Apple has released an optimization that enables the Stable Diffusion AI imager to run on an Apple Silicon processor using Core ML, Apple’s proprietary platform for machine learning models.
The optimization will allow application developers to use Apple’s Neural Engine hardware to run Stable Diffusion roughly twice as fast as previous Mac-based methods.
Released in August, the Stable Diffusion Neural Network is an open source image synthesis model that generates “digital art” from given text, like a number of other popular image generators.
In November, the app was made available for Apple devices, but users complained about slow imaging speed.
For example, on Nvidia processors, when running locally on a Windows or Linux PC, it takes about 8.7 seconds to create a 512x512px image with good resolution.
On Apple Silicon Mac without optimization, the same image will appear only after 69.8 seconds, and with optimization – in 35 seconds.
By releasing a new optimization for SD performance, Apple wants to unleash the full potential of image synthesis on its devices, as stated on the Apple Research announcement page:
“With the rise of Stable Diffusion apps, it’s important for developers to be able to leverage technology to build apps that creatives around the world can work with.”
Author:
Ekaterina Alipova
Source: RB
I am Bret Jackson, a professional journalist and author for Gadget Onus, where I specialize in writing about the gaming industry. With over 6 years of experience in my field, I have built up an extensive portfolio that ranges from reviews to interviews with top figures within the industry. My work has been featured on various news sites, providing readers with insightful analysis regarding the current state of gaming culture.
