One of these generators is Stable Diffusion, which has the particularity of being able to be installed and run locally on a Mac. This has not gone unnoticed at Apple offices, and the company’s developers have decided to Optimize iOS and macOS so you can run Stable Diffusion more efficiently. Especially on Macs with M1 or M2 chips.
Apple Silicon, Core ML and great potential
Apple has officially made the announcement in its Machine Learning section, defending the local execution of Stable Diffusion as a more private way to experiment with this generator and that allows us to save the costs of hiring a server.
We will notice the optimization on iOS 16.2 and macOS 13.1, thanks to some changes that Apple developers have applied to the CoreML engine. This should reduce the imaging wait time, which is usually quite long (I’ve tried imaging with my basic MacBook Air M2 and the wait was about 6-8 minutes).
To improve these times, we will have to wait for these updates, which we should have ready in time for this Christmas. That’s it, two weeks at the most. It will be interesting to see how far those times improve: generating images using artificial intelligence is a very intense process and Apple Silicon can benefit greatly.