app.py

import gradio as gr
import numpy as np
import random
from diffusers import DiffusionPipeline
import torch
from PIL import Image

device = "cuda" if torch.cuda.is_available() else "cpu"

LORA=False

if torch.cuda.is_available():
    torch.cuda.max_memory_allocated(device=device)
    pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16, variant="fp16", use_safetensors=True)
    pipe.enable_xformers_memory_efficient_attention()
    pipe = pipe.to(device)
else: 
    pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", use_safetensors=True)
    pipe = pipe.to(device)

MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 1024





def infer(prompt, negative_prompt, seed, randomize_seed, image_style,cfg,lora_scale):
    pipe.unload_lora_weights()
    if image_style == "No style":
        pipe.unload_lora_weights()
        
    elif image_style == "style_detailed":
        pipe.load_lora_weights("ohkarim/LoRA_logos", weight_name="OH_logos.safetensors", adapter_name="OH_logos")
        prompt=prompt+", detailed, close up, unique background color, OH_logos"
        
    elif image_style == "style_blacknwhite":
        pipe.load_lora_weights("ohkarim/lora_logo_blacknwhite", weight_name="bel_blacknwhite_lora.safetensors", adapter_name="bel_blacknwhite")
        prompt=prompt+", black and white, balck and white, minimalist, unique background color, bel_blacknwhite"

    elif image_style == "style_cartoon":
        pipe.load_lora_weights("ohkarim/lora_logo_modern", weight_name="lora_modern.safetensors", adapter_name="oh_bel_modern")
        prompt=prompt+", simple, modern, unique background color, oh_bel_modern"


    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
        
    generator = torch.Generator().manual_seed(seed)
    
    griddd = []
    for _ in range(4):
        image = pipe(
            prompt = prompt, 
            negative_prompt = negative_prompt,
            cross_attention_kwargs={"scale": lora_scale},
            guidance_scale = cfg, 
            num_inference_steps = 20, 
            width = 512, 
            height = 512,
            generator = generator
        ).images[0] 
        griddd.append(image)
    
    return griddd


css="""
#col-container {
    margin: 0 auto;
    max-width: 520px;
}
"""

if torch.cuda.is_available():
    power_device = "GPU"
else:
    power_device = "CPU"


# Function to read CSS from file
def read_css_from_file(filename):
    with open(filename, "r") as file:
        return file.read()

# Read CSS from file
css = read_css_from_file("style.css")

PTI_SD_DESCRIPTION = '''
<div id="content_align">
  <span style="color:darkred;font-size:32px;font-weight:bold">  
    Create your own logo now!
  </span>
</div>
<div id="content_align">
  <span style="color:blue;font-size:16px;font-weight:bold">  
    There is 3 styles so far, detailed, black & white and cartoon
  </span>
</div>
<div id="content_align" style="margin-top: 10px;">
</div>
'''

# Creating Gradio interface
with gr.Blocks(css=css) as demo:
    gr.Markdown(PTI_SD_DESCRIPTION)
    with gr.Row():   
        with gr.Column():
            text_prompt = gr.Textbox(label="Input Prompt", placeholder="Example: logo of a coffee shop, cup of coffee, mug, brown shades ", lines=2)
            negative_prompt = gr.Textbox(label="Negative Prompt (optional)", placeholder="Example: blurry, unfocused, complicated", lines=2)
            image_style = gr.Dropdown(label="Choose a style", choices=["No style", "style_detailed", "style_blacknwhite", "style_cartoon"], value="No LoRA")

        with gr.Column():
            cfg = gr.Slider(label="Prompt guidance", minimum=1, maximum=20, step=1, value=4) 
            lora_scale = gr.Slider(label="Style guidance", minimum=0, maximum=2, step=0.01, value=0.9) 
            seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0) 
            
            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
                      
    generate_button = gr.Button("Generate Image", variant='primary')
        
    with gr.Row():
        output_image1 = gr.Image(type="pil", label="Image 1", width=256, height=400)
        output_image2 = gr.Image(type="pil", label="Image 2", width=256, height=400)  
        output_image3 = gr.Image(type="pil", label="Image 3", width=256, height=400)
        output_image4 = gr.Image(type="pil", label="Image 4", width=256, height=400)

    generate_button.click(
        fn = infer,
        inputs = [text_prompt, negative_prompt, seed, randomize_seed,image_style,cfg,lora_scale],
        outputs = [output_image1,output_image2,output_image3,output_image4],
        show_progress=True
    )
demo.queue().launch()