SimianLuo-LCM_Dreamshaper_v7-ov

SimianLuo/LCM_Dreamshaper_v7 for OpenVINO

This is the SimianLuo/LCM_Dreamshaper_v7 model converted to OpenVINO for accelerated inference.

Requirements: pip install optimum[openvino,diffusers]

from optimum.intel import OVLatentConsistencyModelPipeline
import time

device = "GPU"  # for best results, use "GPU" for devices with Intel iGPU or dGPU. "CPU" can also be used.
model_id = "helenai/SimianLuo-LCM_Dreamshaper_v7-ov"
pipeline = OVLatentConsistencyModelPipeline.from_pretrained(model_id, device=device, compile=False)
pipeline.reshape(1,512,512,1)  # reshape to static shapes for better performance
pipeline.compile()
prompt = "a robot discovering nature"

start = time.perf_counter()
image = pipeline(prompt, num_inference_steps=4, guidance_scale=8.0).images[0]
end = time.perf_counter()
print(f"Duration: {end-start:.2f}")
image.save("robot.png")