app.py

import streamlit as st
import torch
from diffusers import StableDiffusionXLPipeline, UNet2DConditionModel, EulerDiscreteScheduler
from huggingface_hub import hf_hub_download
from safetensors.torch import load_file

# Define a function to generate the image
def generate_image(prompt, num_inference_steps):
    base = "stabilityai/stable-diffusion-xl-base-1.0"
    repo = "ByteDance/SDXL-Lightning"
    ckpt = "sdxl_lightning_2step_unet.safetensors" # Use the correct ckpt for your step setting!

    # Load model.
    unet = UNet2DConditionModel.from_config(base, subfolder="unet").to(torch.device("cpu"), torch.float16)
    unet.load_state_dict(load_file(hf_hub_download(repo, ckpt), device=torch.device("cpu")))
    pipe = StableDiffusionXLPipeline.from_pretrained(base, unet=unet, torch_dtype=torch.float16, variant="fp16").to(torch.device("cpu"))

    # Ensure sampler uses "trailing" timesteps.
    pipe.scheduler = EulerDiscreteScheduler.from_config(pipe.scheduler.config, timestep_spacing="trailing")

    # Generate image
    image = pipe(prompt, num_inference_steps=num_inference_steps, guidance_scale=0).images[0]

    return image

# Main function for Streamlit app
def main():
    st.title("AI Image Generator")

    # Input fields
    prompt = st.text_input("Enter prompt")
    num_inference_steps = st.slider("Number of Inference Steps", min_value=1, max_value=10, value=2)

    if st.button("Generate Image"):
        # Check if prompt is provided
        if prompt:
            # Generate image
            generated_image = generate_image(prompt, num_inference_steps)
            # Save image
            generated_image.save("output.png")
            # Display image
            st.image(generated_image, caption='Generated Image', use_column_width=True)
        else:
            st.error("Please enter a prompt.")

if __name__ == "__main__":
    main()