Research map / 3D vision / Face and shape
Neural Articulated Radiance Field
Atsuhiro Noguchi, Xiao Sun, Stephen Lin, Tatsuya Harada
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 selector that masks irrelevant parts (Disentangled instead of Holistic) (ablation) | NARF_H · Single object, novel pose/novel view | PSNR: 23.98 -> 27.24 | Improved |
| add selector that masks irrelevant parts (Disentangled instead of Holistic) (ablation) | NARF_H · Single object, novel pose/same view | PSNR: 24.09 -> 27.93 | Improved |
| share one network with selector instead of part-wise networks (ablation) | NARF_P · Single object, novel pose/novel view | PSNR: 26.50 -> 27.24 | Improved |
| use explicit local-coordinate transformation (RT-NeRF instead of RP-NeRF) (ablation) | RP-NeRF · Rigid bulldozer | Mask L2 error: 4511.0 -> 16.9 | Improved |
| add pose-dependent color input xi (ablation) | RT-NeRF · Rigid bulldozer | Mask L2 error: 29.2 -> 16.9 | Improved |
| add mask loss (ablation) | NARF_H · Single object, novel pose/novel view | PSNR: 20.53 -> 23.98 | Improved |
| add mask loss (ablation) | NARF_D · Single object, novel pose/novel view | PSNR: 25.71 -> 27.24 | Improved |
| add mask loss (ablation) | NARF_D · Single object, same pose/same view | PSNR: 31.39 -> 30.86 | Hurt |
| use softmax instead of sigmoid in selector (ablation) | NARF_D · Single object, novel pose/novel view | PSNR: 26.17 -> 27.24 | Improved |
| use tau=100 instead of tau=0.01 in part-wise softmax (ablation) | NARF_P (width 32) · Single object, novel pose/novel view | PSNR: 23.72 -> 26.24 | Improved |
| use tau=100 instead of tau=1000 (ablation) | NARF_P (width 32) · Single object, novel pose/novel view | PSNR: 26.24 -> 26.08 | No clear effect |
| increase width of part networks from 32 to 256 (ablation) | NARF_P · Single object, novel pose/novel view | PSNR: 25.74 -> 26.50 | Improved |
Closest papers by meaning
- A-NeRF: Articulated Neural Radiance Fields for Learning Human Shape, Appearance, and Pose (NeurIPS 2021 · similarity 0.93)
- D-NeRF: Neural Radiance Fields for Dynamic Scenes (CVPR 2021 · similarity 0.90)
- Joint Optimization of Neural Radiance Fields and Continuous Camera Motion from a Monocular Video (CVPR 2025 · similarity 0.89)
- Language-driven Object Fusion into Neural Radiance Fields with Pose-Conditioned Dataset Updates (CVPR 2024 · similarity 0.88)
- PoseGen: Learning to Generate 3D Human Pose Dataset with NeRF (AAAI 2024 · similarity 0.88)
- Mask-Based Modeling for Neural Radiance Fields (ICLR 2024 · similarity 0.88)
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.
