app.py

import gradio as gr
import spaces
from diffusers import StableDiffusionInstructPix2PixPipeline
from loguru import logger
from PIL import Image

pipe = StableDiffusionInstructPix2PixPipeline.from_pretrained(
    "timbrooks/instruct-pix2pix"
).to("cuda")


@logger.catch(reraise=True)
@spaces.GPU(duration=180)
def generate(
    prompt: str,
    init_image: Image.Image,
    progress=gr.Progress(),
):
    logger.info(f"Starting image generation: {dict(prompt=prompt, image=init_image)}")

    # Downscale the image
    init_image.thumbnail((1024, 1024))

    logger.debug(f"Generating image: {dict(prompt=prompt)}")

    def progress_callback(pipe, step_index, timestep, callback_kwargs):
        logger.trace(
            f"Callback: {dict(num_timesteps=pipe.num_timesteps, step_index=step_index, timestep=timestep)}"
        )
        progress((step_index + 1, pipe.num_timesteps))
        return callback_kwargs

    images = pipe(
        prompt=prompt,
        image=init_image,
        callback_on_step_end=progress_callback,
    ).images
    return images[0]


demo = gr.Interface(
    fn=generate,
    inputs=[
        gr.Text(label="Prompt"),
        gr.Image(label="Init image", type="pil"),
    ],
    outputs=[gr.Image(label="Output")],
)

demo.launch()