Research map / 3D vision / Face and shape
Surface Geometry Processing: An Efficient Normal-Based Detail Representation
Wuyuan Xie, Miaohui Wang, Di Lin, Boxin Shi, Jianmin Jiang
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 |
|---|---|---|---|
| use normal-domain detail transfer instead of Laplace mesh editing (LAPL) | geometry detail transfer · DiLiGenT | shape MAE (deg): 3.56 -> 2.24 | Improved |
| use normal-domain detail transfer instead of LAPL | geometry detail transfer · DiLiGenT | detail MAE (deg): 18.31 -> 6.07 | Improved |
| use normal-domain detail transfer instead of LAPL | geometry detail transfer · DiLiGenT | memory (Mb): 11833.11 -> 19.59 | Improved |
| use normal-domain detail transfer instead of LAPL | geometry detail transfer · DiLiGenT | time (sec): 689.01 -> 0.48 | Improved |
| run RDN-Net on the detail component instead of PU-Net | 3D surface super-resolution · Panno 4x | MAE (deg): 24.39 -> 4.92 | Improved |
| run RDN-Net on the detail component instead of PU-Net | 3D surface super-resolution · Fish 4x | MAE (deg): 25.96 -> 3.28 | Improved |
| synthesize in detail-component domain instead of DGTS | geometric texture synthesis · Lizard | detail SSIM: 0.7740 -> 0.9901 | Improved |
| synthesize in detail-component domain instead of DGTS | geometric texture synthesis · Textile-to-Cat | detail SSIM: 0.7180 -> 0.9903 | Improved |
| increase input resolution from 5010 to 49532 points | DSPL baseline · Circular-to-Bunny transfer | detail MAE (deg): 16.02 -> 33.79 | Hurt |
| increase input resolution from 5010 to 306491 points | Proposed · Circular-to-Bunny transfer | detail MAE (deg): 10.71 -> 10.86 | No clear effect |
| increase input resolution from 5010 to 306491 points | Proposed · Circular-to-Bunny transfer | shape MAE (deg): 4.97 -> 2.28 | Improved |
| re-extract detail component repeatedly (k=2 to 4) | detail component · Circular | SSIM vs original detail: 0.9996 -> 0.9984 | No clear effect |
Closest papers by meaning
- SuperNormal: Neural Surface Reconstruction via Multi-View Normal Integration (CVPR 2024 · similarity 0.86)
- Flatten Anything: Unsupervised Neural Surface Parameterization (NeurIPS 2024 · similarity 0.85)
- World Tracing: Pixel-Aligned Geometry Beyond the Visible (NeurIPS 2026 · similarity 0.85)
- PMNI: Pose-free Multi-view Normal Integration for Reflective and Textureless Surface Reconstruction (CVPR 2025 · similarity 0.85)
- VCR-GauS: View Consistent Depth-Normal Regularizer for Gaussian Surface Reconstruction (NeurIPS 2024 · similarity 0.85)
- Normal-GS: 3D Gaussian Splatting with Normal-Involved Rendering (NeurIPS 2024 · similarity 0.85)
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.
