Novel View Synthesis on Deep Blending (PSNR, SSIM, LPIPS, #G/M)
30.18PSNRGaussianPOP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GaussianPOPTraining Iterations=40,0002026.02 | 30.18 | 0.912 | 0.249 | 0.208 | |
| GaussianSpaTraining Iterations=40,0002026.02 | 30.11 | 0.911 | 0.25 | 0.216 | |
| Mini-Splatting2026.02 | 30.05 | 0.909 | 0.254 | 0.397 | |
| GaussianSpaTraining Iterations=30,0002026.02 | 29.98 | 0.91 | 0.252 | 0.262 | |
| GaussianPOPTraining Iterations=30,0002026.02 | 29.95 | 0.909 | 0.252 | 0.275 | |
| CompGS2026.02 | 29.9 | 0.907 | 0.251 | 0.55 | |
| EAGLES2026.02 | 29.86 | 0.91 | 0.25 | 1.19 | |
| MaskGaussianTraining Iterations=40,0002026.02 | 29.8 | 0.905 | 0.257 | 0.342 | |
| Compact3DGS2026.02 | 29.79 | 0.901 | 0.258 | 1.06 | |
| 3DGS2026.02 | 29.42 | 0.9 | 0.25 | 2.78 | |
| LightGaussianTraining Iterations=40,0002026.02 | 27.57 | 0.824 | 0.298 | 0.302 |