git clone https://www.modelscope.cn/caai/caai-stable-diffusion-lora.git
import cv2
from modelscope.pipelines import pipeline
pipe = pipeline('efficient-diffusion-tuning', 'caai/caai-stable-diffusion-lora')
inputs = {'prompt': 'a painting of a road with houses and a church in the background'}
outputs = pipe(inputs)
cv2.imwrite('result.png', outputs['output_imgs'][0])
这是在Stable Diffusion的几个不同风格大模型基础上训练的LoRA模型,包括3D风格,watercolor风格,sketch风格,flat illustration风格,油画风格以及日漫风格