Single-image novel view synthesis on DTU MVS 15 scenes (test)
16.048PSNRpixelNeRF
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| pixelNeRFFine-tuned=LMSE2022.12 | 16.048 | 0.564 | 0.515 | |
| pixelNeRFFine-tuned=false2022.12 | 15.55 | 0.537 | 0.535 | |
| NeRDiDescription=Applied to test scenes without additional fine-tuning on DTU training set2022.12 | 14.472 | 0.465 | 0.421 | |
| DietPixelNeRFDescription=Fine-tuned model with CLIP feature consistency loss2022.12 | 14.242 | 0.481 | 0.487 | |
| NeRFDescription=Standard NeRF implementation2022.12 | 8 | 0.286 | 0.703 |