Research map / 3D vision / Face and shape
WIR3D: Visually-Informed and Geometry-Aware 3D Shape Abstraction
Richard Liu, Daniel Fu, Noah Tan, Itai Lang, Rana Hanocka
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 intermediate CLIP layer supervision (ablation) | WIR3D · shape abstraction ablation | LPIPS: 0.294 -> 0.227 | Improved |
| add stage 1 geometry optimization (ablation) | WIR3D · shape abstraction ablation | LPIPS: 0.248 -> 0.227 | Improved |
| add stage 1 geometry optimization (ablation) | WIR3D · shape abstraction ablation | Coverage: 0.016 -> 0.008 | Improved |
| add SDF loss (ablation) | WIR3D · shape abstraction ablation | Coverage: 0.012 -> 0.008 | Improved |
| add SDF loss (ablation) | WIR3D · shape abstraction ablation | LPIPS: 0.229 -> 0.227 | No clear effect |
| add localized keypoint loss (ablation) | WIR3D · shape abstraction ablation | LPIPS: 0.233 -> 0.227 | Improved |
| add localized keypoint loss (ablation) | WIR3D · shape abstraction ablation | Coverage: 0.009 -> 0.008 | No clear effect |
| add intermediate CLIP layer supervision (ablation) | WIR3D · shape abstraction ablation | CLIP img similarity: 0.891 -> 0.909 | Improved |
| replace 3Doodle with WIR3D | 3Doodle · shape abstraction (novel views) | Coverage: 0.020 -> 0.008 | Improved |
| replace 3Doodle with WIR3D | 3Doodle · shape abstraction (novel views) | CLIP img similarity: 0.900 -> 0.909 | Improved |
| replace 3Doodle with WIR3D | 3Doodle · user study (N=96) | preference: 0.12 -> 0.88 | Improved |
| use WIR3D curve handles instead of ARAP for deformation | ARAP · deformation user study (N=42) | preference: 80% | Improved |
Closest papers by meaning
- Geometry-Aware Network for Non-Rigid Shape Prediction From a Single View (CVPR 2018 · similarity 0.88)
- ShapeClipper: Scalable 3D Shape Learning From Single-View Images via Geometric and CLIP-Based Consistency (CVPR 2023 · similarity 0.87)
- "NeuSDFusion: A Spatial-Aware Generative Model for 3D Shape Completion, Reconstruction, and Generation" (ECCV 2024 · similarity 0.87)
- Geometry-Guided Modeling of Foundation Features Enables Generalizable Object Shape Deformation Learning (ICML 2026 · similarity 0.87)
- DeFormer: Integrating Transformers with Deformable Models for 3D Shape Abstraction from a Single Image (ICCV 2023 · similarity 0.86)
- $\texttt{NAISR}$: A 3D Neural Additive Model for Interpretable Shape Representation (ICLR 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.
