Research map / 3D vision / NeRF and radiance fields
Non-Rigid Neural Radiance Fields: Reconstruction and Novel View Synthesis of a Dynamic Scene From Monocular Video
Edgar Tretschk, Ayush Tewari, Vladislav Golyanik, Michael Zollhöfer, Christoph Lassner, Christian Theobalt
Read the paperCheck your idea against 137,217 papers
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 |
|---|---|---|---|
| add the divergence loss (ablation) | NR-NeRF · Dynamic scenes, novel views | foreground stability (qualitative) | Improved |
| add the rigidity network and offsets losses together (ablation) | NR-NeRF · Dynamic scenes, novel views | background stability (std of pixel colour over time) | Improved |
| apply offsets loss to raw offsets and rigidity separately rather than to masked offsets (ablation) | NR-NeRF · Dynamic scenes, novel views | background stability | Improved |
| add the rigidity network (ablation) | NR-NeRF · Dynamic scenes, novel views | background stability | Improved |
| deform with ray bending instead of conditioning the volume on latent code | NR-NeRF vs naive NR-NeRF · Dynamic scenes (18 scenes, test frames) | SSIM: 0.738 -> 0.758 | Improved |
| ray bending instead of naive latent conditioning | NR-NeRF vs naive NR-NeRF · Dynamic scenes (18 scenes, test frames) | PSNR: 25.83 -> 24.70 | Hurt |
| ray bending instead of naive latent conditioning | NR-NeRF vs naive NR-NeRF · Dynamic scenes (18 scenes, test frames) | LPIPS: 0.226 -> 0.197 | Improved |
| deform with ray bending vs rigid NeRF (view cond.) | NR-NeRF · Dynamic scenes (18 scenes, test frames) | PSNR: 22.24 -> 24.70 | Improved |
| add approximate view conditioning | NR-NeRF · Dynamic scenes (18 scenes, test frames) | PSNR: 24.70 -> 25.15 | Improved |
| add exact view conditioning | NR-NeRF · Dynamic scenes (18 scenes, test frames) | LPIPS: 0.197 -> 0.190 | Improved |
| condition on viewing direction in multi-view data | NR-NeRF · Multi-view capture | smoke-like artifacts (qualitative) | Improved |
| use ray bending vs Neural Volumes on monocular input | NR-NeRF · Dynamic scenes (18 scenes, test frames) | PSNR: 14.13 -> 24.70 | Improved |
Closest papers by meaning
- D-NeRF: Neural Radiance Fields for Dynamic Scenes (CVPR 2021 · similarity 0.92)
- Joint Optimization of Neural Radiance Fields and Continuous Camera Motion from a Monocular Video (CVPR 2025 · similarity 0.92)
- NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis (ECCV 2020 · similarity 0.92)
- BLiRF: Bandlimited Radiance Fields for Dynamic Scene Modeling (AAAI 2024 · similarity 0.91)
- CF-NeRF: Camera Parameter Free Neural Radiance Fields with Incremental Learning (AAAI 2024 · similarity 0.90)
- NeRFPrior: Learning Neural Radiance Field as a Prior for Indoor Scene Reconstruction (CVPR 2025 · similarity 0.90)
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.
