from diffusers import StableDiffusionPipeline
import gradio as gr
import torch

models = [
  "DGSpitzer/Cyberpunk-Anime-Diffusion"
]

prompt_prefixes = {
  models[0]: "dgs illustration style "
}

current_model = models[0]
pipe = StableDiffusionPipeline.from_pretrained(current_model, torch_dtype=torch.float16)
if torch.cuda.is_available():
  pipe = pipe.to("cuda")

device = "GPU 🔥" if torch.cuda.is_available() else "CPU 🥶"

def on_model_change(model):

    global current_model
    global pipe
    if model != current_model:
        current_model = model
        pipe = StableDiffusionPipeline.from_pretrained(current_model, torch_dtype=torch.float16)
        if torch.cuda.is_available():
            pipe = pipe.to("cuda")

def inference(prompt, guidance, steps):

    prompt = prompt_prefixes[current_model] + prompt
    image = pipe(prompt, num_inference_steps=int(steps), guidance_scale=guidance, width=512, height=512).images[0]
    return image

with gr.Blocks() as demo:
    gr.HTML(
        """
            <div style="text-align: center; max-width: 700px; margin: 0 auto;">
              <div
                style="
                  display: inline-flex;
                  align-items: center;
                  gap: 0.8rem;
                  font-size: 1.75rem;
                "
              >
                <h1 style="font-weight: 900; margin-bottom: 7px;">
                  DGS Diffusion Space
                </h1>
              </div>
              <p style="margin-bottom: 10px; font-size: 94%">
               Demo for Cyberpunk Anime Diffusion. Based of Finetuned Diffusion by anzorq <a href="https://twitter.com/hahahahohohe">
              </p>
            </div>
        """
    )
    with gr.Row():
        
        with gr.Column():
            model = gr.Dropdown(label="Model", choices=models, value=models[0])
            prompt = gr.Textbox(label="Prompt", placeholder="{} is added automatically".format(prompt_prefixes[current_model]))
            guidance = gr.Slider(label="Guidance scale", value=7.5, maximum=15)
            steps = gr.Slider(label="Steps", value=27, maximum=100, minimum=2)
            run = gr.Button(value="Run")
            gr.Markdown(f"Running on: {device}")
        with gr.Column():
            image_out = gr.Image(height=512)

    model.change(on_model_change, inputs=model, outputs=[])
    run.click(inference, inputs=[prompt, guidance, steps], outputs=image_out)
    gr.Examples([
        ["portrait of anime girl", 7.5, 27],
        ["a beautiful perfect face girl, Anime fine details portrait of school girl in front of modern tokyo city landscape on the background deep bokeh, anime masterpiece by studio ghibli, 8k, sharp high quality anime, artstation", 7.5, 27],
        ["cyberpunk city landscape with fancy car", 7.5, 27],
        ["portrait of liu yifei girl, soldier working in a cyberpunk city, cleavage, intricate, 8k, highly detailed, digital painting, intense, sharp focus", 7.5, 27],
        ["portrait of a muscular beard male in dgs illustration style, half-body, holding robot arms, strong chest", 7.5, 27],
    ], [prompt, guidance, steps], image_out, inference, cache_examples=torch.cuda.is_available())
    gr.Markdown('''
      Models and Space by [@DGSpitzer](https://huggingface.co/DGSpitzer)❤️<br>
      [![Twitter Follow](https://img.shields.io/twitter/follow/DGSpitzer?label=%40DGSpitzer&style=social)](https://twitter.com/DGSpitzer)

      ![visitors](https://visitor-badge.glitch.me/badge?page_id=dgspitzer_DGS_Diffusion_Space)
    ''')

demo.queue()
demo.launch()