This project provides a ComfyUI node for managing and inputting Stable Diffusion prompts, specifically designed to address the usability challenges for non-English speakers. It offers a categorized library of thousands of prompts, enabling quick selection and input without requiring English proficiency, thereby streamlining the prompt engineering process for a broader user base.

How It Works

The core of the project is the SixGodPrompt node for ComfyUI. It allows users to double-click an input field to open a searchable interface populated with a categorized library of prompts. The system supports custom prompt libraries, Chinese input, prompt weighting, positional adjustments, and dynamic random prompts using a specific syntax (e.g., #[red,blue,yellow]). It also integrates with large language models (LLMs) in .gguf format, such as Qwen, for translation and random inspiration.

Quick Start & Requirements

• Install via ComfyUI custom nodes: sh https://github.com/thisjam/comfyui-sixgod_prompt.git
• Prerequisites: llama_cpp_python (with GPU support recommended, e.g., CUDA 12.1+).
• LLM Model: Requires .gguf format models (e.g., qwen1_5-4b-chat-q2_k.gguf).
• Documentation: https://abetlen.github.io/llama-cpp-python/whl/

Highlighted Details

• Thousands of categorized prompts to facilitate non-English users.
• Custom prompt library support to prevent overwrites during updates.
• Dynamic random prompt generation syntax: #[red,blue,yellow].
• Integration with LLMs for translation and inspiration.

Maintenance & Community

• Active community support via QQ group: 390942276.
• Prompt library maintained by community contributors.

Licensing & Compatibility

• License not explicitly stated in the README.
• Compatibility with ComfyUI and Stable Diffusion WebUI (Forge) versions.

Limitations & Caveats

The project's license is not specified, which may impact commercial use. While it aims to be user-friendly for non-English speakers, the LLM integration for translation and inspiration requires separate setup and model downloads.