app.py

import gradio as gr
import torch
from diffusers import StableDiffusion3Pipeline
import os

# Retrieve the API token from the environment variable
huggingface_token = os.getenv("HUGGINGFACE_TOKEN")
if huggingface_token is None:
    raise ValueError("HUGGINGFACE_TOKEN environment variable is not set.")

# Check if CUDA is available
device = "cuda" if torch.cuda.is_available() else "cpu"

# Load the Stable Diffusion model
repo = "stabilityai/stable-diffusion-3-medium-diffusers"
image_gen = StableDiffusion3Pipeline.from_pretrained(repo, text_encoder_3=None, tokenizer_3=None, use_auth_token=huggingface_token)
image_gen = image_gen.to(device)

def generate_image(prompt, num_inference_steps=50, guidance_scale=7.5):
    # Generate the image
    result = image_gen(
        prompt=prompt,
        num_inference_steps=num_inference_steps,
        guidance_scale=guidance_scale,
        negative_prompt="blurred, ugly, watermark, low resolution, blurry",
        height=512,
        width=512
    )
    # Get the generated image
    image = result.images[0]
    return image

# Create the Gradio interface
iface = gr.Interface(
    fn=generate_image,
    inputs=[
        gr.Textbox(label="Enter a prompt"),
        gr.Slider(label="Number of inference steps", minimum=1, maximum=100, value=50),
        gr.Slider(label="Guidance scale", minimum=1.0, maximum=20.0, value=7.5)
    ],
    outputs=gr.Image(label="Generated Image"),
    title="Stable Diffusion Image Generator",
    description="Enter a prompt to generate an image using the Stable Diffusion model."
)

# Launch the Gradio app
iface.launch()