Image Generation

[kbuechl]

Putting this as an issue todo later after I get a gpu since this will be hard without it.

I think it would be fun to let hammy create art with a keyword.
Looking around I found stable-diffusion-prompt-generator which we can host in ollama. When the user gives us a prompt to generate, we can run it through this first to enrich it (maybe this should be adjustable in settings) and then feed that to a hosted stable-diffusion model (that does not run in ollama, would be a separate docker image).