LoRA in the style of Chinese actresses around the 1960s, based on SDXL1.0 training.
Trigger word: cn1960s
Training materials: 36 old photos of actresses, after simple processing, the borders and texts were removed, and the pixels were adjusted to 1024*1024.
The generalization performance is average. If the generated pixels are adjusted to 1024*1536, the performance is not very good.
60年代左右中国女星LoRA,基于SDXL1.0训练。
触发词:cn1960s
训练素材:36张女星老照片,经过简单处理,去除了边框和文字,调整了像素。
基础模型:sd_xl_base_1.0_0.9vae.safetensors
泛化性能一般,若把生成像素调整为1024*1536,表现不太好。