使用stable diffusion生成图片

准备

 硬件：有GPU的主机，显存6G+ 系统：windows10 或者 ubuntu 20.04 软件: anaconda

1.下载安装anaconda

 wget https://repo.anaconda.com/archive/Anaconda3-2022.10-Linux-x86_64.sh sh Anaconda3-2022.10-Linux-x86_64.sh 验证是否安装成功 conda --version //输出 conda 4.5.11

2.创建虚拟环境

 conda create -n diffusers python==3.9

3.安装需要的python库

 source activate diffusers pip install -r requirements.txt pip install torch==2.0.0+cu117 torchvision==0.15.1+cu117 torchaudio==2.0.0+cu117 -f https://download.pytorch.org/whl/torch_stable.html

requirements.txt 内容如下

 diffusers transformers accelerate ftfy spacy filelock charset-normalizer tqdm opencv-python datasets

4.python脚本

 from diffusers import StableDiffusionPipeline import torch import matplotlib.pyplot as plt import cv2 import numpy as np model_id = "runwayml/stable-diffusion-v1-5" def generateImage(pipe, prompt): image = pipe(prompt).images[0] img_bgr = cv2.cvtColor(np.asarray(image), cv2.COLOR_RGB2BGR) cv2.imshow('img_bgr', img_bgr) cv2.waitKey(1000) if __name__=='__main__': print(torch.cuda.is_available()) pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16) pipe.safety_checker = lambda images, clip_input: (images, False) pipe = pipe.to("cuda") while True: prompt= input('输入描述来生成图片:') generateImage(pipe, prompt)

保存为stable_diffusion.py并执行

 python stable_diffusion.py

效果图

codepythondiffusionpromptstablediffusionlinuxb2bnumpyhivepython库createpython脚本效果图ivaelotransformerscliptransformerhtmlwindows虚拟环境cligpupytorchurl