The Trigger word is cumb€7t. Somehow civitai butchered it in my image prompts.

This is my best lora so far, it can do everything! Belt, - Earrings, -chocker, -necklace, holding in hand, mouth etc. I used this tutorial here do do it:
https://youtu.be/6_PEzbPKk4g
But I did use captioning instead of sdxl keyword style.

Whats the difference between the versions?

Normal Version is mainly trained on belt and filled lying around. For Earrings, -Chocker / Necklace and holding in mouth reduce lora weight of the normal version to 0.75 or
use the less trained version. Also, the normal version tends to create faces resembling my training images.

You can try to specify things like “She is holding three filled in her left hand”
In 7 out of 10 cases Flux gets the number right.

If you have trouble getting results, Instead of “she is wearing a belt” Use things like“ she is wearing a [any color] belt with filled attached to it. Also get inspired by the example image prompts.

If you enjoyed this and want to see more, please consider supporting me on my Safe For Work (SFW) Patreon: https://patreon.com/AIKnowledgeCentral or by subscribing to my YouTube channel: https://www.youtube.com/@AIKnowledge2Go.