AI人工智慧python試作Keras CV實作之Stable Diffusion

iPython code在此

print('試試stable diffusion')
print('參閱https://keras.io/api/keras_cv/models/stable_diffusion/')

Linux指令

!cat /etc/os-release

!nvidia-smi

!python --version

安裝keras-cv，匯入套件

!pip install --upgrade keras-cv

from keras_cv.models import StableDiffusion
from PIL import Image

產生假圖

model = StableDiffusion(img_height=512, img_width=512 ,jit_compile=True)

img = model.text_to_image(
    prompt="cute magical flying black shiba inu, fantasy art,"
    "high quality, highly detailed, elegant, sharp focus,"
    "concept art, character concepts, digital painting, mystery, adventure",
    batch_size=3,  # How many images to generate at once
    num_steps=25,  # Number of iterations (controls image quality)
  #  seed=123,  # Set this to always get the same image from the same prompt
)

轉成圖檔，秀圖，下載假圖

for i in range(3):
  Image.fromarray(img[i]).save(str(i)+"shiba.png")
  print(f"saved at {i}shiba.png")

from google.colab.patches import cv2_imshow
import cv2

for i in range(3):
  im=cv2.imread(str(i)+'shiba.png')
  cv2_imshow(im)

from google.colab import files
for i in range(3):
  files.download(str(i)+'shiba.png')