Diffusers

Diffusers is a native Mac app to generate images from a text description. It uses state-of-the-art models contributed by the community to the Hugging Face Hub, optimized and converted to Core ML for maximum performance. It is the Mac counterpart to the Diffusers open source library.

Main Features:

• Privacy: Models are downloaded on demand and cached in the Mac's hard drive. No credits for online services are needed, and there are no long queues. Prompts and images never leave the computer unless chosen to share.
• Performance: Leverages a native Swift app designed for macOS; Core ML models optimized for CPU, GPU, and Apple Neural Engine; Diffusers library ported to Swift for maximum performance.
• Model Selection: Launches with 5 models: Stable Diffusion 1.4 (the original model), Stable Diffusion 1.5 (trained on additional images with a focus on aesthetics), Stable Diffusion 2 (improved and heavily retrained), Stable Diffusion 2.1 (works great with negative prompts), and OFA small v0 (a distilled model half the size, running faster and requiring less RAM).
• Generation Settings: Allows tweaking of the model, prompt and negative prompt (what you don't want), guidance scale (how much the image should resemble the prompt, good values between 7 and 10), seed (to reproduce a previous generation), and steps (how many diffusion steps, default is 25).
• What's New (Version 1.5): Support for Stable Diffusion 3 Medium; new models like SDXL and SDXL with refiner step for large resolution and high quality generation; quantized and palettized versions of popular models for faster creation and less memory usage.

Pricing & Open Source: The app is completely free. Diffusers is fully open source with a permissive license allowing users to build their own products. App size is 1.6 MB and requires macOS 13.1 or later.