ENFUGUE is an open-source web application designed for generating studio-quality images and video using generative AI. It caters to both beginners and professionals, offering a unified interface for various AI generation tasks and aiming to simplify the workflow for users of all skill levels.

How It Works

ENFUGUE leverages a unified pipeline that seamlessly integrates text-to-image, image-to-video, and other advanced operations through a layered canvas and input roles. It extends the diffusers.StableDiffusionPipeline with additional arguments and supports TensorRT for optimized performance. The application prioritizes efficient resource management, releasing GPU memory when not in use and cleaning up unnecessary files.

Quick Start & Requirements

• Installation and execution are handled via a provided script:
  • Windows: .\enfugue.bat
  • Linux: ./enfugue.sh
• The script accepts flags for automatic installation type, updates, and optional dependency installation (e.g., MMPose).
• Access the UI via https://app.enfugue.ai:45554 or http://127.0.0.1:45555.
• Official documentation and quick-start guides are available on the project's Wiki.

Highlighted Details

• One-click installation script for simplified setup.
• Integrated CivitAI browser for easy model management.
• Optional authentication and authorization for multi-user environments.
• JSON API and Python extensions for programmatic access.
• Efficient GPU utilization and resource cleanup.
• In-app tooltips and documentation for user guidance.

Maintenance & Community

• The project encourages community engagement via a Discord server.
• Further resources and discussions are available on the project's Discussion Boards.

Licensing & Compatibility

• The README does not explicitly state a license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

• Running directly from source has only been tested on Windows (via Cygwin) and Linux, with no macOS support mentioned.
• The Wiki documentation is noted as "in progress."