Novel View Synthesis on Mip-NeRF 360
30.69PSNRStructGS
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| StructGSMSRN=true2025.03 | 30.69 | 0.928 | 0.036 | — | — | — | — | |
| ImprovedGS+Time=62m10s, #G(10^6)=1.688, Iterations=30,0002026.03 | 29.38 | 0.877 | — | — | — | — | — | |
| Mip-NeRF 3602025.03 | 29.23 | 0.844 | 0.207 | — | — | — | — | |
| StructGSMSRN=false2025.03 | 29.21 | 0.881 | 0.155 | — | — | — | — | |
| Mip-Splatting2025.03 | 29.09 | 0.871 | 0.184 | — | — | — | — | |
| AugSFramework=MCMC, SH Order=32026.02 | 28.96 | 0.849 | 0.17 | — | — | — | — | |
| 3DTSreproduced=true2025.12 | 28.95 | 0.876 | 0.153 | — | — | — | — | |
| AugSFramework=MCMC, SH Order=22026.02 | 28.89 | 0.848 | 0.171 | — | — | — | — | |
| Scaffold-GS2025.03 | 28.84 | 0.848 | 0.22 | — | — | — | — | |
| ImprovedGS+Gaussian Budget=1M, Time=45m33s, #G(10^6)=0.867, Iterations=30,0002026.03 | 28.79 | 0.859 | — | — | — | — | — | |
| UTrice2025.12 | 28.7 | 0.866 | 0.163 | — | — | — | — | |
| 3DGSreproduced=false2025.12 | 28.69 | 0.87 | 0.182 | — | — | — | — | |
| 3DGS2025.03 | 28.69 | 0.87 | 0.182 | — | — | — | — | |
| MCMCGaussian Budget=1M, Time=62m14s, #G(10^6)=1.000, Iterations=30,0002026.03 | 28.64 | 0.848 | — | — | — | — | — | |
| DBSIterations=30k2026.02 | 28.6 | 0.844 | 0.182 | — | — | — | — | |
| 2DGSreproduced=false2025.12 | 28.56 | 0.862 | 0.19 | — | — | — | — | |
| Zip-NeRF2026.02 | 28.54 | 0.828 | 0.189 | — | — | — | — | |
| Zip-NeRF2023.12 | 28.53 | 0.828 | 0.19 | — | — | — | — | |
| Zip-NeRFReal-time=false2023.12 | 28.49 | 0.829 | 0.142 | — | — | — | — | |
| TexGau# Gaussians=500K, Memory (MB)=2352025.12 | 28.46 | 0.834 | 0.182 | — | — | — | — | |
| ASAP-Textured Gaussians# Gaussians=500K, Memory (MB)=1282025.12 | 28.44 | 0.833 | 0.182 | — | — | — | — | |
| AugSFramework=3DGS, SH Order=32026.02 | 28.39 | 0.834 | 0.184 | — | — | — | — | |
| BBSplat# Gaussians=200K, Memory (MB)=7812025.12 | 28.37 | 0.844 | 0.181 | — | — | — | — | |
| 3DGRTreproduced=true2025.12 | 28.32 | 0.859 | 0.235 | — | — | — | — | |
| 3DGS-MCMCFramework=MCMC2026.02 | 28.29 | 0.84 | 0.21 | — | — | — | — | |
| AdamW-GS (MC8)Base Pipeline=3DGS-MCMC, ∆Na=+4.52%, time/mins=39.772026.01 | 28.219 | 0.84 | 0.182 | — | — | — | — | |
| 2DGS# Gaussians=500K, Memory (MB)=1132025.12 | 28.18 | 0.828 | 0.19 | — | — | — | — | |
| Spec-Gaussian2026.02 | 28.18 | 0.835 | 0.176 | — | — | — | — | |
| ADCTime=76m02s, #G(10^6)=2.741, Iterations=30,0002026.03 | 28.1 | 0.848 | — | — | — | — | — | |
| BBSplat# Gaussians=100K, Memory (MB)=3912025.12 | 28.05 | 0.834 | 0.192 | — | — | — | — | |
| Original MCMC∆Na=-3.75%, time/mins=46.812026.01 | 27.948 | 0.833 | 0.199 | — | — | — | — | |
| GaussianSpaStorage=25 MB2024.11 | 27.85 | 0.825 | 0.214 | — | — | — | — | |
