AQnodes for ComfyUI
Nodes for ComfyUI, mostly for llm calling and automatizing pulid multiface.
Nodes (26)
Collapse a whole batch of images into one — mean or median
Photoshop blend modes, without leaving ComfyUI
CLIP skip as a standalone node (SD 1.5 anime users, this one's for you)
Steal a reference photo's color grade for a whole batch of images
Turn half-baked album-cover ideas into a full prompt (with a free negative prompt)
JSON out, images in
One Gemini call that returns a whole song's metadata — lyrics, BPM, cover art, YouTube text
Pump the fine detail from one image into another
Pick between up to four image+mask pairs with one number
Pad an image with transparent borders — with an RGBA caveat
A counter that ticks up every run — for seeds, indices, and rotating variants
Turn a base64 image string into a ComfyUI IMAGE (and a mask from the alpha)
Split a mask+image batch into per-item lists for multiface PuLID
PuLID Flux identity for several faces at once — the pack's flagship
The same song-metadata schema, but via OpenAI's Responses API (GPT-5 class)
Run the acestep music-metadata flow against any OpenAI-compatible API
Split any JSON into the acestep output slots
Run Qwen2.5-VL locally — vision, JSON mode, no API key
Load a Qwen2.5-VL model once, feed it to AQ_Qwen
A random number that ignores its input — and that's fine
Save a WEBP to disk and hand the base64 straight to your own frontend
POST your generated images to any endpoint from inside the graph
One image, one audio track, straight to .mp4
The text-encode node for Qwen-Image-Edit style conditioning
The Qwen-Image-Edit text encoder, now with three reference images and a dozen prompt templates
Upload a finished mp4 to YouTube straight from the ComfyUI graph
ComfyUI-AQnodes
A collection of custom nodes for ComfyUI that enhance your workflow with additional functionality for image processing, AI model integration, and API interactions.
Installation
- Clone this repository into your ComfyUI custom_nodes directory:
cd ComfyUI/custom_nodes
git clone https://github.com/yourusername/ComfyUI-AQnodes.git
- Install the required dependencies:
cd ComfyUI-AQnodes
pip install -r requirements.txt
Available Nodes
Image Processing Nodes
AQ_MasksAndImagesAsList
- Purpose: Combines multiple masks and images into a list format for loop processing
- Use Case: When you need to process multiple images with masks and send them to AQ_multiface_ApplyPulidFlux node
- Example: Combining multiple segmentation masks for batch image processing
AQ_multiface_ApplyPulidFlux
- Purpose: Applies PULID effects to multiple faces in an image
- Example: Feed with multiple images and masks from AQ_MasksAndImagesAsList node
Transformer Model Nodes
AQ_QwenLoader
- Purpose: Loads the Qwen transformer model for text (with vision) generation and processing
- Use Case: When you need to initialize the Qwen model for text-based operations (extract image feautures and transform them into something new)
- Example: Setting up Qwen for text generation or image analysis tasks
AQ_Qwen
- Purpose: Interfaces with the Qwen transformer model for text generation (with vision) and processing
- Use Case: Text generation, completion, or image analysis using the Qwen model
- Example: Generating descriptions for images or creating creative text content (extract image feautures and transform them into something new)
API Integration Nodes
AQ_Gemini
- Purpose: Integrates with Google's Gemini AI model through its API
- Use Case: When you need advanced AI capabilities provided by Gemini
- Example: Using Gemini for complex text generation or analysis tasks (extract image feautures and transform them into something new)
- Note: Requires API key configuration
AQ_SendImageToAPI
- Purpose: Sends images to specified API endpoints
- Use Case: When you need to integrate with external image processing services or don't want to pool comfy to see if generation is ready (when using api)
- Example: Sending images for analysis, storage, or processing to external services
- Note: Requires API endpoint configuration
Configuration
Some nodes may require additional configuration:
-
For API nodes (AQ_Gemini, AQ_SendImageToAPI):
- API keys should be stored securely
- Configure endpoint URLs as needed
-
For transformer nodes (AQ_Qwen):
- Ensure sufficient system resources
- Models will be downloaded automatically on first use
Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
License
See the LICENSE file for details.