Research map / 3D vision / Face and shape
Multiview Neural Surface Reconstruction by Disentangling Geometry and Appearance
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 IDR instead of Colmap trim=0 | DTU fixed-camera watertight reconstruction · DTU | Chamfer-L1 (mm): 1.36 -> 0.9 | Helped |
| Use IDR instead of Colmap trim=0 | DTU fixed-camera reconstruction · DTU | PSNR: 20.58 -> 23.20 | Helped |
| Use IDR instead of Colmap trim=7 | DTU fixed-camera trimmed reconstruction · DTU | Chamfer-L1 (mm): 0.65 -> 0.9 | Hurt |
| Use IDR instead of Colmap trim=7 | DTU fixed-camera trimmed reconstruction · DTU | PSNR: 18.54 -> 23.20 | Helped |
| Optimize camera parameters jointly with geometry instead of fixing inaccurate initial cameras (ablation) | IDR · DTU scan 114 | reconstruction quality | Helped |
| Add viewing direction to the renderer (ablation) | IDR · DTU scan 114 | reconstruction quality | Helped |
| Add surface normal to the renderer (ablation) | IDR · DTU scan 114 and geometry-appearance transfer | geometry-appearance separation | Helped |
| Add global geometry feature vector to the renderer (ablation) | IDR · DTU scan 114 | reconstruction quality | Helped |
| Jointly estimate cameras and reconstruct geometry instead of using Colmap | IDR · Fountain-P11 | R_error (deg): 0.03 -> 0.02 | Helped |
| Jointly estimate cameras and reconstruct geometry instead of using Colmap | IDR · Fountain-P11 | t_error (mm): 2.86 -> 2.02 | Helped |
| Use IDR instead of Colmap | Fountain-P11 · Fountain-P11 | PSNR: 21.99 -> 26.48 | Helped |
| Evaluate on held-out images rather than training images | IDR · DTU fixed ground-truth cameras | PSNR: 23.34 -> 22.55 | Hurt |
Closest papers by meaning
- Multi-View Mesh Reconstruction With Neural Deferred Shading (CVPR 2022 · similarity 0.91)
- Just Look at the Image: Viewpoint-Specific Surface Normal Prediction for Improved Multi-View Reconstruction (CVPR 2016 · similarity 0.91)
- Differentiable Neural Surface Refinement for Modeling Transparent Objects (CVPR 2024 · similarity 0.90)
- PMNI: Pose-free Multi-view Normal Integration for Reflective and Textureless Surface Reconstruction (CVPR 2025 · similarity 0.88)
- UniSDF: Unifying Neural Representations for High-Fidelity 3D Reconstruction of Complex Scenes with Reflections (NeurIPS 2024 · similarity 0.88)
- GURecon: Learning Detailed 3D Geometric Uncertainties for Neural Surface Reconstruction (AAAI 2025 · 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.