Research map / 3D vision / 3D generation
Edit in 2D, Verify in 3D: Reinforcement Learning for Multi-view Consistent Scene Editing
Jiyuan Wang, Chunyu Lin, Lei Sun, Zhi Cao, Yuyang Yin, Lang Nie, Zhenlong Yuan, Xiangxiang Chu, Yunchao Wei, Kang Liao, Guosheng Lin
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 VGGT depth and point-confidence rewards (ablation) | RL3DEdit · face scenes ablation | VIEScore: 2.11 -> 5.26 | Improved |
| add VGGT depth and point-confidence rewards (ablation) | RL3DEdit · face scenes ablation | photometric reprojection loss: 0.193 -> 0.077 | Improved |
| add relative pose reward (ablation) | RL3DEdit · face scenes ablation | photometric reprojection loss: 0.131 -> 0.077 | Improved |
| add relative pose reward (ablation) | RL3DEdit · face scenes ablation | VIEScore: 4.77 -> 5.26 | Improved |
| add anchor reward (ablation) | RL3DEdit · face scenes ablation | VIEScore: 4.34 -> 5.26 | Improved |
| add anchor reward (ablation) | RL3DEdit · face scenes ablation | photometric reprojection loss: 0.091 -> 0.077 | Hurt |
| replace VGGT reward with SfM-based reward (ablation) | RL3DEdit · face scenes ablation | VIEScore: 5.26 -> 0.97 | Hurt |
| replace VGGT reward with reprojection warping reward (ablation) | RL3DEdit · face scenes ablation | VIEScore: 5.26 -> 1.41 | Hurt |
| replace VGGT reward with reprojection warping reward (ablation) | RL3DEdit · face scenes ablation | photometric reprojection loss: 0.077 -> 0.065 | Improved |
| use Qwen-Image-Edit backbone instead of FLUX-Kontext (ablation) | RL3DEdit · face scenes ablation | VIEScore: 5.26 -> 5.43 | No clear effect |
| use RL3DEdit instead of EditSplat w/ FLUX-Kontext | EditSplat · 3D editing, 100 test cases | VIEScore: 3.23 -> 5.48 | Improved |
| use RL3DEdit instead of EditSplat w/ FLUX-Kontext | EditSplat · 3D editing, 100 test cases | editing time: 40min -> 1.5min | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 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 4 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
- ConsistDreamer: 3D-Consistent 2D Diffusion for High-Fidelity Scene Editing (CVPR 2024 · similarity 0.90)
- SyncNoise: Geometrically Consistent Noise Prediction for Instruction-based 3D Editing (AAAI 2025 · similarity 0.89)
- ProEdit: Simple Progression is All You Need for High-Quality 3D Scene Editing (NeurIPS 2024 · similarity 0.89)
- EditSplat: Multi-View Fusion and Attention-Guided Optimization for View-Consistent 3D Scene Editing with 3D Gaussian Splatting (CVPR 2025 · similarity 0.88)
- EditRoom: LLM-parameterized Graph Diffusion for Composable 3D Room Layout Editing (ICLR 2025 · similarity 0.87)
- CTRL-D: Controllable Dynamic 3D Scene Editing with Personalized 2D Diffusion (CVPR 2025 · 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.
