app.py

import torch
from diffusers import StableDiffusion3Pipeline
import gradio as gr
import os
import spaces
from huggingface_hub import snapshot_download

HF_TOKEN = os.getenv("HF_TOKEN")

model_path = snapshot_download(
    repo_id="stabilityai/stable-diffusion-3-medium", 
    revision="refs/pr/26",
    repo_type="model", 
    ignore_patterns=["*.md", "*..gitattributes"],
    local_dir="stable-diffusion-3-medium",
    token=HF_TOKEN,
    )

if torch.cuda.is_available():
    device = "cuda"
    print("Using GPU")
else:
    device = "cpu"
    print("Using CPU")

# Initialize the pipeline and download the model
pipe = StableDiffusion3Pipeline.from_pretrained(model_path, torch_dtype=torch.float16)
pipe.to(device)

# Define the image generation function
@spaces.GPU(duration=60)
def generate_image(prompt, negative_prompt, num_inference_steps, height, width, guidance_scale, num_images_per_prompt):
    output = pipe(
        prompt=prompt,
        negative_prompt=negative_prompt,
        num_inference_steps=num_inference_steps,
        height=height,
        width=width,
        guidance_scale=guidance_scale,
        num_images_per_prompt=num_images_per_prompt
    ).images
    return output

# Create the Gradio interface

prompt = gr.Textbox(label="Prompt", info="Describe the image you want", placeholder="A cat...")

negative_prompt = gr.Textbox(label="Negative Prompt", info="Describe what you don't want in the image", placeholder="Ugly, bad anatomy...")

num_inference_steps = gr.Number(label="Number of Inference Steps", precision=0, value=25)

height = gr.Slider(label="Height", info="Height of the Image", minimum=256, maximum="1536", step=32, value=1024)

width = gr.Slider(label="Width", info="Width of the Image", minimum=256, maximum="1536", step=32, value=1024)

guidance_scale = gr.Number(minimum=0.1, value=7.5, label="Guidance Scale", info="The number of denoising steps of the image. More denoising steps usually lead to a higher quality image at the cost of slower inference")

num_images_per_prompt = gr.Slider(label="Number of Images to generate with the settings",minimum=1, maximum=4, step=1, value=1)

interface = gr.Interface(
    fn=generate_image,
    inputs=[prompt, negative_prompt, num_inference_steps, height, width, guidance_scale, num_images_per_prompt],
    outputs="image",
    title="Stable Diffusion 3 Medium",
    description="Made by <a href='https://linktr.ee/Nick088' target='_blank'>Nick088</a> \n Join https://discord.gg/osai to talk about Open Source AI"
)

# Launch the interface
interface.launch(share = False)