-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathtrain.sh
34 lines (30 loc) · 3.41 KB
/
train.sh
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
# For real scenes, add this command to eliminate the impact of background.
# env_scope_center and env_scope_radius represent a sphere domain where the environment light takes effect.
# https://github.com/gapszju/3DGS-DR
#############################################################################################
#### For better results, more training steps (e.g., 60k steps, like 3DGS-DR) can be set, ###
#### although we observe that the model already yields good results after 30k steps. ###
#############################################################################################
# python train-real.py -s data/ref_real/sedan -r 8 --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --env_scope_center -0.032 0.808 0.751 --env_scope_radius 2.138 --init_until_iter 700 --xyz_axis 2.0 1.0 0.0
# python train-real.py -s data/ref_real/gardenspheres -r 6 --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --env_scope_center -0.2270 1.9700 1.7740 --env_scope_radius 0.974 --init_until_iter 700 --xyz_axis 2.0 1.0 0.0
# python train-real.py -s data/ref_real/toycar -r 6 --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --env_scope_center 0.486 1.108 3.72 --env_scope_radius 2.507 --init_until_iter 1500 --xyz_axis 0.0 2.0 1.0
# python train.py -s data/refnerf/helmet --eval --run_dim 256 --albedo_bias 0
# python train.py -s data/refnerf/car --eval --run_dim 256 --albedo_bias 0
# python train.py -s data/refnerf/ball --eval --run_dim 256 --albedo_bias 0
# python train.py -s data/refnerf/teapot --eval --run_dim 256 --albedo_bias 0
# python train.py -s data/refnerf/coffee --eval --run_dim 256 --albedo_bias 0 --albedo_lr 0.002
# python train.py -s data/refnerf/toaster --eval --run_dim 256 --albedo_bias 0
# python train-NeRF.py -s data/nerf_synthetic/ship --eval --run_dim 64 --albedo_bias 0 --gsrgb_loss --albedo_lr 0.002
# python train-NeRF.py -s data/nerf_synthetic/ficus --eval --run_dim 64 --albedo_bias 0 --gsrgb_loss --albedo_lr 0.002
# python train-NeRF.py -s data/nerf_synthetic/lego --eval --run_dim 64 --albedo_bias 0 --gsrgb_loss --albedo_lr 0.002
# python train-NeRF.py -s data/nerf_synthetic/mic --eval --run_dim 64 --albedo_bias 0 --gsrgb_loss --albedo_lr 0.002
# python train-NeRF.py -s data/nerf_synthetic/hotdog --eval --run_dim 64 --albedo_bias 0 --gsrgb_loss --albedo_lr 0.002
# python train-NeRF.py -s data/nerf_synthetic/chair --eval --run_dim 64 --albedo_bias 0 --gsrgb_loss --albedo_lr 0.002
# python train-NeRF.py -s data/nerf_synthetic/materials --eval --run_dim 256 --albedo_bias 0
# python train-NeRF.py -s data/nerf_synthetic/drums --eval --run_dim 64 --albedo_bias 0 --albedo_lr 0.002
# python train-NeRO.py -s data/GlossySynthetic/bell_blender --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --init_until_iter 3000
# python train-NeRO.py -s data/GlossySynthetic/tbell_blender --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --init_until_iter 3000
# python train-NeRO.py -s data/GlossySynthetic/potion_blender --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --init_until_iter 3000
# python train-NeRO.py -s data/GlossySynthetic/teapot_blender --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --init_until_iter 3000
# python train-NeRO.py -s data/GlossySynthetic/luyu_blender --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --init_until_iter 3000
# python train-NeRO.py -s data/GlossySynthetic/cat_blender --eval --run_dim 256 --albedo_bias 2 --albedo_lr 0.0005 --init_until_iter 3000