app.py

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

device = "cuda" if torch.cuda.is_available() else "cpu"
dtype = torch.float16 if torch.cuda.is_available() else torch.float32
model_repo_id = "AiArtLab/sdxs-1b"

pipe = DiffusionPipeline.from_pretrained(
    model_repo_id,
    torch_dtype=dtype,
    trust_remote_code=True
).to(device)

MAX_SEED = np.iinfo(np.int32).max
MIN_IMAGE_SIZE = 768
MAX_IMAGE_SIZE = 1408
STEP = 64

@spaces.GPU(duration=60)
def infer(
    prompt: str,
    negative_prompt: str,
    seed: int,
    randomize_seed: bool,
    width: int,
    height: int,
    guidance_scale: float,
    num_inference_steps: int,
    refine_prompt: bool,
    progress=gr.Progress(track_tqdm=True),
) -> Tuple[Image.Image, int, str]:
    
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
    
    # Используем новую выделенную функцию улучшения промпта
    if refine_prompt:
        refined_list = pipe.refine_prompts(prompt)
        prompt = refined_list[0] # Метод возвращает список, берем первый элемент
    
    output = pipe(
        prompt=prompt,
        negative_prompt=negative_prompt,
        guidance_scale=guidance_scale,
        num_inference_steps=num_inference_steps,
        width=width,
        height=height,
        seed=seed,
    )

    image = output.images[0]
    
    # Возвращаем улучшенный промпт, чтобы он отобразился в интерфейсе
    return image, seed, prompt

examples = [
    "A young woman with striking blue eyes and pointed ears, adorned with a floral kimono and a tattoo. Her hair is styled in a braid, and she wears a pair of ears",
    "A frozen river, surrounded by snow-covered trees, reflects the clear blue sky, with a warm glow from the setting sun.",
    "There is a young male character standing against a vibrant, colorful graffiti wall. he is wearing a straw hat, a black jacket adorned with gold accents, and black shorts.",
    "A man with dark hair and a beard is meticulously carving an intricate design on a piece of pottery. He is wearing a traditional scarf and a white shirt, and he is focused on his work.",
    "girl, smiling, red eyes, blue hair, white shirt"
]

css = """
#col-container {
    margin: 0 auto;
    max-width: 640px;
}
"""

with gr.Blocks(css=css) as demo:
    with gr.Column(elem_id="col-container"):
        gr.Markdown(" # Simple Diffusion (sdxs)")

        with gr.Row():
            prompt = gr.Text(
                label="Prompt",
                show_label=False,
                max_lines=5,
                placeholder="Enter your prompt",
                value ="cat",
                container=False,
            )

            run_button = gr.Button("Run", scale=0, variant="primary")

        result = gr.Image(label="Result", show_label=False)

        with gr.Accordion("Advanced Settings", open=False):
            refine_prompt = gr.Checkbox(label="Refine Prompt", value=True)
            
            negative_prompt = gr.Text(
                label="Negative prompt",
                max_lines=1,
                placeholder="Enter a negative prompt",
                value ="bad quality grainy image with low details, incomplete text, despite numerous technical flaws and distorted figures"
            )

            seed = gr.Slider(
                label="Seed",
                minimum=0,
                maximum=MAX_SEED,
                step=1,
                value=0,
            )
            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)

            with gr.Row():
                width = gr.Slider(
                    label="Width",
                    minimum=MIN_IMAGE_SIZE,
                    maximum=MAX_IMAGE_SIZE,
                    step=STEP,
                    value=1024,
                )

                height = gr.Slider(
                    label="Height",
                    minimum=MIN_IMAGE_SIZE,
                    maximum=MAX_IMAGE_SIZE,
                    step=STEP,
                    value=MAX_IMAGE_SIZE,
                )

            with gr.Row():
                guidance_scale = gr.Slider(
                    label="Guidance scale",
                    minimum=0.0,
                    maximum=10.0,
                    step=0.5,
                    value=4.0,
                )

                num_inference_steps = gr.Slider(
                    label="Number of inference steps",
                    minimum=1,
                    maximum=50,
                    step=1,
                    value=40,
                )

        gr.Examples(examples=examples, inputs=[prompt])
    
    gr.on(
        triggers=[run_button.click, prompt.submit],
        fn=infer,
        inputs=[
            prompt,
            negative_prompt,
            seed,
            randomize_seed,
            width,
            height,
            guidance_scale,
            num_inference_steps,
            refine_prompt,
        ],
        outputs=[result, seed, prompt],
    )

if __name__ == "__main__":
    demo.launch()