app_aot.py

# PyTorch 2.8 (temporary hack)
import os
os.system('pip install --upgrade --pre --extra-index-url https://download.pytorch.org/whl/nightly/cu126 "torch<2.9" spaces')

# Actual demo code
import spaces
import torch
from diffusers.pipelines.wan.pipeline_wan_i2v import WanImageToVideoPipeline
from diffusers.models.transformers.transformer_wan import WanTransformer3DModel
from diffusers.utils.export_utils import export_to_video
import gradio as gr
import tempfile
import numpy as np
from PIL import Image
import random
import gc
import logging
from optimization import optimize_pipeline_

# Hugging Face
from huggingface_hub import HfApi, upload_file
import uuid
from datetime import datetime
from queue import Queue
from threading import Thread
import time

# -----------------------------
# Constants
# -----------------------------
MODEL_ID = "Wan-AI/Wan2.2-I2V-A14B-Diffusers"
LANDSCAPE_WIDTH = 832
LANDSCAPE_HEIGHT = 480
MAX_SEED = np.iinfo(np.int32).max
FIXED_FPS = 16
MIN_FRAMES_MODEL = 8
MAX_FRAMES_MODEL = 81
MIN_DURATION = round(MIN_FRAMES_MODEL/FIXED_FPS,1)
MAX_DURATION = round(MAX_FRAMES_MODEL/FIXED_FPS,1)
HF_MODEL = os.environ.get("HF_UPLOAD_REPO", "rahul7star/VideoExplain")

default_prompt_i2v = "make this image come alive, cinematic motion, smooth animation"
default_negative_prompt = "色调艳丽, 过曝, 静态, 细节模糊不清, 字幕, 风格, 作品, 画作, 画面, 静止, 整体发灰, 最差质量, 低质量, JPEG压缩残留, 丑陋的, 残缺的, 多余的手指, 画得不好的手部, 画得不好的脸部, 畸形的, 毁容的, 形态畸形的肢体, 手指融合, 静止不动的画面, 杂乱的背景, 三条腿, 背景人很多, 倒着走"

# -----------------------------
# Initialize Pipeline
# -----------------------------
pipe = WanImageToVideoPipeline.from_pretrained(MODEL_ID,
    transformer=WanTransformer3DModel.from_pretrained(
        'cbensimon/Wan2.2-I2V-A14B-bf16-Diffusers',
        subfolder='transformer',
        torch_dtype=torch.bfloat16,
        device_map='cuda',
    ),
    transformer_2=WanTransformer3DModel.from_pretrained(
        'cbensimon/Wan2.2-I2V-A14B-bf16-Diffusers',
        subfolder='transformer_2',
        torch_dtype=torch.bfloat16,
        device_map='cuda',
    ),
    torch_dtype=torch.bfloat16,
).to('cuda')

# Clean GPU
for _ in range(3):
    gc.collect()
    torch.cuda.synchronize()
    torch.cuda.empty_cache()

# Optimize pipeline
optimize_pipeline_(pipe,
    image=Image.new('RGB', (LANDSCAPE_WIDTH, LANDSCAPE_HEIGHT)),
    prompt='prompt',
    height=LANDSCAPE_HEIGHT,
    width=LANDSCAPE_WIDTH,
    num_frames=MAX_FRAMES_MODEL,
)

# -----------------------------
# Upload Queue for Lazy Upload
# -----------------------------
upload_queue = Queue()

def upload_worker():
    while True:
        try:
            video_path, summary_text = upload_queue.get()
            logging.info(f"⏳ Uploading video in background: {video_path}")
            upscale_and_upload_4k(video_path, summary_text)
            logging.info(f"✅ Background upload finished: {video_path}")
        except Exception as e:
            logging.error(f"Upload failed: {e}")
        time.sleep(3)  # Delay to avoid HF rate limits
        upload_queue.task_done()

Thread(target=upload_worker, daemon=True).start()

# -----------------------------
# Helper Functions
# -----------------------------
def resize_image(image: Image.Image) -> Image.Image:
    if image.height > image.width:
        transposed = image.transpose(Image.Transpose.ROTATE_90)
        resized = resize_image_landscape(transposed)
        return resized.transpose(Image.Transpose.ROTATE_270)
    return resize_image_landscape(image)

def resize_image_landscape(image: Image.Image) -> Image.Image:
    target_aspect = LANDSCAPE_WIDTH / LANDSCAPE_HEIGHT
    width, height = image.size
    in_aspect = width / height
    if in_aspect > target_aspect:
        new_width = round(height * target_aspect)
        left = (width - new_width) // 2
        image = image.crop((left, 0, left + new_width, height))
    else:
        new_height = round(width / target_aspect)
        top = (height - new_height) // 2
        image = image.crop((0, top, width, top + new_height))
    return image.resize((LANDSCAPE_WIDTH, LANDSCAPE_HEIGHT), Image.LANCZOS)

def get_duration(input_image, prompt, steps, negative_prompt, duration_seconds, guidance_scale, guidance_scale_2, seed, randomize_seed, progress):
    return int(steps) * 15

# -----------------------------
# Upscale + HF Upload Function
# -----------------------------
import subprocess

def upscale_and_upload_4k(input_video_path: str, summary_text: str) -> str:
    logging.info(f"Upscaling video to 4K for upload: {input_video_path}")

