3D-GPT enables procedural 3D modeling by leveraging large language models (LLMs) to generate control scripts for 3D asset creation. It targets researchers and developers interested in AI-driven content generation, offering a novel approach to automating complex 3D design workflows.

How It Works

The system utilizes an agent-based architecture where LLMs interpret natural language prompts to generate Python scripts. These scripts then control the Infinigen framework, a procedural 3D modeling tool, to create specific 3D assets. This approach allows for flexible and intuitive control over 3D generation through high-level language commands.

Quick Start & Requirements

• Install Infinigen: Follow instructions from the official Infinigen repository: https://github.com/princeton-vl/infinigen
• Install Dependencies:

  git clone git@github.com:Chuny1/3DGPT.git
  pip install openai==0.27.8
  

• Setup: Requires an OpenAI API key, which must be added to parser.py. Users can select their preferred GPT model via OpenAI's platform.
• Run: python LLM/parser.py

Highlighted Details

• Agent implementation for procedural 3D modeling with LLMs.
• Generates Python scripts to control Infinigen for asset creation.
• Example usage and function documentation format provided.
• Future releases include modified Infinigen and a configuration file output for easier LLM control and human editing.

Maintenance & Community

The project is associated with authors from the 3DV Conference. Updates are planned, including modifications to Infinigen and the LLM parser.

Licensing & Compatibility

The repository does not explicitly state a license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The current release focuses on agent implementation and script generation. Upcoming releases promise enhancements for user-friendliness and LLM control, suggesting the current version may be less refined or user-friendly. An OpenAI API key is a mandatory requirement.