Novel View Synthesis on Real-world Data (test)
24.64PSNRGPS-Gaussian
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPS-GaussianOptimization Strategy=feed-forward, Inference Engine=TensorRT, Hardware=RTX 3090 GPU, Input Resolution=1024 x 1024, Source Images=22023.12 | 24.64 | 0.917 | 0.088 | 25 | |
| NeRDIllumination=Fixed2020.12 | 23.86 | 0.88 | — | — | |
| NeRFIllumination=Fixed2020.12 | 23.34 | 0.85 | — | — | |
| ENERFOptimization Strategy=feed-forward, Hardware=RTX 3090 GPU, Input Resolution=1024 x 1024, Source Images=22023.12 | 23.26 | 0.893 | 0.118 | 5 | |
| 3D-GSOptimization Strategy=per-subject, Hardware=RTX 3090 GPU, Input Resolution=1024 x 1024, Source Images=22023.12 | 22.97 | 0.839 | 0.125 | — | |
| NeRF-AIllumination=Fixed2020.12 | 22.87 | 0.83 | — | — | |
| FloRenOptimization Strategy=feed-forward, Inference Engine=TensorRT, Hardware=RTX 3090 GPU, Input Resolution=1024 x 1024, Source Images=22023.12 | 22.8 | 0.872 | 0.136 | 15 | |
| IBRNetOptimization Strategy=feed-forward, Hardware=RTX 3090 GPU, Input Resolution=1024 x 1024, Source Images=22023.12 | 22.63 | 0.852 | 0.177 | 0.25 |