I came into some free time and finally decided to set up a training environment based on Diffusion Pipe and the WSL tutorial:

My goal is to match dtwr434's quality with some concepts that are either new or haven't been done well enough to my liking.

I have a couple Sybian LORA's in the pipeline, namely mounting, grinding and bouncing. This is the first training run that seems to have produced something functional.

Honestly I think the most annoying part of LORA training is acquiring, processing and tagging footage. Holy crap does that take a long time. With that said I don't think my tagging is good enough yet, and experimentation will be necessary.

Since it takes many hours to train anything the number of variables are large, I can't promise amazing quality anytime soon, but I'll try!

Alas! Without further ado, please check out my first ever LORA using my personalized workflow embedded in the showcase videos.