Research map / 3D vision / NeRF and radiance fields
Strivec: Sparse Tri-Vector Radiance Fields
Quankai Gao, Qiangeng Xu, Hao Su, Ulrich Neumann, Zexiang Xu
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 multi-scale tensors (0.6, 0.3, 0.15) instead of single scale 0.3 (ablation) | Strivec · NeRF Synthetic | PSNR: 32.73 -> 33.55 | Improved |
| use single smallest scale (0.15) instead of 0.3 (ablation) | Strivec · NeRF Synthetic | PSNR: 32.73 -> 31.96 | Hurt |
| use CP tri-vector local tensors instead of vector-matrix local tensors (ablation) | Strivec · NeRF Synthetic | PSNR: 32.99 -> 33.02 with fewer parameters (21.64M -> 9.15M) | Improved |
| increase number of components from 12 to 48 (ablation) | Strivec · NeRF Synthetic | PSNR: 32.94 -> 33.55 | Improved |
| increase number of components from 48 to 96 (ablation) | Strivec · NeRF Synthetic | PSNR: 33.55 -> 33.59 | No clear effect |
| use local sparse tri-vector tensors instead of global TensoRF-VM under axis rotation | radiance field representation · NeRF Synthetic lego, rotation 45 deg | PSNR: 36.52 -> 34.97 vs TensoRF-VM 36.46 -> 32.31 | Improved |
| use Strivec Ours-24 instead of TensoRF-VM | radiance field representation · NeRF Synthetic | parameters (M): 17.95 -> 7.07 with PSNR 33.14 -> 33.24 | Improved |
| use Strivec instead of TensoRF-CP/VM | radiance field representation · NeRF Synthetic | training time: 17.4m -> 35.7m | Hurt |
| use Strivec Ours-48 instead of TensoRF-VM | radiance field representation · Tanks and Temples | PSNR: 28.56 -> 28.70 | Improved |
| use faster Strivec configuration | Strivec · NeRF Synthetic | PSNR: 33.55 -> 33.09 PSNR, training 959s vs TensoRF-VM 915s | Mixed |
Closest papers by meaning
- TensoRF: Tensorial Radiance Fields (ECCV 2022 · similarity 0.91)
- Mask-Based Modeling for Neural Radiance Fields (ICLR 2024 · similarity 0.88)
- Synergistic Integration of Coordinate Network and Tensorial Feature for Improving Neural Radiance Fields from Sparse Inputs (ICML 2024 · similarity 0.87)
- HyRF: Hybrid Radiance Fields for Memory-efficient and High-quality Novel View Synthesis (NeurIPS 2025 · similarity 0.87)
- ColNeRF: Collaboration for Generalizable Sparse Input Neural Radiance Field (AAAI 2024 · similarity 0.87)
- Learning Robust Generalizable Radiance Field with Visibility and Feature Augmented Point Representation (ICLR 2024 · similarity 0.87)
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.
