruction workflow built around an iPhone app, a Three.js browser viewer, and the scan-to-threejs agent skill. It records room geometry and photographs on an iPhone, then guides an AI coding agent through measurement, modeling, and matched-camera visual comparisons to produce an editable Three.js scene. The project includes a detailed office demo and focuses on rooms and interiors.

Key Features

• iPhone app records room geometry and photographs for reconstruction.
• scan-to-threejs agent skill guides an AI coding agent through measurement, modeling, and matched-camera visual comparisons.
• Three.js viewer offers dollhouse, walking, overhead, wireframe, and knolling views.
• Includes Wes's office demo with a reconstructed room and selected textures.
• Supports exporting captures through the system share sheet for use in the agent workflow.
• Browser viewer includes camera presets, render quality controls, and copyable view links.

Use Cases

• Rebuild a captured room or interior as an editable Three.js scene using the scan-to-threejs skill.
• Explore reconstructed rooms in a browser through dollhouse, walk-around, overhead, wireframe, and knolling views.
• Import and review scans on non-LiDAR devices.

Getting Started

• Use Node.js 22.12+ or a newer supported release for the viewer. Run npm ci and npm run dev, then open the local viewer on port 5188.
• To capture your own space, build the iPhone app with Xcode, scan and share a capture, extract it into data/captures-in/<room>/, and ask an agent that supports local skills to use scan-to-threejs with the extracted export.
• Helpers need Python 3; photo comparisons need Pillow, and optional depth reconstruction needs NumPy, Open3D, and Pillow. Capture scripts use the agent-browser CLI.

Deployment & Requirements

• Node.js 22.12+ or a newer supported release for running the viewer.
• Xcode for building the iPhone app; scanning needs a LiDAR-equipped iPhone or iPad.
• Python 3 for helpers, Pillow for photo comparisons, and NumPy, Open3D, and Pillow for optional depth reconstruction.
• The agent-browser CLI for capture scripts.

Before You Adopt

• License: Other. Review its terms before using, modifying, or distributing the project.
• The reconstruction workflow is iterative and aimed at rooms and interiors, not one-click perfect conversion.
• Scanning requires LiDAR hardware; non-LiDAR devices can only import and review scans.
• The app is source-available, but an App Store release is separate and should not be assumed.
• Only the office example is distributed, and captures should remain private; inspect assets before publishing demos.