app.py

import spaces
from diffusers import StableDiffusionPipeline, StableDiffusionImg2ImgPipeline, UniPCMultistepScheduler
import gradio as gr
import torch
from PIL import Image

model_id = 'Norod78/sd2-simpsons-blip'
prefix = None

pipe = StableDiffusionPipeline.from_pretrained(
  model_id,
  torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32)

pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)

pipe_i2i = StableDiffusionImg2ImgPipeline.from_pretrained(
  model_id,
  torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32)

pipe_i2i.scheduler = UniPCMultistepScheduler.from_config(pipe_i2i.scheduler.config)

if torch.cuda.is_available():
  pipe = pipe.to("cuda")
  pipe_i2i = pipe_i2i.to("cuda")

def error_str(error, title="Error"):
    return f"""#### {title}
            {error}"""  if error else ""

@spaces.GPU
def inference(prompt, guidance, steps, width=512, height=768, seed=0, img=None, strength=0.5, neg_prompt=""):

  if torch.cuda.is_available():
    generator = torch.Generator('cuda').manual_seed(seed) if seed != 0 else None
  else:
    if seed != 0:      
      generator = torch.Generator()
      generator.manual_seed(seed)
    else:
      generator = None

  try:
    if img is not None:
      return img_to_img(prompt, neg_prompt, img, strength, guidance, steps, width, height, generator), None
    else:
      return txt_to_img(prompt, neg_prompt, guidance, steps, width, height, generator), None
  except Exception as e:
    return None, error_str(e)

def txt_to_img(prompt, neg_prompt, guidance, steps, width, height, generator):

    result = pipe(
      prompt,
      negative_prompt = neg_prompt,
      num_inference_steps = int(steps),
      guidance_scale = guidance,
      width = width,
      height = height,
      generator = generator)
    
    return replace_nsfw_images(result)

def img_to_img(prompt, neg_prompt, img, strength, guidance, steps, width, height, generator):

    ratio = min(height / img.height, width / img.width)
    img = img.resize((int(img.width * ratio), int(img.height * ratio)), Image.Resampling.LANCZOS)
    result = pipe_i2i(
        prompt,
        negative_prompt = neg_prompt,
        image = img,
        num_inference_steps = int(steps),
        strength = strength,
        guidance_scale = guidance,
        generator = generator)
        
    return replace_nsfw_images(result)

def replace_nsfw_images(results):

    for i in range(len(results.images)):
      if 'nsfw_content_detected' in results and results.nsfw_content_detected[i]:
        results.images[i] = Image.open("nsfw.png")
    return results.images[0]

css = """.main-div div{display:inline-flex;align-items:center;gap:.8rem;font-size:1.75rem}.main-div div h1{font-weight:900;margin-bottom:7px}.main-div p{margin-bottom:10px;font-size:94%}a{text-decoration:underline}.tabs{margin-top:0;margin-bottom:0}#gallery{min-height:20rem}
"""
with gr.Blocks(css=css) as demo:
    gr.HTML(
        f"""
            <div class="main-div">
              <div>
                <h1>SDv2 Simpsons</h1>
              </div>
              <p>
               Demo for <a href="https://huggingface.co/Norod78/sd2-simpsons-blip">SD2 Simpsons BLIP</a> Stable Diffusion 2, fine-tuned model.<br>
               {"Add the following tokens to your prompts for the model to work properly: <b>prefix</b>" if prefix else ""}
              </p>
              Running on {"<b>GPU 🔥</b>" if torch.cuda.is_available() else f"<b>CPU 🥶</b>. For faster inference it is recommended to <b>upgrade to GPU in <a href='https://huggingface.co/spaces/Norod78/sd2-simpsons-blip/settings'>Settings</a></b>"}<br><br>
              <a style="display:inline-block" href="https://huggingface.co/spaces/Norod78/sd2-simpsons-blip?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a>
            </div>
        """
    )
    with gr.Row():
        
        with gr.Column(scale=55):
          with gr.Group():
              with gr.Row():
                prompt = gr.Textbox(label="Prompt", show_label=False, max_lines=2,placeholder="[your prompt]")
                generate = gr.Button(value="Generate")

              image_out = gr.Image()
          error_output = gr.Markdown()

        with gr.Column(scale=45):
          with gr.Tab("Options"):
            with gr.Group():
              neg_prompt = gr.Textbox(label="Negative prompt", placeholder="What to exclude from the image", value="deformed, extra limbs, grainy, unfocused, blurry, NSFW, nude, naked")              

              with gr.Row():
                guidance = gr.Slider(label="Guidance scale", value=7.5, maximum=15)
                steps = gr.Slider(label="Steps", value=24, minimum=4, maximum=50, step=1)

              with gr.Row():
                width = gr.Slider(label="Width", value=512, minimum=64, maximum=1024, step=64)
                height = gr.Slider(label="Height", value=512, minimum=64, maximum=1024, step=64)

              seed = gr.Slider(0, 2147483647, label='Seed (0 = random)', value=0, step=1)

          with gr.Tab("Image to image"):
              with gr.Group():
                image = gr.Image(label="Image", type="pil")
                strength = gr.Slider(label="Transformation strength", minimum=0, maximum=1, step=0.01, value=0.5)

    inputs = [prompt, guidance, steps, width, height, seed, image, strength, neg_prompt]
    outputs = [image_out, error_output]
    prompt.submit(inference, inputs=inputs, outputs=outputs)
    generate.click(inference, inputs=inputs, outputs=outputs)

    gr.HTML("""
    <div style="border-top: 1px solid #303030;">
      <br>
      <p>This space was created using <a href="https://huggingface.co/spaces/anzorq/sd-space-creator">SD Space Creator</a>.</p>
    </div>
    """)

demo.queue()
demo.launch()