Research map / 3D vision / 3D generation
DiGA3D: Coarse-to-Fine Diffusional Propagation of Geometry and Appearance for Versatile 3D Inpainting
Jingyi Pan, Dan Xu, Qiong Luo
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 TG-SDS loss (fine stage) (ablation) | DiGA3D · Object removal (SPIn-NeRF) | PSNR: 20.66 -> 20.71 | Improved |
| add TG-SDS loss (fine stage) (ablation) | DiGA3D · Object removal (SPIn-NeRF) | LPIPS: 0.29 -> 0.28 | Improved |
| add TG-SDS loss (fine stage) (ablation) | DiGA3D · Object removal (SPIn-NeRF) | SSIM: 0.57 -> 0.58 | Improved |
| add AFP and DDIM inversion in 2D inpainter (coarse stage) (ablation) | DiGA3D · Object removal (SPIn-NeRF) | PSNR: 20.45 -> 20.66 | Improved |
| add AFP and DDIM inversion in 2D inpainter (coarse stage) (ablation) | DiGA3D · Object removal (SPIn-NeRF) | LPIPS: 0.30 -> 0.29 | Improved |
| increase reference views K from 1 to 3 (ablation) | DiGA3D · SPIn-NeRF Scene 1 | PSNR: 19.87 -> 19.94 | Improved |
| increase reference views K from 1 to 3 (ablation) | DiGA3D · SPIn-NeRF Scene 1 | memory (GB): 41G -> 47G | Hurt |
| increase reference views K from 2 to 3 (ablation) | DiGA3D · SPIn-NeRF Scene 1 | LPIPS: 0.3412 -> 0.3330 | Improved |
| select reference views by K-means instead of randomly (ablation) | DiGA3D · Object removal (SPIn-NeRF) | consistency | Improved |
| select reference views by K-means instead of randomly (coarse stage) (ablation) | DiGA3D · Object removal (SPIn-NeRF) | consistency | No clear effect |
| DiGA3D vs GScream on masked LPIPS | GScream · Object removal (SPIn-NeRF) | Masked LPIPS: 0.54 -> 0.56 | Hurt |
| DiGA3D vs GaussCtrl for re-texturing | GaussCtrl · Object re-texturing | CLIP_dir: 0.0742 -> 0.1751 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- GaussCtrl — GaussCtrl: Multi-View Consistent Text-Driven 3D Gaussian Splatting Editing (ECCV 2024) · better on 3 of 3 shared results
- MVIP-NeRF — MVIP-NeRF: Multi-view 3D Inpainting on NeRF Scenes via Diffusion Prior (CVPR 2024) · better on 1 of 1 shared result
- NeRFiller — NeRFiller: Completing Scenes via Generative 3D Inpainting (CVPR 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- In-N-Out: Lifting 2D Diffusion Prior for 3D Object Removal via Tuning-Free Latents Alignment (NeurIPS 2024 · similarity 0.88)
- InstaInpaint: Instant 3D-Scene Inpainting with Masked Large Reconstruction Model (NeurIPS 2025 · similarity 0.88)
- MVIP-NeRF: Multi-view 3D Inpainting on NeRF Scenes via Diffusion Prior (CVPR 2024 · similarity 0.88)
- NeRFiller: Completing Scenes via Generative 3D Inpainting (CVPR 2024 · similarity 0.87)
- IMFine: 3D Inpainting via Geometry-guided Multi-view Refinement (CVPR 2025 · similarity 0.87)
- Sculpt3D: Multi-View Consistent Text-to-3D Generation with Sparse 3D Prior (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.
