from diffusers import StableDiffusionPipeline, EulerDiscreteScheduler
import torch
import numpy as np
import gradio as gr

model_dict = {
    'Stable Diffusion 1.4': "CompVis/stable-diffusion-v1-4",
    'Stable Diffusion 2.1': "stabilityai/stable-diffusion-2-1",
}

model_num_of_layers = {
    'Stable Diffusion 1.4': 12,
    'Stable Diffusion 2.1': 22,
}


# global variable
device = "cuda" if torch.cuda.is_available() else "cpu"
dtype = torch.float16 if str(device).__contains__("cuda") else torch.float32



def get_images(prompt, skip_layers, model, seed):
    model_name = model_dict[model]
    pipeline = StableDiffusionPipeline.from_pretrained(
        model_name,
        torch_dtype=dtype,
        variant="fp16",
        add_watermarker=False,
    )
    # Move the pipeline to the device
    pipeline.to(device)
    pipeline.scheduler = EulerDiscreteScheduler.from_config(pipeline.scheduler.config)
    print('inside get images')
    layer = model_num_of_layers[model] - skip_layers
    gr.Info(f"Generating an from image layer number {layer}")
    print(f'skipping {skip_layers}')
    pipeline_output = pipeline(prompt, clip_skip=skip_layers, num_images_per_prompt=1, return_tensors=False, seed=seed)
    print('after pipeline')
    images = pipeline_output.images
    print('got images')
    return images