We compare our results with some popular frameworks and official releases in terms of speed.
- Python 3.7
- PyTorch 1.10
- CUDA 11.1
- CUDNN 8.1.0
test data |
PSNR |
SSIM |
NeRF |
XRNeRF |
NeRF |
XRNeRF |
blender_chair |
33.927 | 34.528 | 0.967 | 0.985 |
blender_drums |
25.600 | 25.685 | 0.925 | 0.946 |
blender_ficus |
30.13 | 29.300 | 0.964 | 0.972 |
blender_hotdog |
36.18 | 35.905 | 0.974 | 0.985 |
blender_materials |
29.62 | 29.014 | 0.949 | 0.967 |
blender_mic |
32.58 | 32.95 | 0.980 | 0.986 |
blender_ship |
28.65 | 29.46 | 0.856 | 0.932 |
llff_fern |
25.17 | 26.277 | 0.792 | 0.892 |
llff_flower |
27.40 | 26.592 | 0.827 | 0.884 |
llff_fortress |
31.16 | 31.485 | 0.881 | 0.952 |
llff_horns |
27.45 | 26.162 | 0.828 | 0.895 |
llff_leaves |
20.92 | 19.749 | 0.690 | 0.668 |
test data |
PSNR |
SSIM |
elapsed_time(ms) |
KiloNeRF |
XRNeRF |
KiloNerf |
XRNeRF |
KiloNerf |
XRNeRF |
nsvf_Synthetic_NeRF_chair |
33.044 | 33.037 | 0.971 | 0.979 | 384.98 | 407.78 |
nsvf_Synthetic_NeRF_drums |
25.327 | 25.308 | 0.931 | 0.949 | 413.03 | 353.62 |
nsvf_Synthetic_NeRF_ficus |
30.1 | 30.176 | 0.967 | 0.975 | 351.04 | 337.22 |
nsvf_Synthetic_NeRF_hotdog |
32.316 | 33.408 | 0.974 | 0.986 | 484.22 | 491.49 |
nsvf_Synthetic_NeRF_lego |
33.398 | 33.381 | 0.971 | 0.982 | 379.1 | 365.16 |
nsvf_Synthetic_NeRF_materials |
29.193 | 29.175 | 0.951 | 0.966 | 380.28 | 358.57 |
nsvf_Synthetic_NeRF_mic |
33.186 | 33.346 | 0.982 | 0.987 | 370.31 | 346.71 |
nsvf_Synthetic_NeRF_ship |
28.892 | 29.295 | 0.874 | 0.933 | 491.92 | 488.35 |
Average |
30.68 | 30.89102 | 0.9526 | 0.9697 | 406.86 | 393.61 |
MultiScale Blender |
PSNR |
800x800 |
400x400 |
200x200 |
100x100 |
Jax |
XRNeRF |
Jax |
XRNeRF |
Jax |
XRNeRF |
Jax |
XRNeRF |
blender_ship |
29.599 | 28.522 | 31.955 | 30.754 | 33.845 | 32.848 | 34.868 | 33.754 |
blender_mic |
33.739 | 32.478 | 36.353 | 35.008 | 38.837 | 37.958 | 39.011 | 38.064 |
blender_materials |
30.128 | 29.278 | 31.424 | 30.505 | 33.163 | 32.192 | 34.174 | 33.122 |
blender_lego |
33.971 | 32.803 | 35.248 | 34.123 | 35.796 | 34.848 | 35.223 | 34.382 |
blender_hotdog |
36.457 | 35.803 | 38.382 | 37.631 | 39.831 | 39.096 | 39.935 | 39.038 |
blender_ficus |
31.490 | 29.222 | 32.267 | 30.093 | 33.255 | 31.655 | 33.606 | 31.785 |
blender_drums |
25.297 | 24.790 | 26.463 | 26.020 | 27.808 | 27.510 | 28.791 | 28.369 |
blender_chair |
33.351 | 32.429 | 36.517 | 35.618 | 38.056 | 37.342 | 37.950 | 37.257 |
Average |
31.754 | 30.666 | 33.576 | 32.469 | 35.074 | 34.181 | 35.445 | 34.472 |
test data |
PSNR |
InstantNGP |
XRNeRF |
blender_chair |
32.927 | 32.71 |
blender_drums |
26.02 | 26.9 |
blender_ficus |
33.51 | 33.97 |
blender_hotdog |
37.40 | 37.17 |
blender_lego |
36.39 | 35.1 |
blender_materials |
29.78 | 30.73 |
blender_mic |
36.22 | 34.05 |
blender_ship |
31.1 | 30.0 |
average |
32.92 | 32.58 |
test data |
PSNR |
SSIM |
Neural Body |
XRNeRF |
Neural Body |
XRNeRF |
313 |
35.21 | 37.76 | 0.985 | 0.993 |
315 |
33.07 | 35.99 | 0.988 | 0.992 |
377 |
33.86 | 33.86 | 0.985 | 0.986 |
386 |
36.07 | 34.24 | 0.984 | 0.984 |
387 |
31.39 | 31.99 | 0.975 | 0.979 |
390 |
34.48 | 35.45 | 0.980 | 0.984 |
392 |
35.76 | 35.11 | 0.984 | 0.986 |
393 |
33.24 | 33.50 | 0.979 | 0.985 |
394 |
34.31 | 35.61 | 0.980 | 0.984 |
test data (Novel pose) |
PSNR |
SSIM |
Animatable NeRF |
XRNeRF |
Animatable NeRF |
XRNeRF |
S1 |
30.11 | 31.98 | 0.981 | 0.984 |
S5 |
32.60 | 33.25 | 0.987 | 0.990 |
S6 |
29.49 | 30.12 | 0.972 | 0.974 |
S7 |
31.54 | 34.47 | 0.984 | 0.988 |
S8 |
30.77 | 32.01 | 0.983 | 0.985 |
S9 |
31.94 | 28.61 | 0.980 | 0.976 |
S11 |
33.12 | 33.43 | 0.986 | 0.986 |
test data |
PSNR |
SSIM |
GNR |
XRNeRF |
GNR |
XRNeRF |
amanda |
23.62 | 25.35 | 0.93 | 0.95 |
barry |
29.28 | 30.71 | 0.94 | 0.95 |
fuzhizhi |
21.96 | 21.42 | 0.90 | 0.89 |
jinyutong |
23.90 | 24.08 | 0.90 | 0.91 |
joseph |
26.30 | 24.46 | 0.94 | 0.92 |
maria |
21.51 | 23.69 | 0.90 | 0.90 |
mahaoran |
28.41 | 30.93 | 0.93 | 0.94 |
natacha |
28.71 | 27.98 | 0.91 | 0.91 |
soufianou |
27.64 | 28.83 | 0.93 | 0.93 |
zhuna |
25.40 | 24.32 | 0.93 | 0.92 |