    # Temporary file for upscaled video
    with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as tmp_upscaled:
        upscaled_path = tmp_upscaled.name

    cmd = [
        "ffmpeg",
        "-i", input_video_path,
        "-vf", "scale=3840:2160:flags=lanczos",
        "-c:v", "libx264",
        "-crf", "18",
        "-preset", "slow",
        "-y",
        upscaled_path,
    ]
    try:
        subprocess.run(cmd, check=True, capture_output=True)
        logging.info(f"✅ Upscaled video created at: {upscaled_path}")
    except subprocess.CalledProcessError as e:
        logging.error(f"FFmpeg failed:\n{e.stderr.decode()}")
        raise

    # HF Upload folder
    today_str = datetime.now().strftime("%Y-%m-%d")
    unique_subfolder = f"upload_{uuid.uuid4().hex[:8]}"
    hf_folder = f"{today_str}-WAN-I2V/{unique_subfolder}"

    # Upload video
    video_filename = os.path.basename(input_video_path)
    video_hf_path = f"{hf_folder}/{video_filename}"
    upload_file(
        path_or_fileobj=upscaled_path,
        path_in_repo=video_hf_path,
        repo_id=HF_MODEL,
        repo_type="model",
        token=os.environ.get("HUGGINGFACE_HUB_TOKEN"),
    )

    # Upload summary
    summary_file = tempfile.NamedTemporaryFile(delete=False, suffix=".txt").name
    with open(summary_file, "w", encoding="utf-8") as f:
        f.write(summary_text)

    summary_hf_path = f"{hf_folder}/summary.txt"
    upload_file(
        path_or_fileobj=summary_file,
        path_in_repo=summary_hf_path,
        repo_id=HF_MODEL,
        repo_type="model",
        token=os.environ.get("HUGGINGFACE_HUB_TOKEN"),
    )
    logging.info(f"✅ Uploaded summary to HF: {summary_hf_path}")

    # Cleanup
    os.remove(upscaled_path)
    os.remove(summary_file)
    return hf_folder

# -----------------------------
# Video Generation Function
# -----------------------------
@spaces.GPU(duration=get_duration)
def generate_video(
    input_image,
    prompt,
    steps = 4,
    negative_prompt=default_negative_prompt,
    duration_seconds = MAX_DURATION,
    guidance_scale = 1,
    guidance_scale_2 = 1,    
    seed = 42,
    randomize_seed = False,
    progress=gr.Progress(track_tqdm=True),
):
    if input_image is None:
        raise gr.Error("Please upload an input image.")
    
    num_frames = np.clip(int(round(duration_seconds * FIXED_FPS)), MIN_FRAMES_MODEL, MAX_FRAMES_MODEL)
    current_seed = random.randint(0, MAX_SEED) if randomize_seed else int(seed)
    resized_image = resize_image(input_image)

    output_frames_list = pipe(
        image=resized_image,
        prompt=prompt,
        negative_prompt=negative_prompt,
        height=resized_image.height,
        width=resized_image.width,
        num_frames=num_frames,
        guidance_scale=float(guidance_scale),
        guidance_scale_2=float(guidance_scale_2),
        num_inference_steps=int(steps),
        generator=torch.Generator(device="cuda").manual_seed(current_seed),
    ).frames[0]

    with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as tmpfile:
        video_path = tmpfile.name

    export_to_video(output_frames_list, video_path, fps=FIXED_FPS)

    # Queue upload in background
    upload_queue.put((video_path, prompt))
    logging.info(f"Video queued for background upload: {video_path}")

    return video_path, current_seed

# -----------------------------
# Gradio UI
# -----------------------------
with gr.Blocks() as demo:
    gr.Markdown("# Fast 4 steps Wan 2.2 I2V (14B) with Lightning LoRA")
    gr.Markdown("run Wan 2.2 in just 4-8 steps, with Lightning LoRA, fp8 quantization & AoT compilation")

    with gr.Row():
        with gr.Column():
            input_image_component = gr.Image(type="pil", label="Input Image (auto-resized)")
            prompt_input = gr.Textbox(label="Prompt", value=default_prompt_i2v)
            duration_seconds_input = gr.Slider(minimum=MIN_DURATION, maximum=MAX_DURATION, step=0.1, value=MAX_DURATION, label="Duration (seconds)")

            with gr.Accordion("Advanced Settings", open=False):
                negative_prompt_input = gr.Textbox(label="Negative Prompt", value=default_negative_prompt, lines=3)
                seed_input = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=42, interactive=True)
                randomize_seed_checkbox = gr.Checkbox(label="Randomize seed", value=True, interactive=True)
                steps_slider = gr.Slider(minimum=1, maximum=30, step=1, value=6, label="Inference Steps") 
                guidance_scale_input = gr.Slider(minimum=0.0, maximum=10.0, step=0.5, value=1, label="Guidance Scale - high noise stage")
                guidance_scale_2_input = gr.Slider(minimum=0.0, maximum=10.0, step=0.5, value=1, label="Guidance Scale 2 - low noise stage")

            generate_button = gr.Button("Generate Video", variant="primary")
        with gr.Column():
            video_output = gr.Video(label="Generated Video", autoplay=True, interactive=False)

    ui_inputs = [
        input_image_component, prompt_input, steps_slider,
        negative_prompt_input, duration_seconds_input,
        guidance_scale_input, guidance_scale_2_input, seed_input, randomize_seed_checkbox
    ]
    generate_button.click(fn=generate_video, inputs=ui_inputs, outputs=[video_output, seed_input])

if __name__ == "__main__":
    demo.queue().launch(mcp_server=True)