Novel View Synthesis on ShapeNet (test)
28.88PSNRPixelNeRF
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PixelNeRFNumber of input views=22022.06 | 28.88 | 0.936 | 0.076 | 1.1 | — | |
| FWD-UNumber of input views=22022.06 | 28.43 | 0.931 | 0.043 | 336 | — | |
| SODAApproach=Diffusion-based (Geometry-Free)2023.11 | 27.42 | 0.947 | 0.039 | — | 0.74 | |
| PixelNeRFNumber of input views=12022.06 | 26.8 | 0.91 | 0.108 | 1.2 | — | |
| SODA (shorter sampling)Sampling steps=20, Approach=Diffusion-based (Geometry-Free)2023.11 | 26.71 | 0.946 | 0.046 | — | 1.31 | |
| FWD-UNumber of input views=12022.06 | 26.66 | 0.911 | 0.055 | 364 | — | |
| PixelNeRFApproach=Geometry-Aware2023.11 | 26.58 | 0.94 | 0.073 | — | 25.34 | |
| SRTApproach=Geometry-Free2023.11 | 26.31 | 0.934 | 0.073 | — | 17.96 | |
| SODA w/o modulationFeature modulation=false, Approach=Diffusion-based (Geometry-Free)2023.11 | 25.02 | 0.944 | 0.048 | — | 2.97 | |
| Vanilla Auto-EncoderApproach=Geometry-Free2023.11 | 24.71 | 0.943 | 0.096 | — | 42.23 | |
| NeRF-VAEApproach=Geometry-Aware2023.11 | 24.6 | 0.915 | 0.101 | — | 45.79 | |
| SODA w/o bottleneckBottleneck=false, Approach=Diffusion-based (Geometry-Free)2023.11 | 24.31 | 0.949 | 0.051 | — | 2.83 | |
| SRNNumber of input views=12022.06 | 23.28 | 0.849 | 0.139 | 24 | — | |
| DVRNumber of input views=12022.06 | 22.7 | 0.86 | 0.13 | 1.5 | — | |
| DALL-E 2Approach=Diffusion-based (Geometry-Free)2023.11 | 18.75 | 0.823 | 0.101 | — | 6.54 | |
| PaletteApproach=Diffusion-based (Geometry-Free)2023.11 | 14.44 | 0.582 | 0.177 | — | 6.93 |