>>35834011uhhh this
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Textual-InversionIn practice it gives you a "supertoken" for your prompts that invokes whatever that embedding was trained on. I just browsed danbooru for good Reines against plain backgrounds and cropped and resized them to 768x768 and now the AI is doing stuff, hopefully the end result is a file you can put in your embeddings folder and use to generate spicy Reines. Picrel, preview at 34k steps of training for the prompt 1girl,(pavolia reine),virtual youtuber,silver hair,aqua eyes,blue eyes,two tone eyes,white pupils,reinembed
