This is my first LoRA, and my low vram card struggles with SDXL LoRA training. For one thing, max resolution is 768x768 😐 But hey, this is just a test, and it's my very first attempt at training a LoRA. I'm planning on running several more versions on a Runpod or similar cloud computing system, and doing a lot more testing and finetuning. Wanted to just make this v1 available for other people to try.

WD14 Captioning with handmade edits all include "straitjacket" and either "arms crossed" or "crossed arms" (sorry for the inconsistency, I realized it after already training this test model. This will be fixed to be more consistent in a future iteration if this project seems to be worth pursuing). Captioning/trigger wise, straitjackets have also all been identified with multiple various descriptors such as "tight black latex straitjacket", "heavy canvas straitjacket", and "extreme straitjacket"; as well as several reference images including "D rings".

Let me know what you think, as I plan to begin making a handful of other bdsm-themed SDXL LoRAs; including tied to bed/spread eagle, ankle bondage, shibari, and more.

Thanks!

Shloren