app.py

import gradio as gr
import torch
from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
import os
from huggingface_hub import HfApi, login

token = os.getenv("HF_TOKEN")
login(token=token)  # Logs in with the token in Hugging Face Spaces

# Load Stable Diffusion model and ControlNet reference-only model
model_id = "stabilityai/stable-diffusion-3.5-large-turbo"
controlnet_id = "lllyasviel/control_v11p_sd15_inpaint"  # Use an appropriate ControlNet variant

controlnet = ControlNetModel.from_pretrained(controlnet_id, torch_dtype=torch.float32)
pipeline = StableDiffusionControlNetPipeline.from_pretrained(
    model_id,
    controlnet=controlnet,
    torch_dtype=torch.float32
)

# Define the Gradio interface function
def generate_image(prompt, reference_image):
    # Process reference image
    reference_image = reference_image.resize((512, 512))
    # Generate image with reference-only style transfer
    generated_image = pipeline(
        prompt=prompt,
        image=reference_image,
        controlnet_conditioning_scale=1.0,
        guidance_scale=7.5,
        num_inference_steps=50
    ).images[0]
    return generated_image

# Set up Gradio interface
interface = gr.Interface(
    fn=generate_image,
    inputs=[
        gr.Textbox(label="Prompt"),
        gr.Image(type="pil", label="Reference Image (Style)")
    ],
    outputs="image",
    title="Image Generation with Reference-Only Style Transfer",
    description="Generate an image based on a text prompt and style reference image using Stable Diffusion 3.5 with ControlNet (reference-only mode)."
)

# Launch the Gradio interface
interface.launch()