Research map / 3D vision / Face and shape
Topologically Consistent Multi-View Face Inference Using Volumetric Sampling
Tianye Li, Shichen Liu, Timo Bolkart, Jiayi Liu, Hao Li, Yajie Zhao
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 volumetric ToFu instead of 3DMM regression (with post-processing) | multi-view face mesh inference · Light Stage multi-view face test set | median scan-to-mesh distance (mm): 1.659 -> 0.585 | Improved |
| use volumetric ToFu instead of 3DMM regression (with post-processing) | multi-view face mesh inference · Light Stage multi-view face test set | median vertex-to-vertex distance (mm): 2.88 -> 1.97 | Improved |
| use ToFu instead of DFNRMVS | multi-view face mesh inference · Light Stage multi-view face test set | median vertex-to-vertex distance (mm): 4.565 -> 1.973 | Improved |
| use ToFu instead of 3DMM regression | multi-view face mesh inference · Light Stage multi-view face test set | vertices below 1 mm scan-to-mesh (%): 33.1 -> 78.3 | Improved |
| use ToFu instead of 3DMM regression | multi-view face mesh inference · Light Stage multi-view face test set | median landmark error (mm): 3.21 -> 2.02 | Improved |
| use ToFu instead of DFNRMVS for speed | multi-view face mesh inference · Base mesh, 15 views | runtime (s): 4.5 -> 0.385 | Improved |
| use ToFu instead of traditional MVS plus registration pipeline | multi-view face mesh inference · Base mesh, 15 views | runtime (s): 600+ -> 0.385 | Improved |
| use mesh-based correspondence instead of optical flow | 2D correspondence warping · Image 1366x1003 | runtime (s): 30 -> 1 | Improved |
| add normal displacement weights to the mesh upsampling function (ablation) | ToFu upsampling operator · Light Stage test set | shape detail quality | Improved |
| reduce the number of input camera views (ablation) | ToFu · Light Stage test set | scan-to-mesh error | Hurt |
| add successive local refinement stages after the global stage (ablation) | ToFu · Light Stage test set | scan-to-mesh distance | Improved |
Closest papers by meaning
- VGGTFace: Topologically Consistent Facial Geometry Reconstruction in the Wild (AAAI 2026 · similarity 0.91)
- Learning Complete 3D Morphable Face Models From Images and Videos (CVPR 2021 · similarity 0.90)
- Pixel3DMM: Versatile Screen-Space Priors for Single-Image 3D Face Reconstruction (ICLR 2026 · similarity 0.89)
- 3D Reconstruction of “In-the-Wild” Faces in Images and Videos (TPAMI 2018 · similarity 0.89)
- 3D Facial Expressions through Analysis-by-Neural-Synthesis (CVPR 2024 · similarity 0.89)
- AvatarMe++: Facial Shape and BRDF Inference With Photorealistic Rendering-Aware GANs (TPAMI 2021 · similarity 0.89)
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.
