A quick look under the hood of Stable Diffusion Open Source architecture.

Analyzing the model that is powering a new wave of generation models.

If you haven’t lived under a rock for the past year, you’ve probably heard that the text-to-image generation space is undergoing a massive revolution. Last week, an AI startup called Stability AI unveiled its first version of its Stable Diffusion text-to-image synthesis model.

The good (I could say great) news is that they released it as an open-source model for free. That is significant!

And being an open-source model means that, if you have a sufficiently powerful graphics card, you can download and run the model on your computer ( I did it and even already built a full creative project on it.. you can read about it here).

Because it is an open-source model, we can use it for non-commercial purposes and commercial under the terms of the license called Creative ML OpenRAIL-M — which is fine enough to impose some usage restrictions, such as not using it to break applicable laws, generating false information, discriminate against individuals, or provide medical advice.

I’ve seen an explosion of innovation in the last few days around what people can do with Stable Diffusion, which matches the quality of…