Research map / Image generation / Image restoration
DynIBaR: Neural Dynamic Image-Based Rendering
Zhengqi Li, Qianqian Wang, Forrester Cole, Richard Tucker, Noah Snavely
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use DynIBaR instead of the strongest baseline, NSFF | Dynamic scene view synthesis · Nvidia Dynamic Scene Dataset, Full | LPIPS: 0.062 -> 0.027 | Improved |
| Use DynIBaR instead of the strongest baseline, NSFF | Dynamic scene view synthesis · Nvidia Dynamic Scene Dataset, Dynamic Only | LPIPS: 0.144 -> 0.062 | Improved |
| Use DynIBaR instead of the strongest baseline, NSFF | Dynamic scene view synthesis · UCSD Dynamic Scenes Dataset, Full | PSNR: 31.75 -> 36.47 | Improved |
| Use DynIBaR instead of the strongest baseline, NSFF | Dynamic scene view synthesis · UCSD Dynamic Scenes Dataset, Full | LPIPS: 0.034 -> 0.014 | Improved |
| Use DynIBaR instead of the strongest baseline, DVS | Dynamic scene view synthesis · UCSD Dynamic Scenes Dataset, Dynamic Only | PSNR: 26.57 -> 28.01 | Improved |
| Use DynIBaR instead of the strongest baseline, DVS | Dynamic scene view synthesis · UCSD Dynamic Scenes Dataset, Dynamic Only | LPIPS: 0.096 -> 0.042 | Improved |
| Add cross-time temporal consistency (ablation) | DynIBaR · Nvidia Dynamic Scene Dataset, Full | PSNR: 27.57 -> 30.77 | Improved |
| Use motion trajectories rather than scene flow fields for feature aggregation (ablation) | DynIBaR · Nvidia Dynamic Scene Dataset, Full | PSNR: 29.42 -> 30.77 | Improved |
| Use motion trajectories rather than multiple scene-flow vectors to nearby times (ablation) | DynIBaR · Nvidia Dynamic Scene Dataset, Full | PSNR: 29.59 -> 30.77 | Improved |
| Add a time-invariant static scene model (ablation) | DynIBaR · Nvidia Dynamic Scene Dataset, Full | PSNR: 28.19 -> 30.77 | Improved |
| Add masked reconstruction supervision from motion segmentation masks (ablation) | DynIBaR · Nvidia Dynamic Scene Dataset, Full | PSNR: 29.95 -> 30.77 | Improved |
| Add regularization loss (ablation) | DynIBaR · Nvidia Dynamic Scene Dataset, Full | PSNR: 29.46 -> 30.77 | Improved |
Closest papers by meaning
- Generalizable Patch-Based Neural Rendering (ECCV 2022 · similarity 0.86)
- DynMF: Neural Motion Factorization for Real-time Dynamic View Synthesis with 3D Gaussian Splatting (ECCV 2024 · similarity 0.85)
- Geometry-aware Reconstruction and Fusion-refined Rendering for Generalizable Neural Radiance Fields (CVPR 2024 · similarity 0.83)
- Ced-NeRF: A Compact and Efficient Method for Dynamic Neural Radiance Fields (AAAI 2024 · similarity 0.82)
- DyST: Towards Dynamic Neural Scene Representations on Real-World Videos (ICLR 2024 · similarity 0.81)
- DyBluRF: Dynamic Neural Radiance Fields from Blurry Monocular Video (CVPR 2024 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
