ComfyUI-PromptEnhance

A ComfyUI custom node for enhancing prompts using OpenAI-compatible APIs (like DeepSeek, OpenAI, etc.).

Features

• Prompt Enhancement: Transform simple prompts into detailed, visual descriptions suitable for image generation models
• OpenAI-compatible API: Works with any OpenAI-compatible API (DeepSeek, OpenAI, Azure OpenAI, etc.)
• Customizable Template: Use the default prompt enhancement template or provide your own
• Configurable Parameters: Adjust temperature, max tokens, and timeout settings

Installation

1. Navigate to your ComfyUI custom nodes directory:

   cd ComfyUI/custom_nodes/
   

2. Clone or copy this folder:

   git clone <repo_url> ComfyUI-PromptEnhance
   # or simply copy the folder
   

3. Install dependencies (if not already installed):

   pip install requests
   

4. Restart ComfyUI

Nodes

Prompt Enhance (LLM)

Basic prompt enhancement node.

Inputs:

• prompt (required): The original prompt to enhance
• api_endpoint (required): OpenAI-compatible API endpoint (default: DeepSeek)
• api_key (required): Your API key
• model (required): Model name (e.g., deepseek-chat, gpt-4)
• prompt_template (optional): Custom enhancement template (use {prompt} as placeholder)
• temperature (optional): Sampling temperature (0.0-2.0, default: 0.7)
• max_tokens (optional): Maximum response tokens (100-8192, default: 2048)
• timeout (optional): Request timeout in seconds (10-300, default: 60)

Output:

• enhanced_prompt: The enhanced prompt string

Prompt Enhance Advanced (LLM)

Advanced version with additional options.

Additional Inputs:

• system_prompt (optional): System prompt for the LLM
• top_p (optional): Top-p sampling parameter (0.0-1.0, default: 1.0)

Outputs:

• enhanced_prompt: The enhanced prompt string
• original_prompt: The original input prompt (for comparison/reference)

API Endpoints

DeepSeek (Default)

https://api.deepseek.com/v1/chat/completions

Model: deepseek-chat

OpenAI

https://api.openai.com/v1/chat/completions

Model: gpt-4, gpt-3.5-turbo, etc.

Azure OpenAI

https://<your-resource>.openai.azure.com/openai/deployments/<deployment>/chat/completions?api-version=2024-02-15-preview

Default Prompt Template

The default template is designed to transform user prompts into detailed, visual descriptions that are:

• Faithful to the original intent
• Rich in visual details
• Free of metaphors and emotional rhetoric
• Directly usable by text-to-image models

Example Usage

1. Add the "Prompt Enhance (LLM)" node to your workflow
2. Connect a text input or type directly in the prompt field
3. Configure your API endpoint and key
4. Connect the enhanced_prompt output to your CLIP Text Encode or other text processing nodes

Example Workflows

Ready-to-use example workflows are available in the examples/ directory:

• SDXL Prompt Enhance Workflow: A complete SDXL workflow demonstrating prompt enhancement with visual preview of enhanced prompts. Perfect for learning how to integrate Prompt Enhance nodes into your workflows.

See the examples/README.md for detailed instructions on using the example workflows.

License

MIT License