| from diffusers import AutoPipelineForText2Image | |
| import torch | |
| pipeline = AutoPipelineForText2Image.from_pretrained( | |
| "runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16, variant="fp16" | |
| ).to("cuda") | |
| image = pipeline( | |
| "stained glass of darth vader, backlight, centered composition, masterpiece, photorealistic, 8k" | |
| ).images[0] | |
| image |