This ComfyUI extension provides advanced prompt generation and image analysis capabilities, targeting users who want to enhance their AI image creation workflows. It offers nodes for GPT-4 powered text generation, image description via GPT-4 Vision, local LLM integration with Ollama, and sophisticated prompt structuring with dynamic category-based generation.

How It Works

The extension introduces several custom nodes. PromptGenerator and APNextNode allow for structured and randomized prompt creation, pulling elements from user-defined JSON files organized into categories. GPT4VisionNode leverages GPT-4 Vision to analyze images and generate detailed descriptions, with options for output detail and length. GPT4MiniNode and OllamaNode provide text generation capabilities using OpenAI's GPT-4 and local Ollama models, respectively, supporting custom base prompts and output formatting. A PGSD3LatentGenerator is also included for Stable Diffusion 3 latent creation.

Quick Start & Requirements

• Installation: Add the repository to your ComfyUI custom nodes directory.
• OpenAI API Key: Required for GPT4VisionNode and GPT4MiniNode. Set as an environment variable: OPENAI_API_KEY=sk-your-api-key-here.
• Ollama: Required for OllamaNode.
• Dependencies: Additional Python packages as specified in import statements.
• Custom Categories: Create JSON files in comfyui_dagthomas/data/next/[CATEGORY_NAME]/ for APNextNode customization.
• Example Workflow: Download apntest.json.
• Documentation: Detailed documentation is in progress.

Highlighted Details

• Dynamic Prompt Generation: APNextNode allows users to define custom categories and fields via JSON files, enabling highly flexible and repeatable prompt construction.
• GPT-4 Vision Integration: Enables image-to-text analysis for detailed image descriptions, useful for prompt seeding or content moderation.
• Local LLM Support: OllamaNode allows integration with local language models, offering an alternative to cloud-based APIs.
• SD3 Latent Generation: Includes a node specifically for generating latents compatible with Stable Diffusion 3 pipelines.

Maintenance & Community

• The project is marked as "beta" with documentation in progress.
• No specific community links (Discord, Slack) or notable contributors are mentioned in the README.

Licensing & Compatibility

• The README does not explicitly state a license.

Limitations & Caveats

• The project is in beta, and detailed documentation is still being developed.
• Functionality relies heavily on correctly structured JSON files for custom categories.
• Usage of OpenAI nodes requires an API key and incurs costs.