Research map / 3D vision / Depth and stereo
Skullptor: High Fidelity 3D Head Reconstruction in Seconds with Multi-View Normal Prediction
Noé Artru, Rukhshanda Hussain, Emeline Got, Alexandre Messier, David B. Lindell, Abdallah Dib
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 view-aware cross-attention and multi-view training to DAViD | normal estimator · NPHM | normal gradient error: 0.190 -> 0.166 | Improved |
| add view-aware cross-attention and multi-view training to DAViD | normal estimator · Multiface | normal gradient error: 0.250 -> 0.234 | Improved |
| add view-aware cross-attention and multi-view training to DAViD | normal estimator · Multiface | mean angular error: 9.16 -> 9.13 | No clear effect |
| fine-tune DAViD on the head-scan data with single-view inputs (ablation) | DAViD · NPHM | normal angular error: 7.80 -> 8.43 | Hurt |
| fine-tune DAViD on the head-scan data with single-view inputs (ablation) | DAViD · NPHM | mesh depth error (mm): 2.68 -> 3.53 | Hurt |
| train on multi-view inputs without cross-attention (ablation) | DAViD · NPHM | normal angular error: 8.43 -> 7.13 | Improved |
| add cross-attention on top of multi-view training (ablation) | DAViD_ft_multi · NPHM | mesh depth error (mm): 2.87 -> 2.44 | Improved |
| add cross-attention on top of multi-view training (ablation) | DAViD_ft_multi · Multiface | mesh depth error (mm): 3.70 -> 3.29 | Improved |
| add cross-attention on top of multi-view training (ablation) | DAViD_ft_multi · Multiface | normal angular error: 8.59 -> 8.54 | No clear effect |
| reduce input views from 23 to 10 | Skullptor · NPHM | depth error (mm): 2.33 -> 2.36 | No clear effect |
| reduce input views from 26 to 10 | Skullptor · Multiface | depth error (mm): 2.43 -> 2.99 | Hurt |
| initialize mesh from an aligned template head instead of a unit sphere | Skullptor · mesh optimization | reconstruction time and quality | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Sapiens 2B — Sapiens: Foundation for Human Vision Models (ECCV 2024) · better on 7 of 8 shared results
- DAViD — DAViD: Modeling Dynamic Affordance of 3D Objects Using Pre-trained Video Diffusion Models (ICCV 2025) · better on 4 of 4 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
- H3D-Net: Few-Shot High-Fidelity 3D Head Reconstruction (ICCV 2021 · similarity 0.90)
- Implicit Shape and Appearance Priors for Few-Shot Full Head Reconstruction (TPAMI 2025 · similarity 0.89)
- FaceLift: Learning Generalizable Single Image 3D Face Reconstruction from Synthetic Heads (ICCV 2025 · similarity 0.88)
- Large-Scale High-Quality 3D Gaussian Head Reconstruction from Multi-View Captures (ECCV 2026 · similarity 0.87)
- SuperNormal: Neural Surface Reconstruction via Multi-View Normal Integration (CVPR 2024 · similarity 0.87)
- PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing (CVPR 2025 · similarity 0.86)
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.
