forked from ashawkey/torch-ngp
-
Notifications
You must be signed in to change notification settings - Fork 0
/
main_tensoRF.py
171 lines (134 loc) · 8.48 KB
/
main_tensoRF.py
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
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
import argparse
from nerf.gui import NeRFGUI
from nerf.provider import NeRFDataset
from tensoRF.utils import *
# torch.autograd.set_detect_anomaly(True)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('path', type=str)
parser.add_argument('-O', action='store_true', help="equals --fp16 --cuda_ray --preload")
parser.add_argument('--test', action='store_true', help="test mode")
parser.add_argument('--workspace', type=str, default='workspace')
parser.add_argument('--seed', type=int, default=0)
### training options
parser.add_argument('--iters', type=int, default=30000, help="training iters")
parser.add_argument('--lr0', type=float, default=2e-2, help="initial learning rate for embeddings")
parser.add_argument('--lr1', type=float, default=1e-3, help="initial learning rate for networks")
parser.add_argument('--ckpt', type=str, default='latest')
parser.add_argument('--num_rays', type=int, default=4096, help="num rays sampled per image for each training step")
parser.add_argument('--cuda_ray', action='store_true', help="use CUDA raymarching instead of pytorch")
parser.add_argument('--max_steps', type=int, default=1024,
help="max num steps sampled per ray (only valid when using --cuda_ray)")
parser.add_argument('--num_steps', type=int, default=512,
help="num steps sampled per ray (only valid when NOT using --cuda_ray)")
parser.add_argument('--update_extra_interval', type=int, default=16,
help="iter interval to update extra status (only valid when using --cuda_ray)")
parser.add_argument('--upsample_steps', type=int, default=0,
help="num steps up-sampled per ray (only valid when NOT using --cuda_ray)")
parser.add_argument('--max_ray_batch', type=int, default=4096,
help="batch size of rays at inference to avoid OOM (only valid when NOT using --cuda_ray)")
parser.add_argument('--l1_reg_weight', type=float, default=1e-4)
parser.add_argument('--patch_size', type=int, default=1,
help="[experimental] render patches in training, so as to apply LPIPS loss. 1 means disabled, use [64, 32, 16] to enable")
### network backbone options
parser.add_argument('--fp16', action='store_true', help="use amp mixed precision training")
parser.add_argument('--cp', action='store_true', help="use TensorCP")
parser.add_argument('--resolution0', type=int, default=128)
parser.add_argument('--resolution1', type=int, default=300)
parser.add_argument("--upsample_model_steps", type=int, action="append", default=[2000, 3000, 4000, 5500, 7000])
### dataset options
parser.add_argument('--color_space', type=str, default='srgb', help="Color space, supports (linear, srgb)")
parser.add_argument('--preload', action='store_true',
help="preload all data into GPU, accelerate training but use more GPU memory")
# (the default value is for the fox dataset)
parser.add_argument('--bound', type=float, default=2,
help="assume the scene is bounded in box[-bound, bound]^3, if > 1, will invoke adaptive ray marching.")
parser.add_argument('--scale', type=float, default=0.33, help="scale camera location into box[-bound, bound]^3")
parser.add_argument('--offset', type=float, nargs='*', default=[0, 0, 0], help="offset of camera location")
parser.add_argument('--dt_gamma', type=float, default=1 / 128,
help="dt_gamma (>=0) for adaptive ray marching. set to 0 to disable, >0 to accelerate rendering (but usually with worse quality)")
parser.add_argument('--min_near', type=float, default=0.2, help="minimum near distance for camera")
parser.add_argument('--density_thresh', type=float, default=10, help="threshold for density grid to be occupied")
parser.add_argument('--bg_radius', type=float, default=-1,
help="if positive, use a background model at sphere(bg_radius)")
### GUI options
parser.add_argument('--gui', action='store_true', help="start a GUI")
parser.add_argument('--W', type=int, default=1920, help="GUI width")
parser.add_argument('--H', type=int, default=1080, help="GUI height")
parser.add_argument('--radius', type=float, default=5, help="default GUI camera radius from center")
parser.add_argument('--fovy', type=float, default=50, help="default GUI camera fovy")
parser.add_argument('--max_spp', type=int, default=64, help="GUI rendering max sample per pixel")
### experimental
parser.add_argument('--error_map', action='store_true', help="use error map to sample rays")
parser.add_argument('--clip_text', type=str, default='', help="text input for CLIP guidance")
parser.add_argument('--rand_pose', type=int, default=-1,
help="<0 uses no rand pose, =0 only uses rand pose, >0 sample one rand pose every $ known poses")
opt = parser.parse_args()
if opt.O:
opt.fp16 = True
opt.cuda_ray = True
opt.preload = True
if opt.patch_size > 1:
