import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig


prompt = "The camera smoothly arcs counterclockwise, offering a serene view of a secluded island surrounded by crystal-clear waters. The lush greenery and rugged cliffs create a picturesque contrast against the deep blue sea, while distant boats dot the horizon under a clear sky."

# pipe = WanVideoPipeline.from_pretrained(
#     torch_dtype=torch.bfloat16,
#     device="cuda",
#     model_configs=[
#         ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
#         ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
#         ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
#     ],
# )
# pipe.enable_vram_management()

# video = pipe(
#     prompt=prompt,
#     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",
#     seed=1, tiled=True
# )
# save_video(video, "outputs/video_Wan2.1-T2V-1.3B.mp4", fps=16, quality=5)

pipe = WanVideoPipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[
        ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
        ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
        ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
    ],
)
state_dict = load_state_dict("models/train/Wan2.1-T2V-1.3B_full-1gpu/epoch-2.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()

video = pipe(
    prompt=prompt,
    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",
    seed=1, tiled=True
)
save_video(video, "outputs/video_Wan2.1-T2V-1.3B_full.mp4", fps=16, quality=5)