I've been playing with it a bit to understand if/how I can use it for manufacturing. It feels like out of the box it's not great for really specific stuff that might be useful industrially - think there are ways to improve it though. A very light summary below

https://www.linkedin.com/pulse/yet-another-stable-diffusion-...

This is an adaptation into the medical field (generating Medical Images) that might help further train classifiers.

https://bluch.github.io/stablecxr/