app.py

import gradio as gr
from PIL import Image
from io import BytesIO
import torch
import os

#os.system("pip install git+https://github.com/fffiloni/diffusers")

from diffusers import DiffusionPipeline, DDIMScheduler
from imagic import ImagicStableDiffusionPipeline

has_cuda = torch.cuda.is_available()
device = "cuda" 

pipe = ImagicStableDiffusionPipeline.from_pretrained(
    "CompVis/stable-diffusion-v1-4",
    safety_checker=None,
    #custom_pipeline=ImagicStableDiffusionPipeline,
    scheduler = DDIMScheduler(beta_start=0.00085, beta_end=0.012, beta_schedule="scaled_linear", clip_sample=False, set_alpha_to_one=False)
).to(device)

generator = torch.Generator("cuda").manual_seed(0)

def train(prompt, init_image, trn_text, trn_steps):
    init_image = Image.open(init_image).convert("RGB")
    init_image = init_image.resize((256, 256))

   
    res = pipe.train(
        prompt,
        init_image,
        guidance_scale=7.5,
        num_inference_steps=50,
        generator=generator,
        text_embedding_optimization_steps=trn_text,
        model_fine_tuning_optimization_steps=trn_steps)
    
    with torch.no_grad():
        torch.cuda.empty_cache()
    
    return "Training is finished !", gr.update(value=0), gr.update(value=0)
   
def generate(prompt, init_image, trn_text, trn_steps):
    init_image = Image.open(init_image).convert("RGB")
    init_image = init_image.resize((256, 256))

   
    res = pipe.train(
        prompt,
        init_image,
        guidance_scale=7.5,
        num_inference_steps=50,
        generator=generator,
        text_embedding_optimization_steps=trn_text,
        model_fine_tuning_optimization_steps=trn_steps)
    
    with torch.no_grad():
        torch.cuda.empty_cache()
    
    
    
    res = pipe(alpha=1)
    
    
    return res.images[0]
    
title = """
    <div style="text-align: center; max-width: 650px; margin: 0 auto;">
        <div
        style="
            display: inline-flex;
            align-items: center;
            gap: 0.8rem;
            font-size: 1.75rem;
        "
        >
        <h1 style="font-weight: 900; margin-top: 7px;">
            Imagic Stable Diffusion • Community Pipeline
        </h1>
        </div>
         <p style="margin-top: 10px; font-size: 94%">
        Text-Based Real Image Editing with Diffusion Models
        <br />This pipeline aims to implement <a href="https://arxiv.org/abs/2210.09276" target="_blank">this paper</a> to Stable Diffusion, allowing for real-world image editing.
        
        </p>
        <br /><img src="https://user-images.githubusercontent.com/788417/196388568-4ee45edd-e990-452c-899f-c25af32939be.png" style="margin:7px 0 20px;"/>
       
        <p style="font-size: 94%">
            You can skip the queue by duplicating this space or run the Colab version: 
            <span style="display: flex;align-items: center;justify-content: center;height: 30px;">
            <a href="https://huggingface.co/spaces/fffiloni/imagic-stable-diffusion?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>       
            </span>
        </p>

    </div>
"""

article = """
    <div class="footer">
        <p><a href="https://github.com/huggingface/diffusers/tree/main/examples/community#imagic-stable-diffusion" target="_blank">Community pipeline</a> 
        baked by <a href="https://github.com/MarkRich" style="text-decoration: underline;" target="_blank">Mark Rich</a> - 
        Gradio Demo by 🤗 <a href="https://twitter.com/fffiloni" target="_blank">Sylvain Filoni</a>
        </p>
    </div>
"""

css = '''
    #col-container {max-width: 700px; margin-left: auto; margin-right: auto;}
    a {text-decoration-line: underline; font-weight: 600;}
    .footer {
            margin-bottom: 45px;
            margin-top: 35px;
            text-align: center;
            border-bottom: 1px solid #e5e5e5;
        }
        .footer>p {
            font-size: .8rem;
            display: inline-block;
            padding: 0 10px;
            transform: translateY(10px);
            background: white;
        }
        .dark .footer {
            border-color: #303030;
        }
        .dark .footer>p {
            background: #0b0f19;
        }
'''


with gr.Blocks(css=css) as block:
    with gr.Column(elem_id="col-container"):
        gr.HTML(title)

        prompt_input = gr.Textbox(label="Target text", placeholder="Describe the image with what you want to change about the subject")
        image_init = gr.Image(source="upload", type="filepath",label="Input Image")
        with gr.Row():
            trn_text = gr.Slider(0, 500, step=50, value=250, label="text embedding")
            trn_steps = gr.Slider(0, 1000, step=50, value=500, label="finetuning steps")
        with gr.Row():
            train_btn = gr.Button("1.Train")
            gen_btn = gr.Button("2.Generate")

        training_status = gr.Textbox(label="training status")
        image_output = gr.Image(label="Edited image")

        #examples=[['a sitting dog','imagic-dog.png', 250], ['a photo of a bird spreading wings','imagic-bird.png',250]]
        #ex = gr.Examples(examples=examples, fn=infer, inputs=[prompt_input,image_init,trn_steps], outputs=[image_output], cache_examples=False, run_on_click=False)
        
        
        gr.HTML(article)

    train_btn.click(fn=train, inputs=[prompt_input,image_init,trn_text,trn_steps], outputs=[training_status, trn_text, trn_steps])
    gen_btn.click(fn=generate, inputs=[prompt_input,image_init,trn_text,trn_steps], outputs=[image_output])
    
block.queue(max_size=12).launch(show_api=False)