lcm_with_consistency_decoder.py

from diffusers import DiffusionPipeline
import torch
from consistencydecoder import ConsistencyDecoder
from PIL import Image
import numpy as np

pipe = DiffusionPipeline.from_pretrained("SimianLuo/LCM_Dreamshaper_v7", torch_dtype=torch.float32)

decoder_consistency = ConsistencyDecoder(device="cuda:0") # Model size: 2.49 GB

# To save GPU memory, torch.float16 can be used, but it may compromise image quality.
pipe.to(torch_device="cuda", torch_dtype=torch.float32)

prompt = "Self-portrait oil painting, a beautiful cyborg with golden hair, 8k"

# Can be set to 1~50 steps. LCM support fast inference even <= 4 steps. Recommend: 1~8 steps.
num_inference_steps = 10

latent = pipe(prompt=prompt, num_inference_steps=num_inference_steps, guidance_scale=8.0, output_type="latent")

latent=latent.images[0]/0.18215
latent=latent.unsqueeze(0)

with torch.no_grad(),torch.amp.autocast("cuda"):
    consistent_latent = decoder_consistency(latent,schedule=[1.0])

image = consistent_latent[0].cpu().numpy()
image = (image + 1.0) * 127.5
image = image.clip(0, 255).astype(np.uint8)
image = Image.fromarray(image.transpose(1, 2, 0))
image.save("con.png")

new result: