Research map / 3D vision / NeRF and radiance fields
Digging into Radiance Grid for Real-Time View Synthesis with Detail Preservation
Jian Zhang, Jinchi Huang, Bowen Cai, Huan Fu, Mingming Gong, Chaohui Wang, Jiaming Wang, Hongchen Luo, Rongfei Jia, Binqiang Zhao, Xing Tang
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 perceptual loss Lp (ablation) | HrSRG · Real 360 Objects "Hat" | LPIPS: 0.061 -> 0.056 | Improved |
| add perceptual loss Lp (ablation) | HrSRG · Real 360 Objects "Hat" | PSNR: 26.43 -> 26.64 | Improved |
| add specular MLP (vs diffuse HrSRG-minus) (ablation) | HrSRG · Real 360 Objects "Hat" | LPIPS: 0.069 -> 0.056 | Improved |
| add specular MLP (vs diffuse HrSRG-minus) (ablation) | HrSRG · Real 360 Objects "Hat" | PSNR: 25.65 -> 26.64 | Improved |
| use 512^3 voxel grid instead of 256^3 (ablation) | HrSRG · Real 360 Objects "Hat" | LPIPS: 0.107 -> 0.056 | Improved |
| directly optimize voxel colors instead of converged Def-NeRF (ablation) | HrSRG vs Def-NeRF-Plus · Real 360 Objects "Hat" | LPIPS: 0.128 -> 0.056 | Improved |
| apply median cut pruning (ablation) | HrSRG · Synthetic-NeRF 800x800 | model size (MB): 210.4 -> 35.8 | Improved |
| apply median cut pruning (ablation) | HrSRG · Synthetic-NeRF 800x800 | PSNR: 31.06 -> 31.04 | No clear effect |
| raise threshold tau1 from 0.01 to 0.9 (ablation) | HrSRG · Shoe | model size: 169.4M -> 35.5M | No clear effect |
| use HrSRG instead of PlenOctree | HrSRG vs PlenOctree · Tanks and Temples | LPIPS: 0.131 -> 0.099 | Improved |
| use HrSRG instead of PlenOctree | HrSRG vs PlenOctree · Tanks and Temples | PSNR: 27.99 -> 27.33 | Hurt |
| use HrSRG instead of PlenOctree-512 | HrSRG vs PlenOctree · Synthetic-NeRF 800x800 | PSNR: 31.67 -> 31.04 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- Mip-NeRF — Mip-NeRF: A Multiscale Representation for Anti-Aliasing Neural Radiance Fields (ICCV 2021) · better on 2 of 4 shared results
- Plenoxels — Plenoxels: Radiance Fields Without Neural Networks (CVPR 2022) · better on 2 of 2 shared results
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
- Direct Voxel Grid Optimization: Super-Fast Convergence for Radiance Fields Reconstruction (CVPR 2022 · similarity 0.94)
- NeRFLight: Fast and Light Neural Radiance Fields Using a Shared Feature Grid (CVPR 2023 · similarity 0.92)
- Cascaded and Generalizable Neural Radiance Fields for Fast View Synthesis (TPAMI 2023 · similarity 0.92)
- NeRF-VPT: Learning Novel View Representations with Neural Radiance Fields via View Prompt Tuning (AAAI 2024 · similarity 0.90)
- Sparse Voxels Rasterization: Real-time High-fidelity Radiance Field Rendering (CVPR 2025 · similarity 0.90)
- Learning with Unreliability: Fast Few-shot Voxel Radiance Fields with Relative Geometric Consistency (CVPR 2024 · 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.
