Based on my Spirals of light dataset. I forgot to enable training text encoder, so this version just slaps bright spriraly lighty thingies on top of everything indiscriminately. The result turned out quite interesting, because now lora adds spiraly details on everything, basically enchancing images. If your prompt has light and glowing efects like spirals, magic, fluids or even absurdly long hair it just enchances it.
Big downside - it completely destroys monotone colors and darkness. See comparisons for reference, all prompts included.
No trigger word needed.
Fun fact: with my detailer at 1.0 an this lora at 1.2 NoobAI vpred base with realistic tag converges to some almost realistic imagery for whatever reason:
Description
No trigger