AUTOMATIC1111のおすすめ拡張機能９選んんんほおお！【Stable Diffusion】

The first expansion function introduced is the Image Browser, which simplifies the process of viewing output images. Instead of manually opening the Outputs Folder, users can view images directly through the web UI. This tool automatically reads and displays images, allowing for easy browsing, metadata inspection, and deletion. A unique feature is the ability to search metadata, enabling users to filter images based on specific prompts, model names, sampling methods, or the name of the expansion function used. Additionally, there's a ranking function that lets users sort images by their assigned rank, which is particularly useful after mass production with the Generate Forever feature. The Image Browser can be installed as an extension and is considered an essential tool for its convenience.

The second feature is Tag Complete, which assists users in inputting prompts by suggesting Danbooru-tags, commonly used on image posting sites like Danbooru. This auto-complete function not only proposes tags as users type but also corrects typos and offers related word suggestions. Each tag is accompanied by a hit count, indicating its popularity and the likelihood of obtaining desired results with that tag. However, some models may not respond well to these tags, so it's advised to use it as an auxiliary input tool. The default number of suggestions is five, but it's recommended to increase this number to about ten for a better experience.

System Info is the third function, aimed at users interested in the system's current state. It provides real-time information about the machine, which is beneficial for monitoring VRAM consumption during image generation. The tool also lists recognized models and learning files, eliminating the need to manually search for them in folders. While not an extension that performs actions, System Info is appreciated for consolidating scattered information into one view.

The fourth tool is the Style Editor, which simplifies the management of saved prompt templates. While users can save prompts as templates using the floppy mark, managing an increasing number of templates can be cumbersome. The Style.csv file, where saved styles are stored, can be edited directly through the web UI with the Style Editor extension. This allows for easy viewing, editing, and deletion of registered styles in real time, streamlining the process of managing and organizing frequently used prompts.

The fifth feature is the Aspect Ratio Selector, designed to help users generate images of a specific size while maintaining the aspect ratio. It adds an icon to the interface with preset buttons for common ratios like 1:1 and 16:9, and a template button to reset to the default size of 512. It also includes a calculator to determine the necessary aspect ratio adjustments when changing resolution settings. This tool simplifies the process of resizing images without distorting their proportions.

The sixth tool, Aspect Ratio Helper, complements the slide bar functionality, providing a preset aspect ratio that maintains the ratio as the user adjusts the bar. It offers a different experience from the selector and adds a small command next to the resolution setting for easy ratio adjustments.

The seventh feature, Config Preset, allows users to save not only their prompt settings but also their sampling and high-resolution settings, which are not saved by default. It adds a pull-down menu under the output results of T2i and I2i, where users can save, edit, and delete presets, making it easy to recall custom settings with each startup.

The eighth feature is After Detailer, a well-known extension that automatically detects and enhances the facial details of illustrations. It works with T2i and I2i and can be used by simply enabling it in the settings. After Detailer can also accommodate prompts for creating variations in facial expressions, offering a simple yet powerful tool for refining illustrations.

The ninth and final feature is InPaint Anything, which utilizes an old Facebook object detection model to allow users to modify illustrations and create masks using Segment Anything. It offers a choice of models based on quality and speed, and once a model is selected and the image is processed, users can segment the image, create masks for specific objects, and apply various processing methods, including InPaint, Cleaner, Control Net, and Mask Only. This feature provides a straightforward way to modify specific parts of an image, such as changing the color of an object or creating a material image from a segment.

In addition to the main features, the Mask Only function allows users to extract only the segment mask, which can be useful for collecting materials for collages or extracting parts of an image. The video concludes by encouraging viewers to make good use of these nine expansions for efficient and convenient operation, and reminds them to stay hydrated.