This repository provides a labeling extension for Automatic1111's Stable Diffusion Web UI, enabling users to generate Booru-style tags for images using various AI models like DeepDanbooru. It's designed for users of Stable Diffusion Web UI who want to automate image tagging for organization or dataset creation.

How It Works

The extension integrates with the Stable Diffusion Web UI, allowing users to select an interrogation model (e.g., Waifu Diffusion 1.4 Tagger, DeepDanbooru models) and process single or multiple images. It leverages pre-trained models to analyze image content and output descriptive tags, similar to how Booru websites categorize artwork.

Quick Start & Requirements

• Installation: Clone the repository into the extensions/ directory of your Stable Diffusion Web UI installation.
• Prerequisites: Stable Diffusion Web UI.
• Model Download: Models are downloaded automatically on first run or can be manually placed in the models/deepdanbooru/ directory.
• Restart: A full restart of the Stable Diffusion Web UI is required after installation.

Highlighted Details

• Supports multiple interrogation models, including Waifu Diffusion 1.4 Tagger and various DeepDanbooru configurations.
• Allows tagging of single or multiple image files.
• Borrowed code from DeepDanbooru and MrSmillingWolf's tagger projects.

Maintenance & Community

This branch is unmaintained. Users are directed to a fork at picobyte/stable-diffusion-webui-wd14-tagger for current development.

Licensing & Compatibility

Copyright is public domain, with exceptions for borrowed code. Specific license details for borrowed components are not explicitly stated.

Limitations & Caveats

The primary repository is unmaintained, indicating potential lack of updates, bug fixes, or support. Users should refer to the linked fork for ongoing development.