Gals in skirts, mainly plaid but also some plain pencil skirts too.
v0.5 - New training method used, model likely overtrained, use low weight.
First attempt at training with new method so not 100% on all the quirks. I think it's over trained rather than unreliable, recommend including on lower strength in combination with other LORAs and tweaking from there.
Starting Settings
Type: SDXL, Illustrious
Model: iLustMix
Steps: 18-40
CFG: 3.5-12.0
Sampler: Euler
Schedule: Normal
DISCLAIMER: All training data and datasets are synthetic. I have almost no idea what I'm doing with making LORAs, it's 90% trial and error, as such I make no promises about the results they return and warn that all may return unexpected results, all my models may return NSFW results. No likeness to real persons were used in training or are intended.