A lightweight middleware that connects RapidRAW to a ComfyUI backend for fast, self-hosted generative AI edits.

Warning: This project is a work in progress and considered unstable for the average user. Official support will begin with the release of RapidRAW v1.4.9.

This server acts as an intelligent cache between RapidRAW and ComfyUI to make generative edits fast.

Instead of sending a huge source image for every prompt change, the full image is sent only once. For all subsequent edits, only the tiny mask and text prompt are transferred. The connector sends the full job to ComfyUI and returns only the cropped, edited patch. This minimizes network traffic and makes the editing experience feel instant.

• A running instance of ComfyUI.
• Python 3.10+

git clone https://github.com/CyberTimon/RapidRAW-AI-Connector.git
cd RapidRAW-AI-Connector
pip install -r requirements.txt

All settings are managed via environment variables. The defaults should work for a standard local ComfyUI setup. You can change them by setting variables like COMFY_HOST and COMFY_PORT before running the script.

python main.py

In RapidRAW's settings, point the Self-Hosted AI Backend to the connector's address (e.g., http://127.0.0.1:5000).

Tweak your generative process by editing the workflow.json file. You can use custom models, nodes, and samplers by updating the workflow and corresponding node IDs in engine.py.

This AI Connector is licensed under the Apache License 2.0. See the LICENSE file for more details.