In the last year, several machine learning models have become available to the public to generate images from textual descriptions. This has been an interesting development in the AI space. However, most of these models have remained closed source for valid ethic reasons. Because of this, even though you can interact with them via some interface, you are limited in the number of things you can test. Until now… The latest of these models is Stable Diffusion, which is an open machine learning model developed by Stability AI to generate digital images from natural language descriptions. This model has become quite popular, mainly because it was the first one to be open-sourced. I’ve already played around with Dall-E and Midjourney, but wanted to try to run a model locally and get more freedom to play around with things. I was able to successfully install and run the model on my M1 Pro and on my Windows desktop. This guide details the steps I’ve followed to get everything working on my Mac. Initial Notes A couple of notes before we get to it. I tried several guides online, and I was unable to get a smooth experience with any of them. I had to try numerous repos, solutions, etc. The main goal of this guide is to provide instructions for how to run Stable Diffusion on an M1, which is the one where I found more challenges. Installing on Windows was way more straightforward. Having said this, the repo I ended up going with has detailed guides for all platforms: Windows, Mac, and Linux. Please don't hesitate to use any of those if you are using another platform or if this guide doesn’t work for you on a Mac.

Read the full article →