Stumbled across a post on xiaohongshu (Rednote, in english (not a direct translation)) and I figured, sure, I'll make a lora out of that. More ai re-feeding.
Trained it with civitai's new experimental trainer. If, for whatever reason, it doesn't show up as grokbot, it will go under "TXQS51HB0FYSWZBSR1Z8ZYCD00"
Thanks civitai, nothing beats privacy like that...
The concept lora doesn't exactly like detailed prompts.
If you exclude the peeking prompt, you can utilize it for characters.
Not particularly sure who came up with the idea first, I assume Yayoi, since that fellow posted 2 hours before Multi's first grokbot picture.
To my surprise, there's already one, for Anima.
Description
Including "peeking into the frame from the bottom corner" changes it slightly, hence you can add it or not.





