Adventures in Imagineering — Where do the GPU-less go?

I’ve been writing a lot about Stable Diffusion (and image generation using AI) lately. That’s because providing a text prompt, getting an AI generated image back after a few seconds, and then improving upon your prompt to get the exact image you want can be so very addictive 🙂

But there’s a flip-side to the coin — those who can’t run AI systems like Stable Diffusion because their machines don’t support it. Where do they go to get some Stable Diffusion action?

In the case of Stable Diffusion, those generally left out are the ones who do not have a supported GPU on their machine. I was not able to run Stable Diffusion on my machine initially because I had an Apple Silicon device and they weren’t quite there in terms of Apple Silicon support yet. (If you are in that boat, you might want to check out this post where I explain how to get Stable Diffusion set up on an Apple Silicon machine quite easily …)

So, at first, I had to rely on Google Colab. If you aren’t familiar with Google Colab, but have worked with Jupyter Notebook, then you should be able to get up to speed fairly quickly since Google Colab is basically just a modified version of Jupyter Notebook running on Google hardware. Compared to working on my old Intel notebook (which was the only option I had due to the support issues for the Apple…