Research map / 3D vision / Depth and stereo
RayMap3R: Inference-Time RayMap for Dynamic 3D Reconstruction
Feiran Wang, Zezhou Shang, Gaowen Liu, Yan Yan
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 dual-branch RayMap dynamic identification (R) (ablation) | CUT3R · 3D reconstruction (sampled sequences) | Chamfer: 0.322 -> 0.245 | Improved |
| add dual-branch RayMap dynamic identification (R) (ablation) | CUT3R · Video depth (sampled sequences) | AbsRel: 0.208 -> 0.186 | Improved |
| add dual-branch RayMap dynamic identification (R) (ablation) | CUT3R · Camera pose (sampled sequences) | ATE: 0.114 -> 0.113 | No clear effect |
| add reset metric alignment (M) on top of R (ablation) | RayMap3R · 3D reconstruction (sampled sequences) | Chamfer: 0.245 -> 0.213 | Improved |
| add state-aware smoothing (S) on top of R (ablation) | RayMap3R · Camera pose (sampled sequences) | ATE: 0.113 -> 0.084 | Improved |
| add M and S to R (full model) (ablation) | RayMap3R · 3D reconstruction (sampled sequences) | Chamfer: 0.245 -> 0.170 | Improved |
| add dual-branch R (ablation) | CUT3R · Video depth KITTI | AbsRel: 0.120 -> 0.102 | Improved |
| add state-aware smoothing (S) on top of R (ablation) | RayMap3R · Camera pose TUM-dyn | ATE: 0.032 -> 0.020 | Improved |
| add reset metric alignment (M) on top of R (ablation) | RayMap3R · 3D reconstruction 7-Scenes | Acc: 0.032 -> 0.028 | Improved |
| use adaptive a_t x sc_t smoothing instead of fixed beta 0.5 (ablation) | State-aware smoothing · Mean over Sintel, TUM-dyn, ScanNet | Mean ATE: 0.092 -> 0.080 | Improved |
| use product a_t x sc_t instead of sc_t alone (ablation) | State-aware smoothing · Mean over Sintel, TUM-dyn, ScanNet | Mean ATE: 0.120 -> 0.080 | Improved |
| add smoothing of any kind vs none (fixed beta 0.3) (ablation) | CUT3R · Mean over Sintel, TUM-dyn, ScanNet | Mean RPE_t: 0.036 -> 0.022 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TTT3R — TTT3R: 3D Reconstruction as Test-Time Training (ICLR 2026) · better on 10 of 10 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
- 4D3R: Motion-Aware Neural Reconstruction and Rendering of Dynamic Scenes from Monocular Videos (NeurIPS 2025 · similarity 0.88)
- Enhancing 3D Reconstruction for Dynamic Scenes (NeurIPS 2025 · similarity 0.88)
- Point3R: Streaming 3D Reconstruction with Explicit Spatial Pointer Memory (NeurIPS 2025 · similarity 0.87)
- Continuous 3D Perception Model with Persistent State (CVPR 2025 · similarity 0.87)
- IGGT4D: Streaming 4D Instance-Grounded Geometry Transformer (NeurIPS 2026 · similarity 0.86)
- BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates (NeurIPS 2026 · similarity 0.86)
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.
