Research map / 3D vision / NeRF and radiance fields
3D-aware Blending with Generative NeRFs
Hyunsu Kim, Gayoung Lee, Yunjey Choi, Jin-Hwa Kim, Jun-Yan Zhu
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 3D-aware alignment of the reference image to 2D baselines | 2D blending baselines · CelebA-HQ and AFHQv2-Cat | KID, LPIPS_m, masked L2 | Improved |
| add Poisson blending on top of the method (ablation) | 3D-aware blending · CelebA-HQ and AFHQv2-Cat | KID, masked L2, LPIPS_m | Improved |
| add Poisson blending to baselines (ablation) | baselines · CelebA-HQ and AFHQv2-Cat | qualitative | No clear effect |
| add density-blending loss (ablation) | 3D-aware blending · hair blending (qualitative) | qualitative | Improved |
| remove 3D-aware alignment of the reference (ablation) | 3D-aware blending · qualitative | qualitative | Hurt |
| add Poisson blending to cut optimization from 200 to 100 iterations | 3D-aware blending · runtime 512x512 | time (s): 26.9s -> 12.1s | Improved |
| optimize W+ latent space instead of W or generator G (ablation) | 3D-aware blending · qualitative | realism and faithfulness | Improved |
| add 3D local alignment (ICP) for loosely aligned data (ablation) | 3D-aware blending · AFHQv2-Cat | qualitative and user study | Improved |
Closest papers by meaning
- NeRF Analogies: Example-Based Visual Attribute Transfer for NeRFs (CVPR 2024 · similarity 0.89)
- SIGNeRF: Scene Integrated Generation for Neural Radiance Fields (CVPR 2024 · similarity 0.88)
- Language-driven Object Fusion into Neural Radiance Fields with Pose-Conditioned Dataset Updates (CVPR 2024 · similarity 0.88)
- Joint Optimization of Neural Radiance Fields and Continuous Camera Motion from a Monocular Video (CVPR 2025 · similarity 0.87)
- MVIP-NeRF: Multi-view 3D Inpainting on NeRF Scenes via Diffusion Prior (CVPR 2024 · similarity 0.87)
- ICON: Incremental CONfidence for Joint Pose and Radiance Field Optimization (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.
