3D-Bionic-Designer

简体中文 | 日本語 | English

Introduction

• An Agent-driven 3D bionic product designer using Comfyui backend + Gradio frontend.
• Simply input a biological reference name, image, and target design (e.g., table, chair, small exhibition hall), and it will generate corresponding bionic products with text descriptions, images, and 3D models.
• The LLM / LVM models first analyze the input information and complete the autonomous design following the steps: "Design concept text" → "Visual description text" → "prompt" → "2D image output" → "3D model output".
• All computations are performed locally using open-source models, requiring no API or internet connection (after installation).

3D_bio_desigenr_demo_mute-3x.mp4

Installation

Install Comfyui

• Install comfyui
• Install comfyui-manager
• Run comfyui and load the workflow file, try to install missing plugins using comfyui-manager.
  • If installation fails, manually install the following comfyui plugins: Comfyui-Ollama-Describer, comfyui-3D-pack, comfyui-mixlab-nodes, Comfyui-Easy-Use

Install ollama

Install Other Dependencies

pip install gradio numpy Pillow websocket-client pymeshlab

Running Instructions

• Start ollama
• Start Comfyui, load the workflow file, and test if it works properly in comfyui.
• Modify server_address in app.py to your Comfyui address, e.g.: server_address = "127.0.0.1:8188"
• Modify comfyui_output_path in app.py to your Comfyui output path, e.g.: comfyui_output_path = "D:/01_DL/ComfyUI_windows_portable/ComfyUI/output", to ensure the Gradio frontend can access ComfyUI's output.
• Start the Gradio frontend.

  python app.py

• Visit 127.0.0.1:7860 to open the interface.