-
Notifications
You must be signed in to change notification settings - Fork 0
/
consistency-coarse.yaml
executable file
·135 lines (124 loc) · 3.24 KB
/
consistency-coarse.yaml
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
name: "consistency-coarse"
tag: "${rmspace:${system.prompt_processor.prompt},_}"
exp_root_dir: "outputs"
seed: 719
data_type: "random-camera-datamodule"
data:
batch_size: 8
width: 64
height: 64
camera_distance_range: [1.5, 2.0]
elevation_range: [-10, 45]
light_sample_strategy: "magic3d"
eval_camera_distance: 2.0
eval_fovy_deg: 45.
batch_uniform_azimuth: true
use_n_views: True
n_views: 4
azimuth_range: [-135, 225]
system_type: "consistency-system"
system:
refinement: false
geometry_type: "implicit-volume"
geometry:
radius: 2.
normal_type: analytic
# coarse to fine hash grid encoding
# to ensure smooth analytic normals
pos_encoding_config:
otype: ProgressiveBandHashGrid
n_levels: 16
n_features_per_level: 2
log2_hashmap_size: 19
base_resolution: 16
per_level_scale: 1.447269237440378 # max resolution 4096
start_level: 8 # resolution ~200
start_step: 0
update_steps: 250
density_bias: "blob_magic3d"
density_activation: softplus
density_blob_scale: 10.
density_blob_std: 0.5
isosurface_resolution: 128
isosurface_threshold: auto
isosurface_coarse_to_fine: true
material_type: "diffuse-with-point-light-material"
material:
ambient_only_steps: 1000
textureless_prob: 0.5
ambient_light_color: [1.0, 1.0, 1.0]
diffuse_light_color: [0.0, 0.0, 0.0]
soft_shading: false
albedo_activation: sigmoid
background_type: "neural-environment-map-background"
background:
color_activation: sigmoid
random_aug: true
renderer_type: "nerf-volume-renderer"
renderer:
radius: ${system.geometry.radius}
num_samples_per_ray: 512
return_comp_normal: true # for geometry training
prompt_processor_type: "stable-diffusion-prompt-processor"
prompt_processor:
pretrained_model_name_or_path: "stabilityai/stable-diffusion-2-1-base"
prompt: ???
guidance_type: "stable-diffusion-guidance"
guidance:
pretrained_model_name_or_path: "stabilityai/stable-diffusion-2-1-base"
weighting_strategy: uniform
guidance_scale: [0, 100., 50., 3000]
recon_std_rescale: 0.0
min_sigma_percent: 0.1
max_sigma_percent: 0.98
min_step_percent: 0.1
max_step_percent: 0.98
loss_type: consistency
cur_steps: [0, 700, 100, 3000]
grad_clip: [0, 2.0, 8.0, 3000]
share_noise: true
perturb_indices: sqrt
perturb_factor: 200
end_gap: 100
freq:
guidance_eval: 0
loggers:
wandb:
enable: false
project: "threestudio"
name: None
loss:
lambda_sds: 1.
lambda_orient: [0, 10., 1000., 2000]
lambda_sparsity: 0.
lambda_opaque: 0.
lambda_normal_smoothness_2d: 0.
optimizer:
name: Adan
args:
eps: 1.0e-8
lr: 0.005
weight_decay: 2.0e-8
max_grad_norm: 5.0
foreach: True
fused: False
params:
geometry.encoding:
lr: 0.05
geometry.density_network:
lr: 0.005
geometry.feature_network:
lr: 0.005
background:
lr: 0.005
trainer:
max_steps: 3000
log_every_n_steps: 1
num_sanity_val_steps: 0
val_check_interval: 500
enable_progress_bar: true
precision: 16-mixed
checkpoint:
save_last: true
save_top_k: -1
every_n_train_steps: ${trainer.max_steps}