Version 1.5 is probably its final form. Drawing on some of what I developed in the Eldritch Candid LoRA, I put together a new database of images and processed them in a similar manner, though with a lighter touch. The result is even more flexible and a bit more analog. unlike with 1.2, all these sample images were just base SDXL with LoRA and no refiner. You can naturally derive more specific aesthetics with the refiner or different fine tuned models.

You don't necessarily need a specific keyword. But it likes any of the following:

Oscilloscope

Vacuum tube

Glowing electronics

brass gears

switches

knobs

dials

control panel