Research map / 3D vision / Face and shape
Self-Supervised Geometry-Aware Encoder for Style-Based 3D GAN Inversion
Read the paperCheck 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 |
|---|---|---|---|
| Use the E3DGE encoder instead of optimization-based PTI inversion | StyleSDF inversion · CelebA-HQ | Time (s): 246 -> 0.45(T)/0.81(S) | Helped |
| Use E3DGE instead of the strongest baseline | StyleSDF inversion · Novel View Synthesis | MAE: .186 ±.016 -> .147 ± .011 | Helped |
| Add local features to synthetic training (ablation) | E3DGE · CelebA-HQ source-view reconstruction | MAE: .245 ± .024 -> .074 ± .007 | Helped |
| Add local features to synthetic training (ablation) | E3DGE · Novel View Synthesis | MAE: .241 ± .011 -> .282 ± .103 | Hurt |
| Use 2D alignment rather than raw local features (ablation) | E3DGE · Novel View Synthesis | MAE: .282 ± .103 -> .178 ± .007 | Helped |
| Use 3D alignment rather than raw local features (ablation) | E3DGE · Novel View Synthesis | MAE: .282 ± .103 -> .150 ± .011 | Helped |
| Add surface geometry supervision (ablation) | E3DGE global encoder · NoW Challenge validation set | Median: 2.43 -> 1.75 | Helped |
| Add around-surface geometry supervision (ablation) | E3DGE global encoder · NoW Challenge validation set | Median: 1.75 -> 1.71 | Helped |
| Add code reconstruction loss (ablation) | E3DGE global encoder · NoW Challenge validation set | Median: 1.71 -> 1.66 | Helped |
| Add 3D supervision to 2D inversion training (ablation) | E3DGE global encoder · NoW Challenge validation set | Mean: 2.43 -> 2.11 | Helped |
Closest papers by meaning
- In-N-Out: Faithful 3D GAN Inversion with Volumetric Decomposition for Face Editing (CVPR 2024 · similarity 0.91)
- Dual Encoder GAN Inversion for High-Fidelity 3D Head Reconstruction from Single Images (NeurIPS 2024 · similarity 0.90)
- WarpGAN: Warping-Guided 3D GAN Inversion with Style-Based Novel View Inpainting (NeurIPS 2025 · similarity 0.89)
- Dual Mapping of 2D StyleGAN for 3D-Aware Image Generation and Manipulation (Student Abstract) (AAAI 2024 · similarity 0.87)
- Controllable 3D Face Generation with Conditional Style Code Diffusion (AAAI 2024 · similarity 0.87)
- Diffusion-driven GAN Inversion for Multi-Modal Face Image Generation (CVPR 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.