Use at weight 0.6 - 1.
Keyword is ayasechihaya. Trained on hair_down_ac and tied_hair_ac. Add one to positive and other to negative prompt.
Outfits include (hakama, hakama_skirt) and (sweater, plaid_tie, plaid_skirt, white_shirt).
Issues
I get some pretty strange image artifacts at lower resolutions. Image generation is a lot better at at least 683x1024 px.
Her eyes are pretty unique - one black dot slightly off-center to toward the top and one white dot. It can be difficult to generate.
Pics 1-4: Anything V5