realtime security diffusion
Have you ever wanted to make realtime ai images from your security camera alerts? I did.
why
The vision was to make something that created dynamic art in an autonomous way.
I wanted to see what it would come up with on its own and to be enthralled by the randomness of it all.
the workflow
- alert image captured
- notification pushed over MQTT
- node-red listener picks up notification
- image analyzed for scene content (stable diffusion)
- rich prompt generated (Ollama)
- image generated (stable diffusion)
- image published and notification generated
- dashboard refreshes
Left: the original image
Center: cropped to 512x512px
Right: generated image
Metadata (captions): generated by stable diffusion image analysis
Metadata (prompt): the rich prompt from Ollama
