ComfyUI Extension: ComfyUI WD 1.4 Tagger

ComfyUI WD 1.4 Tagger Introduction

ComfyUI-WD14-Tagger is an extension for ComfyUI that allows you to automatically generate descriptive tags for images, similar to those used on booru-style image boards. This can be incredibly useful for AI artists who want to quickly and accurately tag their artwork for better organization, searchability, and sharing. The extension leverages advanced models to analyze images and generate relevant tags, saving you time and effort.

How ComfyUI WD 1.4 Tagger Works

The extension works by using pre-trained models to "interrogate" or analyze an image and generate a list of tags that describe its content. Think of it as having a very knowledgeable art critic who can look at your image and instantly tell you what it depicts. The models have been trained on large datasets of images and their associated tags, allowing them to recognize a wide variety of subjects, styles, and elements in your artwork.

When you input an image into the extension, it processes the image through the selected model, which then outputs a list of tags along with confidence scores. These scores indicate how certain the model is that a particular tag applies to the image. You can set thresholds to filter out tags with low confidence scores, ensuring that only the most relevant tags are included.

ComfyUI WD 1.4 Tagger Features

Tagging and Outputting Multiple Batched Inputs

The extension supports tagging multiple images at once, which can be a huge time-saver if you have a large collection of artwork. You can batch process images and get tags for all of them in one go.

Model Selection

You can choose from different interrogation models, each with its own strengths. For example, the MOAT model is the newest and offers advanced tagging capabilities, while the ConvNextV2 model is very popular for its accuracy. You can experiment with different models to see which one works best for your needs.

Threshold Settings

• Threshold: This setting allows you to specify the minimum confidence score a tag must have to be included in the results. For example, setting a threshold of 0.5 means that only tags with a confidence score of 50% or higher will be considered.
• Character Threshold: Similar to the general threshold, but specifically for character tags. This can be useful if you want to ensure that character names are only included if the model is very confident.

Exclude Tags

You can provide a comma-separated list of tags that you want to exclude from the results. This is useful if there are certain tags that you know are irrelevant or incorrect for your images.

Quick Interrogation

You can quickly interrogate images directly from any node that displays an image, such as LoadImage, SaveImage, or PreviewImage nodes. Simply right-click on the node or the image and select WD14 Tagger from the menu.

ComfyUI WD 1.4 Tagger Models

The extension uses models created by SmilingWolf. Here are some of the models you can use:

• MOAT: The newest model, offering advanced tagging capabilities.
• ConvNextV2: A popular model known for its accuracy and reliability. You can try out these models and see which one provides the best results for your specific needs. Different models may perform better with different types of images, so feel free to experiment.

What's New with ComfyUI WD 1.4 Tagger

2023-05-14 Update

• Moved to Own Repository: The extension now has its own dedicated repository, making it easier to manage and update.
• Model Downloading: Models are automatically downloaded at runtime if they are missing, simplifying the setup process.
• Support for Multiple Inputs: You can now tag multiple images at once, streamlining your workflow. These updates make the extension more user-friendly and efficient, allowing you to focus more on your art and less on technical details.

Troubleshooting ComfyUI WD 1.4 Tagger

Common Issues and Solutions

1. Model Not Downloading

• Solution: Ensure you have an active internet connection. The models are downloaded automatically at runtime if they are missing. If the download fails, you can manually download the models and place them in the models folder.

2. Tags Not Accurate

• Solution: Try using a different model or adjusting the threshold settings. Different models may perform better with different types of images.

3. Extension Not Showing Up

• Solution: Make sure the extension is installed in the correct directory (custom_nodes\ComfyUI-WD14-Tagger). Restart ComfyUI after installation.

Frequently Asked Questions

• Can I use the extension offline?
• Yes, you can use the extension offline. Simply download the models manually and place them in the models folder.
• What are the system requirements?
• The extension requires onnxruntime for CPU usage or onnxruntime-gpu for GPU usage. Note that GPU support may have issues, and the author cannot provide support for GPU-related problems.

Learn More about ComfyUI WD 1.4 Tagger

For more information, tutorials, and community support, you can visit the following resources:

• ComfyUI Examples
• WaifuDiffusion v1.4 Tags
• ComfyUI GitHub Repository These resources provide valuable insights and examples to help you get the most out of the ComfyUI-WD14-Tagger extension.