app.py

from diffusers import StableDiffusionXLPipeline, DDIMScheduler
import torch
import gradio as gr
import inversion
import numpy as np
from PIL import Image
import sa_handler

device = "cuda" if torch.cuda.is_available() else "cpu"
scheduler = DDIMScheduler(beta_start=0.00085, beta_end=0.012, beta_schedule="scaled_linear", clip_sample=False, set_alpha_to_one=False)
pipeline = StableDiffusionXLPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16, variant="fp16", use_safetensors=True, scheduler=scheduler).to(device)

def run(image, src_style, src_prompt, prompts, shared_score_shift, shared_score_scale, guidance_scale, num_inference_steps, seed, large=True):
    prompts = prompts.splitlines()
    dim, d = (1024, 128) if large else (512, 64)
    image = image.resize((dim, dim))
    x0 = np.array(image)
    zts = inversion.ddim_inversion(pipeline, x0, src_prompt, num_inference_steps, 2)
    offset = min(5, len(zts) - 1)
    prompts.insert(0, src_prompt)
    shared_score_shift = np.log(shared_score_shift)
    handler = sa_handler.Handler(pipeline)
    sa_args = sa_handler.StyleAlignedArgs(
        share_group_norm=True, share_layer_norm=True, share_attention=True,
        adain_queries=True, adain_keys=True, adain_values=False,
        shared_score_shift=shared_score_shift, shared_score_scale=shared_score_scale,)
    handler.register(sa_args)
    for i in range(1, len(prompts)):
        prompts[i] = f'{prompts[i]}, {src_style}.'
    zT, inversion_callback = inversion.make_inversion_callback(zts, offset=offset)
    g_cpu = torch.Generator(device='cpu')
    
    if seed > 0:
        g_cpu.manual_seed(seed)
    latents = torch.randn(len(prompts), 4, d, d, device='cpu', generator=g_cpu, dtype=pipeline.unet.dtype,).to(device)
    latents[0] = zT
    images_a = pipeline(prompts, latents=latents, callback_on_step_end=inversion_callback, num_inference_steps=num_inference_steps, guidance_scale=guidance_scale).images
    handler.remove()
    torch.cuda.empty_cache()
    return images_a

with gr.Blocks() as demo:
    gr.Markdown("""# Welcome to🌟Tonic's🤵🏻Style📐Align 
Here you can generate images with a style from a reference image using [transfer style from sdxl](https://huggingface.co/docs/diffusers/main/en/using-diffusers/sdxl). Add a reference picture, describe the style and add prompts to generate images in that style. It's the most interesting with your own art! You can also use [stabilityai/stable-diffusion-xl-base-1.0] by cloning this space. 🧬🔬🔍 Simply click here: <a style="display:inline-block" href="https://huggingface.co/spaces/Tonic1/TonicsStyleAlign?duplicate=true"><img src="https://img.shields.io/badge/-Duplicate%20Space-blue?labelColor=white&style=flat&logo=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAP5JREFUOE+lk7FqAkEURY+ltunEgFXS2sZGIbXfEPdLlnxJyDdYB62sbbUKpLbVNhyYFzbrrA74YJlh9r079973psed0cvUD4A+4HoCjsA85X0Dfn/RBLBgBDxnQPfAEJgBY+A9gALA4tcbamSzS4xq4FOQAJgCDwV2CPKV8tZAJcAjMMkUe1vX+U+SMhfAJEHasQIWmXNN3abzDwHUrgcRGmYcgKe0bxrblHEB4E/pndMazNpSZGcsZdBlYJcEL9Afo75molJyM2FxmPgmgPqlWNLGfwZGG6UiyEvLzHYDmoPkDDiNm9JR9uboiONcBXrpY1qmgs21x1QwyZcpvxt9NS09PlsPAAAAAElFTkSuQmCC&logoWidth=14" alt="Duplicate Space"></a></h3> 
Join us : 🌟TeamTonic🌟 is always making cool demos! Join our active builder's🛠️community on 👻Discord: [Discord](https://discord.gg/nXx5wbX9) On 🤗Huggingface: [TeamTonic](https://huggingface.co/TeamTonic) & [MultiTransformer](https://huggingface.co/MultiTransformer) On 🌐Github: [Polytonic](https://github.com/tonic-ai) & contribute to 🌟 [PolyGPT](https://github.com/tonic-ai/polygpt-alpha)
""")
    with gr.Row():
        image_input = gr.Image(label="Reference image", type="pil")
    with gr.Row():
        style_input = gr.Textbox(label="Describe the reference style")
        image_desc_input = gr.Textbox(label="Describe the reference image")
        prompts_input = gr.Textbox(label="Prompts to generate images (separate with new lines)", lines=5)
    with gr.Accordion(label="Advanced Settings"):
        with gr.Row():
            shared_score_shift_input = gr.Slider(value=1.5, label="shared_score_shift", minimum=1.0, maximum=2.0, step=0.05)
            shared_score_scale_input = gr.Slider(value=0.5, label="shared_score_scale", minimum=0.0, maximum=1.0, step=0.05)
            guidance_scale_input = gr.Slider(value=10.0, label="guidance_scale", minimum=5.0, maximum=20.0, step=1)
            num_inference_steps_input = gr.Slider(value=12, label="num_inference_steps", minimum=12, maximum=300, step=1)
            seed_input = gr.Slider(value=0, label="seed", minimum=0, maximum=1000000, step=42)
    with gr.Row():
        run_button = gr.Button("Generate Images")
    with gr.Row():
        output_gallery = gr.Gallery()

    run_button.click(
        run,
        inputs=[image_input, style_input, image_desc_input, prompts_input, shared_score_shift_input, shared_score_scale_input, guidance_scale_input, num_inference_steps_input, seed_input],
        outputs=output_gallery
    )
    examples = [
        ["./download%20(8).jpg", "picasso blue period", "a portrait of a man playing guitar", 
         "an astronaut holding a cocktail glass\nan astronaut in space holding a laptop\nan astronaut in space with an explosion of iridescent powder", 
         1.7, 0.7, 20, 144, False, 245112]
    ]

    examples_component = gr.Examples(
        examples=examples,
        inputs=[image_input, style_input, ref_image_input, prompts_input, shared_score_shift_input, shared_score_scale_input, guidance_scale_input, num_inference_steps_input, large_input, seed_input],
        outputs=output_gallery,
        fn=run,
        cache_examples=True
    )

    demo.add_component(examples_component)

demo.launch()