ImaginAIry is a Python library for generating images and videos using AI models like Stable Diffusion.
The tool addresses the need for straightforward, programmatic access to image and video generation without complex setup. It wraps Stable Diffusion and Stable Video Diffusion models, allowing users to generate both static images and animated videos through a Python interface. The library handles model loading, inference, and output formatting, with support for features like image prompts, upscaling via integrated spandrel models, and multiple video output formats including mp4, webp, and gif.
Developers working on Linux or macOS with M1 chips will find the tool reliable and straightforward to install. Video generation currently requires Nvidia GPUs and does not work on Mac or CPU. The project suits workflows where image or video generation needs to be integrated into Python applications or scripts. For Windows users, manual PyTorch installation is required before use. The library supports both standard Stable Diffusion and SDXL model weights, including those in compvis format, and allows loading models from local paths.
The project shows active maintenance with regular feature additions and bug fixes. Recent work has focused on expanding capabilities, including frame interpolation for smoother video output, image prompt support for guided generation, and integration of upscaling functionality. The maintainers have added cross-platform testing infrastructure to catch installation issues early. Documentation has been formalized with an auto-published documentation website. The codebase has undergone dependency cleanup, removing unnecessary packages and vendorizing components to reduce installation friction.