opt.error_map = False # do not use error_map if use patch-based training
# assert opt.patch_size > 16, "patch_size should > 16 to run LPIPS loss."
assert opt.num_rays % (opt.patch_size ** 2) == 0, "patch_size ** 2 should be dividable by num_rays."
print(opt)
seed_everything(opt.seed)
if opt.cp:
assert opt.bg_radius <= 0, "background model is not implemented for --cp"
from tensoRF.network_cp import NeRFNetwork
else:
from tensoRF.network import NeRFNetwork
model = NeRFNetwork(
resolution=[opt.resolution0] * 3,
bound=opt.bound,
cuda_ray=opt.cuda_ray,
density_scale=1,
min_near=opt.min_near,
density_thresh=opt.density_thresh,
bg_radius=opt.bg_radius,
)
print(model)
criterion = torch.nn.MSELoss(reduction='none')
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
if opt.test:
trainer = Trainer('ngp', opt, model, device=device, workspace=opt.workspace, criterion=criterion, fp16=opt.fp16,
metrics=[PSNRMeter()], use_checkpoint=opt.ckpt)
if opt.gui:
gui = NeRFGUI(opt, trainer)
gui.render()
else:
test_loader = NeRFDataset(opt, device=device, type='test').dataloader()
if test_loader.has_gt:
trainer.evaluate(test_loader) # blender has gt, so evaluate it.
trainer.test(test_loader, write_video=True) # test and save video
# trainer.save_mesh(resolution=256, threshold=0.1)
else:
optimizer = lambda model: torch.optim.Adam(model.get_params(opt.lr0, opt.lr1), betas=(0.9, 0.99), eps=1e-15)
train_loader = NeRFDataset(opt, device=device, type='train').dataloader()
# decay to 0.1 * init_lr at last iter step
scheduler = lambda optimizer: optim.lr_scheduler.LambdaLR(optimizer,
lambda iter: 0.1 ** min(iter / opt.iters, 1))
trainer = Trainer('ngp', opt, model, device=device, workspace=opt.workspace, optimizer=optimizer,
criterion=criterion, ema_decay=None, fp16=opt.fp16, lr_scheduler=scheduler,
scheduler_update_every_step=True, metrics=[PSNRMeter()], use_checkpoint=opt.ckpt,
eval_interval=50)
# calc upsample target resolutions
upsample_resolutions = (np.round(np.exp(
np.linspace(np.log(opt.resolution0), np.log(opt.resolution1), len(opt.upsample_model_steps) + 1)))).astype(
np.int32).tolist()[1:]
print('upsample_resolutions:', upsample_resolutions)
trainer.upsample_resolutions = upsample_resolutions
if opt.gui:
gui = NeRFGUI(opt, trainer, train_loader)
gui.render()
else:
valid_loader = NeRFDataset(opt, device=device, type='val', downscale=1).dataloader()
max_epoch = np.ceil(opt.iters / len(train_loader)).astype(np.int32)
trainer.train(train_loader, valid_loader, max_epoch)
# also test
test_loader = NeRFDataset(opt, device=device, type='test').dataloader()
if test_loader.has_gt:
trainer.evaluate(test_loader) # blender has gt, so evaluate it.
trainer.test(test_loader, write_video=True) # test and save video
# trainer.save_mesh(resolution=256, threshold=0.1)