Trigger: ais-bulbz
Strength: between 0.5 and 1.2 depending on prompt - 1+ will likely affect the quality of the image.
This is the single most frustrating LoRA I've used, let alone made. It is capable of interesting images with fun juxtapositions but you do need to play with the prompt and weights to achieve something great.
It was trained on 'ais-bulbz, [object/scene] inside the bulb, [object/scene] outside the bulb'. And it can do this, but making the strength too high makes the image rubbish. At lower strengths it will try to take a bit of the inside out and a bit to the outside in...
If you know how I can make this work how I want it to or to make the images reliably better at higher strengths, please let me know. I'll send you a big chunk of buzz in return!