model.py

from diffusers import StableDiffusionPipeline
import torch

def load_model(model_name):
    try:
        if model_name == "Stable Diffusion":
            # Check if GPU is available, else use CPU
            if torch.cuda.is_available():
                pipe = StableDiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", torch_dtype=torch.float16)
                pipe = pipe.to("cuda")  # Use GPU
            else:
                pipe = StableDiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4")
                pipe = pipe.to("cpu")  # Use CPU if no GPU is found
            return pipe
    except Exception as e:
        print(f"Error loading model: {e}")
        return None

def generate_image(model, prompt, model_name):
    try:
        if model_name == "Stable Diffusion":
            result = model(prompt, num_inference_steps=10, height=256, width=256)  # Reduced resolution for faster generation
            image = result.images[0]
            return image
    except Exception as e:
        print(f"Error generating image: {e}")
        return None