Research map / 3D vision / Face and shape
Physically-Aware Generative Network for 3D Shape Modeling
Check 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 |
|---|---|---|---|
| Add connectivity and stability losses | IM-Net generation · PartNet Chair shape generation | Validity Ratio (VR): 55.5% -> 70.5% | Helped |
| Add connectivity and stability losses | PQ-Net generation · PartNet Chair shape generation | Validity Ratio (VR): 48.7% -> 60.5% | Helped |
| Add connectivity loss only (ablation) | IM-Net generation · PartNet Chair shape generation | Average Connected Components (CC): 2.01 -> 1.72 | Helped |
| Add stability loss only (ablation) | IM-Net generation · PartNet Chair shape generation | Average Potential Well (PW): 7.32 -> 9.90 | Helped |
| Add stability loss only (ablation) | PQ-Net generation · PartNet Chair shape generation | Connection Ratio (CR): 53.5% -> 56.4% | No clear effect |
| Use connectivity and stability losses together (ablation) | IM-Net generation · PartNet Chair shape generation | MMD: 7.25 -> 7.11 | Hurt |
| Use connectivity and stability losses together (ablation) | PQ-Net generation · PartNet Chair shape generation | MMD: 7.33 -> 7.36 | Hurt |
| Replace decoder with physically-aware decoder using both losses | IM-Net auto-encoder · PartNet Chair shape auto-encoding | Validity Ratio (VR): 64.8% -> 74.7% | Helped |
| Replace decoder with physically-aware decoder using both losses | PQ-Net auto-encoder · PartNet Chair shape auto-encoding | Chamfer Distance (CD): 2.91 -> 3.00 | Hurt |
| Use connectivity loss only (ablation) | IM-Net auto-encoder · PartNet Chair shape auto-encoding | IoU: 67.13 -> 70.42 | Helped |
| Use stability loss only (ablation) | PQ-Net auto-encoder · PartNet Chair shape auto-encoding | IoU: 66.08 -> 65.00 | Hurt |
| Use both physical losses for shape correction | IM-Net correction model · PartNet Chair shape correction | Validity Ratio (VR): 18.9% -> 34.9% | Helped |
Closest papers by meaning
- Physical Simulation Layer for Accurate 3D Modeling (CVPR 2022 · similarity 0.92)
- PhysGen: Physically Grounded 3D Shape Generation for Industrial Design (CVPR 2026 · similarity 0.89)
- Enhancing Implicit Shape Generators Using Topological Regularizations (ICML 2024 · similarity 0.87)
- GPLD3D: Latent Diffusion of 3D Shape Generative Models by Enforcing Geometric and Physical Priors (CVPR 2024 · similarity 0.87)
- Flow-Corrected Shape Optimization: Taming Manifold Drift in High-Dimensional 3D Models (NeurIPS 2026 · similarity 0.86)
- ShapeShifter: 3D Variations Using Multiscale and Sparse Point-Voxel Diffusion (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.