| TexGau# Gaussians=200K, Memory (MB)=93.62025.12 | 27.81 | 0.813 | 0.223 | — | — | — | — | |
| ASAP-Textured Gaussians# Gaussians=200K, Memory (MB)=60.72025.12 | 27.79 | 0.81 | 0.22 | — | — | — | — | |
| VoD-3DGS[L]variant=[L]2026.02 | 27.79 | 0.818 | 0.213 | — | — | — | — | |
| HACBitrate version=highrate2026.03 | 27.77 | 0.811 | 0.23 | — | — | 21.87 | — | |
| ContextGSBackbone=Scaffold-GS, Tradeoff=high-rate2024.05 | 27.75 | 0.811 | 0.231 | — | — | 18.41 | — | |
| AdamW-GS (GS7)Base Pipeline=3DGS, ∆Na=-46.9%, time/mins=19.182026.01 | 27.73 | 0.82 | 0.222 | — | — | — | — | |
| Mip-NeRF 360Train=48h, FPS=0.06, Storage=8.6 MB2023.11 | 27.69 | 0.792 | 0.237 | — | — | — | — | |
| Mip-NeRF 360Real-time=false2023.12 | 27.69 | 0.791 | 0.237 | — | — | — | — | |
| Mip-NeRF 3602026.02 | 27.69 | 0.792 | 0.237 | — | — | — | — | |
| Mip-NeRF 3602023.12 | 27.69 | 0.791 | 0.237 | — | — | — | — | |
| AdamW-GS (GS8)Base Pipeline=3DGS, ∆Na=-49.3%, time/mins=18.532026.01 | 27.678 | 0.822 | 0.22 | — | — | — | — | |
| UniSDF2023.12 | 27.67 | 0.808 | 0.213 | — | — | — | — | |
| SuperGau# Gaussians=500K, Memory (MB)=1682025.12 | 27.66 | 0.814 | 0.198 | — | — | — | — | |
| ProGSBitrate version=HR, Level of Detail (LoD)=52026.03 | 27.66 | 0.817 | 0.21 | — | — | 29.81 | — | |
| ContextGSBackbone=Scaffold-GS, Tradeoff=low-rate2024.05 | 27.62 | 0.808 | 0.237 | — | — | 12.68 | — | |
| Mip-NeRF 360Runtime=≈ 48h2023.11 | 27.569 | 0.793 | 0.234 | — | — | — | — | |
| HAC2024.05 | 27.53 | 0.807 | 0.238 | — | — | 15.26 | — | |
| HACBitrate version=lowrate2026.03 | 27.53 | 0.807 | 0.238 | — | — | 15.26 | — | |
| HybridNeRFReal-time=true2023.12 | 27.52 | 0.806 | 0.167 | — | — | — | — | |
| GaussianPOPTraining Iterations=40,0002026.02 | 27.51 | 0.822 | 0.218 | — | — | — | 0.33 | |
| vanilla 3DGS∆Na=3.098m, time/mins=30.582026.01 | 27.506 | 0.815 | 0.216 | — | — | — | — | |
| RePR∆Na=-41.1%, time/mins=26.732026.01 | 27.503 | 0.815 | 0.218 | — | — | — | — | |
| EAGLESBackbone=3DGS2024.05 | 27.5 | 0.806 | 0.252 | — | — | 253.9 | — | |
| 3DGS2026.02 | 27.5 | 0.813 | 0.221 | — | — | — | 3.111 | |
| 2DGS# Gaussians=200K, Memory (MB)=45.02025.12 | 27.5 | 0.804 | 0.235 | — | — | — | — | |
| Scaffold-GS2026.03 | 27.5 | 0.806 | 0.252 | — | — | 253.9 | — | |
| 3DGS2024.05 | 27.49 | 0.813 | 0.222 | — | — | 744.7 | — | |
| 3DGS2026.03 | 27.49 | 0.97 | 0.031 | — | — | 68.46 | — | |
| MaskGaussion∆Na=-53.1%, time/mins=26.82026.01 | 27.485 | 0.815 | 0.219 | — | — | — | — | |
| LP-3DGS-R2026.02 | 27.47 | 0.812 | 0.227 | — | — | — | 1.959 | |
| 3DGS*Train=24m 07s, FPS=120, Storage=746 MB, GPU=NVIDIA A1002023.11 | 27.46 | 0.812 | 0.222 | — | — | — | — | |
| GaussianSpaTraining Iterations=40,0002026.02 | 27.43 | 0.821 | 0.219 | — | — | — | 0.353 | |
| Mini-Splatting2026.02 | 27.4 | 0.821 | 0.219 | — | — | — | 0.559 | |
