LoRA trained with style of qiandaiyiyu. Dataset consisted on 17 images with autocaptioning enabled, captions were reasonable so I didn't touch anything more.
Training has been performed at 16 alpha, epoch 20, LR OneTrainer default.
Recommended strength: 1.0-1.3
Trigger word: qiandai_style
Description
This is the one you should use. Working perfectly with Qiandai style. Recommended str: 1.3-1.5