extensions/modified-euler-samplers-for-sonar-diffusers/example.py

from schedulers.EulerA import EulerA
# Initialize the Celery app

controlnet = ControlNetModel.from_pretrained(
  "lllyasviel/control_v11p_sd15_openpose", 
  torch_dtype=torch.float16,
  local_files_only=True,
)
pipe = StableDiffusionControlNetPipeline.from_pretrained(
  "runwayml/stable-diffusion-v1-5",
  controlnet=controlnet,
  local_files_only=True, 
  torch_dtype=torch.float16,
  safety_checker=None, 
  requires_safety_checker=False,
).to('cuda')

# import one of the 2 schedulers from this repo
pipe.scheduler = EulerA.from_config(pipe.scheduler.config)

# choose from [0, 'rand_new', 'rand_init']
pipe.scheduler.history_d = 'rand_new'
# number should be between -1 and 1
pipe.scheduler.momentum = 0.95
# number should be between -1 and 1
pipe.scheduler.momentum_hist = 0.75

buffer = open('img0.png', 'rb')
buffer.seek(0)
image_bytes = buffer.read()
images = Image.open(BytesIO(image_bytes))

start_time = time.time()
generator = torch.manual_seed(2733424006)
image=pipe(
  "A person standing in a field of flowers, 4k, realistic",
  images,
  num_inference_steps=20,
  height=512,
  width=512,
  generator=generator
).images[0]
end_time = time.time()
execution_time = end_time - start_time
print("Execution time: {:.2f} seconds".format(execution_time))

# print(image)
image.save('img1.png', format='PNG')