Reproduction of the demo

[yxsysu]

I use the diffuser and run the codes as follows:

from diffusers import StableUnCLIPImg2ImgPipeline
from diffusers.utils import load_image
import torch

pipe = StableUnCLIPImg2ImgPipeline.from_pretrained(
    "stabilityai/stable-diffusion-2-1-unclip", torch_dtype=torch.float16, variation="fp16"
)
pipe = pipe.to("cuda")

url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/stable_unclip/tarsila_do_amaral.png"
init_image = load_image(url)

images = pipe(init_image).images
images[0].save("variation_image.png")

However, I found the results are inferior to the demo: clipdrop.co/stable-diffusion-reimagine

Could you tell me how to reproduce the results of the demo? Thanks for your reply.