app.py

import os
import threading
import time
from datetime import datetime, timedelta

import gradio as gr
import numpy as np
import random
import spaces  # [uncomment to use ZeroGPU]
from diffusers import FluxPipeline
import torch

device = "cuda" if torch.cuda.is_available() else "cpu"
model_repo_id = "black-forest-labs/FLUX.1-dev"

if torch.cuda.is_available():
    torch_dtype = torch.float16
else:
    torch_dtype = torch.float32

pipe = FluxPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
pipe = pipe.to(device)

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


def delete_old_files():
    while True:
        now = datetime.now()
        cutoff = now - timedelta(minutes=10)
        directories = ["./gradio_tmp"]

        for directory in directories:
            for filename in os.listdir(directory):
                file_path = os.path.join(directory, filename)
                if os.path.isfile(file_path):
                    file_mtime = datetime.fromtimestamp(os.path.getmtime(file_path))
                    if file_mtime < cutoff:
                        os.remove(file_path)
        time.sleep(600)


threading.Thread(target=delete_old_files, daemon=True).start()


@spaces.GPU  # [uncomment to use ZeroGPU]
def infer(prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps,
          progress=gr.Progress(track_tqdm=True)):
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)

    generator = torch.Generator().manual_seed(seed)

    image = pipe(
        prompt=prompt,
        guidance_scale=guidance_scale,
        num_images_per_prompt=1,
        num_inference_steps=num_inference_steps,
        width=width,
        height=height,
        generator=generator
    ).images[0]
    return image, seed


examples = [
    "A vintage pink convertible with glossy chrome finishes and whitewall tires sits parked on an open road, surrounded by a field of wildflowers under a clear blue sky. The car's body is a delicate pastel pink, complementing the vibrant greens and colors of the meadow. Its interior boasts cream leather seats and a polished wooden dashboard, evoking a sense of classic elegance. The sun casts a soft light on the vehicle, highlighting its curves and shiny surfaces, creating a picture of nostalgia mixed with dreamy escapism.",
    "A noble black Labrador retriever sits serenely in a sunlit meadow, its glossy coat absorbing the golden rays of a late afternoon sun. The dog's intelligent eyes sparkle with a mixture of curiosity and loyalty, as it gazes off into the distance where the meadow meets a line of tall, slender birch trees. The dog's posture is regal, yet approachable, with its tongue playfully hanging out to the side slightly, suggesting a friendly disposition. The idyllic setting is filled with the vibrant greens of lush grass and the soft colors of wildflowers speckled throughout, creating a peaceful harmony between the dog and its natural surroundings.",
    "A vibrant red-colored dog of medium build stands attentively in an autumn forest setting. Its fur is a deep, rich red, reminiscent of autumn leaves, contrasting with its bright, intelligent eyes, a clear sky blue. The dog's ears perk up, and its tail wags slightly as it looks off into the distance, its posture suggesting alertness and curiosity. Golden sunlight filters through the canopy of russet and gold leaves above, casting dappled light onto the forest floor and the glossy coat of the canine, creating a serene and heartwarming scene."
]

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(f"""
            <div style="text-align: center; font-size: 32px; font-weight: bold; margin-bottom: 20px;">
             CogView3-Plus Huggingface Space🤗
           </div>
           <div style="text-align: center;">
               <a href="https://huggingface.co/THUDM/CogView3-Plus">🤗 Model Hub | 
               <a href="https://github.com/THUDM/CogView3">🌐 Github</a> |
               <a href="https://arxiv.org/abs/2403.05121">📜 arxiv </a>
           </div>
           <div style="text-align: center;display: flex;justify-content: center;align-items: center;margin-top: 1em;margin-bottom: .5em;">
              <span>If the Space is too busy, duplicate it to use privately</span>
              <a href="https://huggingface.co/spaces/THUDM-HF-SPACE/CogView-3-Plus?duplicate=true"><img src="https://huggingface.co/datasets/huggingface/badges/resolve/main/duplicate-this-space-lg.svg" width="160" style="
                margin-left: .75em;
            "></a>
           </div>
           <div style="text-align: center; font-size: 15px; font-weight: bold; color: red; margin-bottom: 20px;">
            ⚠️ This demo is for academic research and experiential use only. 
            </div>
        """)

        with gr.Row():
            prompt = gr.Text(
                label="Prompt",
                show_label=False,
                max_lines=1,
                placeholder="Enter your prompt",
                container=False,
            )

            run_button = gr.Button("Run", scale=0)
        result = gr.Image(label="Result", show_label=False)

        with gr.Accordion("Advanced Settings", open=False):
            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=512,
                    maximum=2048,
                    step=32,
                    value=1024,  # 替换为你的模型默认值
                )

                height = gr.Slider(
                    label="Height",
                    minimum=512,
                    maximum=2048,
                    step=32,
                    value=1024,
                )

            with gr.Row():
                guidance_scale = gr.Slider(
                    label="Guidance scale",
                    minimum=0.0,
                    maximum=10.0,
                    step=0.1,
                    value=7.0,
                )

                num_inference_steps = gr.Slider(
                    label="Number of inference steps",
                    minimum=10,
                    maximum=100,
                    step=1,
                    value=50,
                )

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

demo.queue().launch(server_name="127.0.0.1")