Research map / 3D vision / 3D generation
Edit3r: Instant 3D Scene Editing from Sparse Unposed Images
Jiageng Liu, Weijie Lyu, Xueting Li, Yejie Guo, Ming-Hsuan Yang
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 SAM2 recoloring supervision (ablation) | Edit3r · DL3DV-Edit-Bench | C-FID: 215.0 -> 171.3 | Improved |
| add SAM2 recoloring supervision (ablation) | Edit3r · DL3DV-Edit-Bench | CLIP t2i: 0.243 -> 0.266 | Improved |
| add 3D regularization losses (ablation) | Edit3r · DL3DV-Edit-Bench | C-FID: 278.4 -> 171.3 | Improved |
| add 3D regularization losses (ablation) | Edit3r · DL3DV-Edit-Bench | CLIP t2i: 0.237 -> 0.266 | Improved |
| add SAM2 mask-jitter augmentation (ablation) | Edit3r · DL3DV-Edit-Bench | C-FID: 179.6 -> 171.3 | Improved |
| add random drop of first view Gaussians (p=0.5) (ablation) | Edit3r · DL3DV-Edit-Bench | C-FID: 183.1 -> 171.3 | Improved |
| use Edit3r instead of NoPoSplat on edited inputs | feed-forward reconstruction · DL3DV-Edit-Bench | CLIP t2i: 0.253 -> 0.266 | Improved |
| use Edit3r instead of GaussCtrl | optimization-based editing · DL3DV-Edit-Bench | C-FID: 135.0 -> 171.3 | Hurt |
| use Edit3r instead of per-scene optimization | EditSplat · DL3DV-Edit-Bench | Time (s): 584.46 -> 0.51 | Improved |
| swap inference editor from IP2P to FLUX (ablation) | Edit3r · DL3DV-Edit-Bench | CLIP t2i: 0.266 -> 0.276 | Improved |
| swap inference editor from IP2P to Gemini (ablation) | Edit3r · DL3DV-Edit-Bench | CLIP t2i: 0.266 -> 0.246; C-FID 171.3 -> 150.3 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- EditSplat — EditSplat: Multi-View Fusion and Attention-Guided Optimization for View-Consistent 3D Scene Editing with 3D Gaussian Splatting (CVPR 2025) · better on 4 of 4 shared results
- GaussCtrl — GaussCtrl: Multi-View Consistent Text-Driven 3D Gaussian Splatting Editing (ECCV 2024) · better on 2 of 4 shared results
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
- EditVerse3D: High-Quality 3D Object Editing with Region-Aware Learning (ECCV 2026 · similarity 0.91)
- Omni-3DEdit: Generalized Versatile 3D Editing in One-Pass (CVPR 2026 · similarity 0.90)
- EditSplat: Multi-View Fusion and Attention-Guided Optimization for View-Consistent 3D Scene Editing with 3D Gaussian Splatting (CVPR 2025 · similarity 0.90)
- Customize your NeRF: Adaptive Source Driven 3D Scene Editing via Local-Global Iterative Training (CVPR 2024 · similarity 0.89)
- CTRL-D: Controllable Dynamic 3D Scene Editing with Personalized 2D Diffusion (CVPR 2025 · similarity 0.89)
- SHAP-EDITOR: Instruction-Guided Latent 3D Editing in Seconds (CVPR 2024 · similarity 0.88)
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.
