Research map / 3D vision / Face and shape
Rethinking Directional Parameterization in Neural Implicit Surface Reconstruction
Zijie Jiang, Tianhan Xu, Hiroharu Kato
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 |
|---|---|---|---|
| replace viewing direction with reflection direction (ablation) | NeuS · DTU (15 scenes) | Chamfer distance (total avg): 0.77 -> 1.11 | Hurt |
| use hybrid direction instead of viewing direction | NeuS · DTU (15 scenes) | Chamfer distance (total avg): 0.77 -> 0.72 | Improved |
| use hybrid direction instead of viewing direction | NeuDA · DTU (15 scenes) | Chamfer distance (total avg): 0.67 -> 0.63 | Improved |
| use hybrid direction instead of viewing direction on diffuse objects | NeuS · DTU (diffuse scenes) | Chamfer distance (avg): 0.67 -> 0.67 | No clear effect |
| use hybrid direction instead of viewing direction on specular objects | NeuS · DTU (specular scenes) | Chamfer distance (avg): 1.03 -> 0.86 | Improved |
| use hybrid direction instead of reflection direction | NeuS · Shiny Blender (coffee) | Acc.: 3.30 -> 0.64 | Improved |
| use hybrid direction instead of viewing direction | NeuS · Shiny Blender (4 objects) | Acc.: 2.76 -> 0.45 | Improved |
| replace viewing direction with reflection direction (ablation) | NeuS · Shiny Blender (4 objects) | Acc.: 2.76 -> 1.12 | Improved |
| use hybrid direction instead of reflection direction | NeuDA · Shiny Blender (4 objects) | Acc.: 1.41 -> 0.61 | Improved |
| set a low initial value of gamma_b | Ours (NeuDA) · DTU | reconstruction quality (qualitative) | Hurt |
| detach the SDF gradient in the blend weight | Ours (NeuDA) · DTU | reconstruction quality (qualitative) | Improved |
| fuse view and reflection directions before positional encoding rather than after | Ours (NeuDA) · DTU | reconstruction quality (qualitative) | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- NeuDA — NeuDA: Neural Deformable Anchor for High-Fidelity Implicit Surface Reconstruction (CVPR 2023) · better on 4 of 4 shared results
- NeuS — NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction (NeurIPS 2021) · better on 4 of 4 shared results
- Ref-NeuS — Ref-NeuS: Ambiguity-Reduced Neural Implicit Surface Learning for Multi-View Reconstruction with Reflection (ICCV 2023) · 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
- Differentiable Neural Surface Refinement for Modeling Transparent Objects (CVPR 2024 · similarity 0.89)
- NeRSP: Neural 3D Reconstruction for Reflective Objects with Sparse Polarized Images (CVPR 2024 · similarity 0.88)
- PMNI: Pose-free Multi-view Normal Integration for Reflective and Textureless Surface Reconstruction (CVPR 2025 · similarity 0.87)
- Flatten Anything: Unsupervised Neural Surface Parameterization (NeurIPS 2024 · similarity 0.87)
- High-Fidelity Polarimetric Implicit 3D Reconstruction with View-Dependent Physical Representation (AAAI 2025 · similarity 0.87)
- UniSDF: Unifying Neural Representations for High-Fidelity 3D Reconstruction of Complex Scenes with Reflections (NeurIPS 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.
