Diffusers 0.3 adds image-to-image, inpainting, ONNX export, and Mac MPS support for Stable Diffusion

AI Impact Summary

Diffusers 0.3 expands the diffusion toolkit with an image-to-image pipeline, inpainting, and an experimental ONNX exporter/pipeline, plus RAM optimization via attention slicing and Mac hardware support. The release also broadens concepts and tools, including Textual Inversion, Diffusers Interpret, and new community models like Japanese Stable Diffusion and Waifu Diffusion. Together, these changes lower hardware barriers and enable faster prototyping and broader deployment of Stable Diffusion-based applications.