Research map / 3D vision / Gaussian splatting
Raymap-Guided Coupling for Drift-Robust Unposed Feed-Forward 3D Reconstruction
Xiangyu Sun, Liu.Liu Liu.Liu, Seungkwon Yang, Jingbing Han, Seungtae Nam, Zhizhong Su, Eunbyung Park
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 raymap loss (ablation) | Ours-Large · DL3DV 6v | PSNR: 21.869 -> 23.302 | Improved |
| add rgb loss (ablation) | Ours-Large · DL3DV 6v | PSNR: 12.608 -> 23.302 | Improved |
| add raymap loss (ablation) | Ours-Large · DL3DV 6v | AUC@5: 0.869 -> 0.874 | Improved |
| do not detach raymap in coupling module (ablation) | RGC · DL3DV 6v | PSNR: 20.178 -> 23.302 | Improved |
| use RGC instead of DAv3 Gaussian head (ablation) | Ours · DL3DV 6v | PSNR: 22.407 -> 23.302 | Improved |
| detach both raymap and depth (ablation) | RGC · DL3DV 6v | training | Hurt |
| use proposed sampler instead of original sampler (ablation) | Ours · DL3DV 6v range 0-50 | PSNR: 22.696 -> 23.302 | Improved |
| increase g_max without replay (ablation) | Ours · DL3DV 6v | PSNR: 23.302 -> 22.744 (0-50); 19.542 -> 19.712 (0-150) | Mixed |
| add replay at g_max=infinity (ablation) | Ours · DL3DV 6v range 0-50 | PSNR: 22.744 -> 23.006 | Improved |
| use Ours instead of YonoSplat | Ours · DL3DV 24v | PSNR: 22.174 -> 24.242 | 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.
- YonoSplat — YoNoSplat: You Only Need One Model for Feedforward 3D Gaussian Splatting (ICLR 2026) · better on 8 of 8 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
- COLMAP-Free 3D Gaussian Splatting (CVPR 2024 · similarity 0.90)
- SelfSplat: Pose-Free and 3D Prior-Free Generalizable 3D Gaussian Splatting (CVPR 2025 · similarity 0.89)
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.89)
- UniGS: Unified Geometry-Aware Gaussian Splatting for Multimodal Rendering (NeurIPS 2026 · similarity 0.89)
- Structure Consistent Gaussian Splatting with Matching Prior for Few-shot Novel View Synthesis (NeurIPS 2024 · similarity 0.88)
- Evolving High-Quality Rendering and Reconstruction in a Unified Framework with Contribution-Adaptive Regularization (CVPR 2025 · 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.
