MochiDiffusion/MochiDiffusion

What it solves

Mochi Diffusion is a native macOS app that allows users to generate images locally and offline. It eliminates the need for cloud-based services by leveraging Apple's hardware acceleration to run Stable Diffusion and FLUX.2 Klein models directly on a Mac.

How it works

The app uses two primary pipelines for image generation:

• Core ML Stable Diffusion: It utilizes Apple's Core ML implementation to optimize models for Apple Silicon, enabling high performance and reduced memory usage via the Neural Engine or GPU.
• FLUX.2 Klein: It integrates the Iris pipeline to run FLUX.2 models with native Metal GPU acceleration.

Who it’s for

It is designed for macOS users (Apple Silicon M1 and later) who want to generate AI images privately and efficiently without relying on the cloud.

Highlights

• Local Generation: All processing happens on-device, ensuring complete privacy and no data collection.
• Hardware Optimization: Specifically tuned for Apple Silicon's Neural Engine and GPU for maximum speed.
• Metadata Preservation: Saves images as PNGs with generation settings compatible with AUTOMATIC1111 and Civitai.
• Built-in Gallery: Includes a native SwiftUI gallery for managing and syncing generated images.