Research map / 3D vision / NeRF and radiance fields
DeVRF: Fast Deformable Voxel Radiance Fields for Dynamic Scenes
Jia-Wei Liu, Yan-Pei Cao, Weijia Mao, Wenqiao Zhang, David Junhao Zhang, Jussi Keppo, Ying Shan, Xiaohu Qie, Mike Zheng Shou
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 voxel-based canonical and deformation fields with static-to-dynamic training | DeVRF · Synthetic inward-facing scenes | Time: 18.7\mathrm{hrs} -> 8\mathrm{mins} | Improved |
| Add coarse-to-fine training (ablation) | DeVRF · Synthetic inward-facing scenes | PSNR: 22.44 -> 31.97 | Improved |
| Add total variation regularization (ablation) | DeVRF with c2f · Synthetic inward-facing scenes | PSNR: 31.97 -> 32.73 | Improved |
| Add deformation cycle consistency (ablation) | DeVRF with c2f and TV · Synthetic inward-facing scenes | PSNR: 32.73 -> 33.97 | Improved |
| Add optical-flow supervision (ablation) | DeVRF with c2f, TV, and cycle · Synthetic inward-facing scenes | PSNR: 33.97 -> 34.29 | Improved |
| Add coarse-to-fine training (ablation) | DeVRF · Real-world inward-facing | PSNR: 24.56 -> 27.83 | Improved |
| Add total variation regularization (ablation) | DeVRF with c2f · Real-world inward-facing | PSNR: 27.83 -> 29.35 | Improved |
| Add deformation cycle consistency (ablation) | DeVRF with c2f and TV · Real-world inward-facing | PSNR: 29.35 -> 31.56 | Improved |
| Add optical-flow supervision (ablation) | DeVRF with c2f, TV, and cycle · Real-world inward-facing | PSNR: 31.56 -> 31.68 | Improved |
| Train with static multi-view data in addition to dynamic sequences | D-NeRF · Synthetic inward-facing scenes | PSNR: 17.59 -> 31.83 | Improved |
| Train with static multi-view data in addition to dynamic sequences | Nerfies · Synthetic inward-facing scenes | PSNR: 19.45 -> 33.09 | Improved |
| Train with static multi-view data in addition to dynamic sequences | HyperNeRF · Real-world inward-facing | PSNR: 10.39 -> 28.50 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- D-NeRF — D-NeRF: Neural Radiance Fields for Dynamic Scenes (CVPR 2021) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- D-NeRF: Neural Radiance Fields for Dynamic Scenes (CVPR 2021 · similarity 0.92)
- Direct Voxel Grid Optimization: Super-Fast Convergence for Radiance Fields Reconstruction (CVPR 2022 · similarity 0.92)
- Ced-NeRF: A Compact and Efficient Method for Dynamic Neural Radiance Fields (AAAI 2024 · similarity 0.91)
- BLiRF: Bandlimited Radiance Fields for Dynamic Scene Modeling (AAAI 2024 · similarity 0.90)
- CVT-xRF: Contrastive In-Voxel Transformer for 3D Consistent Radiance Fields from Sparse Inputs (CVPR 2024 · similarity 0.89)
- Joint Optimization of Neural Radiance Fields and Continuous Camera Motion from a Monocular Video (CVPR 2025 · similarity 0.89)
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.
