GeometryCrafter/run.py at main · TencentARC/GeometryCrafter · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
from pathlib import Path
import torch
from decord import VideoReader, cpu
from diffusers.training_utils import set_seed
from fire import Fire
import os
import numpy as np
import torch.nn.functional as F

from third_party import MoGe
from geometrycrafter import (
    GeometryCrafterDiffPipeline,
    GeometryCrafterDetermPipeline,
    PMapAutoencoderKLTemporalDecoder,
    UNetSpatioTemporalConditionModelVid2vid
)

def main(
    video_path: str,
    save_folder: str = "workspace/output/",
    cache_dir: str = "workspace/cache",
    height: int = None,
    width: int = None,
    downsample_ratio: float = 1.0,
    num_inference_steps: int = 5,
    guidance_scale: float = 1.0,
    window_size: int = 110,
    decode_chunk_size: int = 8,
    overlap: int = 25,
    process_length: int = -1,
    process_stride: int = 1,
    seed: int = 42,
    model_type: str = 'diff', # 'determ'
    force_projection: bool = True,
    force_fixed_focal: bool = True,
    use_extract_interp: bool = False,
    track_time: bool = False,
    low_memory_usage: bool = False
):
    assert model_type in ['diff', 'determ']
    set_seed(seed)
    unet = UNetSpatioTemporalConditionModelVid2vid.from_pretrained(
        'TencentARC/GeometryCrafter',
        subfolder='unet_diff' if model_type == 'diff' else 'unet_determ',
        low_cpu_mem_usage=True,
        torch_dtype=torch.float16,
        cache_dir=cache_dir
    ).requires_grad_(False).to("cuda", dtype=torch.float16)
    point_map_vae = PMapAutoencoderKLTemporalDecoder.from_pretrained(
        'TencentARC/GeometryCrafter',
        subfolder='point_map_vae',
        low_cpu_mem_usage=True,
        torch_dtype=torch.float32,
        cache_dir=cache_dir
    ).requires_grad_(False).to("cuda", dtype=torch.float32)
    prior_model = MoGe(
        cache_dir=cache_dir,
    ).requires_grad_(False).to('cuda', dtype=torch.float32)
    if model_type == 'diff':
        pipe = GeometryCrafterDiffPipeline.from_pretrained(
            "stabilityai/stable-video-diffusion-img2vid-xt",
            unet=unet,
            torch_dtype=torch.float16,
            variant="fp16",
            cache_dir=cache_dir
        ).to("cuda")
    else:
        pipe = GeometryCrafterDetermPipeline.from_pretrained(
            "stabilityai/stable-video-diffusion-img2vid-xt",
            unet=unet,
            torch_dtype=torch.float16,
            variant="fp16",
            cache_dir=cache_dir
        ).to("cuda")


    try:
        pipe.enable_xformers_memory_efficient_attention()
    except Exception as e:
        print(e)
        print("Xformers is not enabled")
    # bugs at https://github.com/continue-revolution/sd-webui-animatediff/issues/101
    # pipe.enable_xformers_memory_efficient_attention()
    pipe.enable_attention_slicing()

    video_base_name = os.path.basename(video_path).split('.')[0]
    vid = VideoReader(video_path, ctx=cpu(0))
    original_height, original_width = vid.get_batch([0]).shape[1:3]

    if height is None or width is None:
        height = original_height
        width = original_width

    assert height % 64 == 0
    assert width % 64 == 0

    frames_idx = list(range(0, len(vid), process_stride))
    frames = vid.get_batch(frames_idx).asnumpy().astype(np.float32) / 255.0
    if process_length > 0:
        process_length = min(process_length, len(frames))
        frames = frames[:process_length]
    else:
        process_length = len(frames)
    window_size = min(window_size, process_length)
    if window_size == process_length:
        overlap = 0
    frames_tensor = torch.tensor(frames.astype("float32"), device='cuda').float().permute(0, 3, 1, 2)
    # t,3,h,w

    if downsample_ratio > 1.0:
        original_height, original_width = frames_tensor.shape[-2], frames_tensor.shape[-1]
        frames_tensor = F.interpolate(frames_tensor, (round(frames_tensor.shape[-2]/downsample_ratio), round(frames_tensor.shape[-1]/downsample_ratio)), mode='bicubic', antialias=True).clamp(0, 1)

    save_path = Path(save_folder)
    save_path.mkdir(parents=True, exist_ok=True)

    with torch.inference_mode():
        rec_point_map, rec_valid_mask = pipe(
            frames_tensor,
            point_map_vae,
            prior_model,
            height=height,
            width=width,
            num_inference_steps=num_inference_steps,
            guidance_scale=guidance_scale,
            window_size=window_size,
            decode_chunk_size=decode_chunk_size,
            overlap=overlap,
            force_projection=force_projection,
            force_fixed_focal=force_fixed_focal,
            use_extract_interp=use_extract_interp,
            track_time=track_time,
            low_memory_usage=low_memory_usage
        )

        if downsample_ratio > 1.0:
            rec_point_map = F.interpolate(rec_point_map.permute(0,3,1,2), (original_height, original_width), mode='bilinear').permute(0, 2, 3, 1)
            rec_valid_mask = F.interpolate(rec_valid_mask.float().unsqueeze(1), (original_height, original_width), mode='bilinear').squeeze(1) > 0.5

        np.savez(
            str(save_path / f"{video_base_name}.npz"),
            point_map=rec_point_map.detach().cpu().numpy().astype(np.float16),
            mask=rec_valid_mask.detach().cpu().numpy().astype(np.bool_))

if __name__ == "__main__":
    Fire(main)