Core Features of sd-webui-infinite-image-browsing

Infinite Image Browsing (IIB) is a robust and highly performant image, video, and audio browser designed for Stable Diffusion web UI, ComfyUI, Fooocus, NovelAI, StableSwarmUI, and Invoke.AI, with standalone operation capabilities. It offers a rich set of features to efficiently manage and explore your generated media content.

Image Search & Favorite

IIB transforms the prompt, model, Lora, and other generation information embedded in your images into searchable tags. This allows for highly precise searches within your extensive media library.

• Tag-based Search: Automatically extracts key information (like prompts, models, Lora usage) and converts it into tags. These tags are then sorted by frequency of use, making it easy to find common themes or specific generation parameters.
• Autocomplete and Translation: The search bar offers tag autocomplete suggestions, and for multi-language users, it supports auto-translation of tags to bridge language barriers.
• Advanced Search: Beyond simple keywords, you can perform advanced searches similar to Google, enabling complex queries.
• Fuzzy Search: For less precise queries, you can use fuzzy search to find images based on partial filenames or fragments of generated information.
• Image Favorite: Easily mark images as favorites by toggling custom tags directly from the right-click menu. These custom tags provide a flexible way to categorize and quickly retrieve your preferred content.
• Custom Search Paths: Users can add custom folders to their search paths, simplifying the management of user-created directories.

View Images/Videos & Send To

IIB provides comprehensive viewing capabilities for both images and videos, coupled with seamless integration with other AI tools.

• Generation Information Display: View detailed image generation information (prompts, parameters, etc.) directly on the image grid or in a dedicated full-screen preview mode. This information is dynamically displayed, and you can even drag, resize, or collapse the info panel.
• Full-Screen Preview: Enter a full-screen preview mode for an immersive viewing experience. While in this mode, you can still access image generation information and perform context menu operations.
• Navigation: Easily navigate between images in full-screen preview using arrow keys or dedicated buttons, allowing for quick browsing of previous and next items.
• Video Playback: Play video files directly from a remote server, offering a unified browsing experience for all media types.
• Send To Integrations: Send images to various other tabs and third-party extensions within your AI ecosystem, such as:
  • SD-webui Tabs: Send to txt2img, img2img, inpaint, or extras tabs for further processing.
  • Third-party Extensions: Direct integration with tools like ControlNet (for both txt2img and img2img workflows) and openOutpaint.

Walk Mode

Designed for seamless exploration of deeply nested folder structures, Walk Mode flattens your content for continuous browsing.

• Automatic Folder Loading: This mode automatically loads content from subsequent folders (mimicking os.walk), allowing you to scroll indefinitely through all images without needing to manually navigate in and out of directories.
• Flattened View: When enabled, Walk Mode presents all images from subfolders in a single, continuous stream, eliminating the tedious task of repeatedly entering and exiting folders.
• Performance: Tested to function correctly and efficiently even with large collections, handling over 27,000 files without issues.

File Tree Structure & File Operations

IIB offers an intuitive file management interface, combining traditional file tree navigation with modern batch processing capabilities.

• File Tree-Based Preview: Browse your files and folders using a familiar tree structure, providing a clear overview of your directory hierarchy.
• Automatic Refreshing: The file view automatically refreshes to reflect any changes made to the file system.
• Basic File Operations: Perform essential file management tasks directly within the browser:
  • Delete: Remove unwanted files or empty folders.
  • Move/Copy: Organize your files by moving or copying them to different locations.
  • Create New Folders: Easily establish new directories as needed.
• Multi-Selection: Select multiple items by holding down Ctrl, Shift, or Cmd keys. This enables batch operations for:
  • Deleting multiple files.
  • Moving or copying selected items to another folder.
  • Adding or removing custom tags for a collection of images.
  • Drag-and-drop functionality for selected items.
• "Keep Multi-Select": A dedicated button allows you to maintain the multi-selection state across various actions, facilitating complex batch operations on a consistent set of files.

Image Comparison (similar to Imgsli)

For those who need to compare different generations, IIB includes a side-by-side comparison tool.

• Side-by-Side Viewing: Display two images next to each other for easy visual comparison.
• Generation Info Comparison: Simultaneously view and compare the generation parameters (prompts, models, settings) of both images, highlighting differences that might explain visual variations.

Batch Download

Streamline the process of downloading multiple images or media files from your browsing sessions.

• Multi-Source Integration: Compile a list of files for batch download from various views, including search results, regular image grids, or Walk Mode.
• Flexible Addition: Add images to the processing list through convenient drag-and-drop gestures or by using the "Send To Batch Download" option available in the right-click menu.
• Packaging Options: Download the selected files as a single archive, with options to compress the package for efficiency.

Natural Language Categorization & Search (Experimental)

This experimental feature leverages AI models to understand the semantic content of your images, offering new ways to organize and discover your artwork.

• Semantic Grouping: Images are automatically grouped based on the semantic similarity of their prompts, allowing for intuitive categorization without manual tagging.
• Natural-Language Retrieval: Search your image library using full sentences or descriptive phrases, similar to how Retrieval-Augmented Generation (RAG) systems work. The AI interprets your query to find relevant images.
• Usage Flow:
  1. Select Scope: Choose one or more folders from your QuickMovePaths to define the pool of images for analysis.
  2. Categorize: Click "Refresh" to initiate the process. The system will generate "topic cards" that represent the major themes found within the selected scope. This involves:
     • Prompt Extraction & Normalization: Reading image metadata and cleaning prompts to focus on core semantic elements.
     • Embeddings: Converting these cleaned prompts into numerical vectors using an OpenAI-compatible embedding model.
     • Clustering: Grouping similar vectors (and thus similar image concepts) into clusters.
     • Title Generation: Using a large language model (LLM) to generate concise, human-readable titles and keywords for each cluster.
  3. Search: Input a natural-language query and click "Search". The system will retrieve and display images semantically similar to your query.
• Caching & Incremental Updates: To improve performance and reduce costs, the system caches generated embeddings and topic titles. It intelligently re-vectors or re-generates titles only when source data or rules change.
• Configuration: This feature requires an OpenAI-compatible API endpoint and API key, configurable via environment variables (OPENAI_BASE_URL, OPENAI_API_KEY, EMBEDDING_MODEL, AI_MODEL).