| EfficientGSStorage=98 MB2024.11 | 27.38 | 0.817 | 0.216 | — | — | — | — | |
| ProGSBitrate version=LR, Level of Detail (LoD)=52026.03 | 27.37 | 0.811 | 0.22 | — | — | 20.61 | — | |
| GaussianPOPTraining Iterations=30,0002026.02 | 27.36 | 0.816 | 0.223 | — | — | — | 0.401 | |
| Mip-NeRF 3602022.11 | 27.325 | 0.7942 | 0.6559 | 0.2438 | 0.1077 | — | — | |
| LightGaussianStorage=42 MB2024.11 | 27.28 | 0.805 | 0.243 | — | — | — | — | |
| LightGaussian2026.03 | 27.28 | 0.805 | 0.243 | — | — | 42 | — | |
| CompGS2026.03 | 27.26 | 0.8 | 0.24 | — | — | 16.5 | — | |
| GaussianSpaTraining Iterations=30,0002026.02 | 27.25 | 0.812 | 0.225 | — | — | — | 0.426 | |
| Exact-NeRF2022.11 | 27.23 | 0.7881 | 0.6569 | 0.2452 | 0.1088 | — | — | |
| EAGLES2026.02 | 27.23 | 0.809 | 0.238 | — | — | — | 1.33 | |
| EAGLES2026.03 | 27.23 | 0.81 | 0.24 | — | — | 54 | — | |
| 3DGSTrain=41m 33s, FPS=134, Storage=734 MB, GPU=NVIDIA A60002023.11 | 27.21 | 0.815 | 0.214 | — | — | — | — | |
| 3DGSRuntime=≈ 35m2023.11 | 27.21 | 0.815 | 0.214 | — | — | — | — | |
| 3DGS2026.02 | 27.21 | 0.815 | 0.214 | — | — | — | — | |
| 3D Gaussian splattingReal-time=true2023.12 | 27.16 | 0.805 | 0.153 | — | — | — | — | |
| Navaneet et al.Backbone=3DGS2024.05 | 27.16 | 0.808 | 0.228 | — | — | 50.3 | — | |
| Compact3D2026.03 | 27.16 | 0.808 | 0.228 | — | — | 50.3 | — | |
| Scaffold-GS2024.05 | 27.15 | 0.808 | 0.238 | — | — | 68.89 | — | |
| LP-3DGS-M2026.02 | 27.12 | 0.805 | 0.239 | — | — | — | 1.866 | |
| CompGS2026.02 | 27.12 | 0.806 | 0.24 | — | — | — | 0.845 | |
| ASAP-Textured Gaussians# Gaussians=100K, Memory (MB)=34.62025.12 | 27.11 | 0.785 | 0.259 | — | — | — | — | |
| R-3DGS2026.03 | 27.1 | 0.809 | 0.226 | — | — | 29 | — | |
| TexGau# Gaussians=100K, Memory (MB)=46.92025.12 | 27.09 | 0.787 | 0.267 | — | — | — | — | |
| C3DGSTrain=33m 06s, FPS=128, Storage=48.8 MB, GPU=NVIDIA A1002023.11 | 27.08 | 0.798 | 0.247 | — | — | — | — | |
| Compact3DGSBackbone=3DGS2024.05 | 27.08 | 0.798 | 0.247 | — | — | 48.8 | — | |
| Compact3DGS2026.02 | 27.08 | 0.798 | 0.247 | — | — | — | 1.388 | |
| Lee et al.2026.03 | 27.08 | 0.798 | 0.238 | — | — | 48.8 | — | |
| SOG2026.03 | 27.08 | 0.799 | 0.23 | — | — | 38.4 | — | |
| RDO-Gaussian2026.03 | 27.05 | 0.802 | 0.239 | — | — | 23.46 | — | |
| C3DGS+PPStorage=29.1 MB2023.11 | 27.03 | 0.797 | 0.247 | — | — | — | — | |
| LightGaussianBackbone=3DGS2024.05 | 27 | 0.799 | 0.249 | — | — | 44.54 | — | |
| Compressed3DBackbone=3DGS2024.05 | 26.98 | 0.801 | 0.238 | — | — | 28.8 | — | |
| Compressed3D2026.03 | 26.98 | 0.801 | 0.238 | — | — | 28.8 | — | |
| MaskGaussianTraining Iterations=40,0002026.02 | 26.95 | 0.802 | 0.252 | — | — | — | 0.43 | |
| LightGaussianTraining Iterations=40,0002026.02 | 26.81 | 0.788 | 0.273 | — | — | — | 0.451 |