Nerfstudio
Nerfstudio Dataset includes 10 in-the-wild captures obtained using either a mobile phone or a mirror-less camera with a fisheye lens. We processed the data using either COLMAP or the Polycam app to obtain camera poses and intrinsic parameters.
Web: https://nerf.studio
Authors: Matthew Tancik, Ethan Weber, Evonne Ng, Ruilong Li, Brent Yi, Justin Kerr, Terrance Wang, Alexander Kristoffersen, Jake Austin, Kamyar Salahi, Abhik Ahuja, David McAllister, Angjoo Kanazawa
Method | PSNR | SSIM | LPIPS | Time | GPU Mem. | |
|---|---|---|---|---|---|---|
| Zip-NeRF | 5h 21m 41s | 26.19 GB | ||||
| Instant NGP | 4m 33s | 4.22 GB | ||||
| NerfStudio | 13m 30s | 4.77 GB | ||||
| Gaussian Splatting | N/A | N/A | ||||
| Mip-Splatting | N/A | N/A |
PSNR
Peak Signal to Noise Ratio. The higher the better.
Method | Egypt | person | kitchen | plane | dozer | floating tree | aspen | stump | sculpture | Giannini Hall |
|---|---|---|---|---|---|---|---|---|---|---|
| Zip-NeRF | ||||||||||
| Instant NGP | ||||||||||
| NerfStudio | ||||||||||
| Gaussian Splatting | ||||||||||
| Mip-Splatting |
SSIM
Structural Similarity Index. The higher the better. The implementation matches JAX's SSIM and torchmetrics's SSIM (with default parameters).
Method | Egypt | person | kitchen | plane | dozer | floating tree | aspen | stump | sculpture | Giannini Hall |
|---|---|---|---|---|---|---|---|---|---|---|
| Zip-NeRF | ||||||||||
| NerfStudio | ||||||||||
| Instant NGP | ||||||||||
| Gaussian Splatting | ||||||||||
| Mip-Splatting |
LPIPS
Learned Perceptual Image Patch Similarity. The lower the better. The implementation uses VGG backbone and matches lpips pip package with checkpoint version 0.1
Method | Egypt | person | kitchen | plane | dozer | floating tree | aspen | stump | sculpture | Giannini Hall |
|---|---|---|---|---|---|---|---|---|---|---|
| Zip-NeRF | ||||||||||
| NerfStudio | ||||||||||
| Instant NGP | ||||||||||
| Gaussian Splatting | ||||||||||
| Mip-Splatting |