Research map / 3D vision / Gaussian splatting
High-Resolution and Few-shot View Synthesis from Asymmetric Dual-lens Inputs
Ruikang Xu, Mingde Yao, Yue Li, Yueyi Zhang, Zhiwei Xiong
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 pre-upsampling of wide-angle images to the 3D-GS baseline (ablation) | 3D-GS + HAT · Simulated dataset 10-shot | PSNR: 17.89 -> 18.65 | Improved |
| add depth-wise loss Rd (ablation) | DL-GS pre-upsampling baseline · Simulated dataset 10-shot | PSNR: 18.65 -> 19.03 | Improved |
| add dual-lens-consistent loss Rc (ablation) | DL-GS · Simulated dataset 10-shot | PSNR: 19.03 -> 19.40 | Improved |
| add wide-angle reference branch in refinement module (ablation) | DL-GS · Simulated dataset 90-shot | PSNR: 25.08 -> 25.35 | Improved |
| add telephoto reference branch in refinement module (ablation) | DL-GS · Simulated dataset 90-shot | PSNR: 25.35 -> 25.61 | Improved |
| use dual-lens regularization instead of monocular-depth few-shot methods (FSGS + HAT) | 3D-GS · Real-captured forward-facing 5-shot | PSNR: 23.08 -> 24.05 | Improved |
| DL-GS versus NeRF-SR in dense training | DL-GS · Simulated dataset 20-shot SSIM vs DS-NeRF | PSNR: 21.54 -> 21.77 | Improved |
| replace NeRF-based methods with 3D-GS-based DL-GS | DL-GS · Simulated 10-shot training time | training time: 5h 8min -> 14min 35s | Improved |
| NeRF-SR without few-shot regularization | NeRF-SR · Real-captured forward-facing 5-shot | PSNR: 12.53 (5-shot) vs 30.53 (50-shot) | Hurt |
Closest papers by meaning
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.88)
- FewViewGS: Gaussian Splatting with Few View Matching and Multi-stage Training (NeurIPS 2024 · similarity 0.88)
- SplatFlow: Multi-View Rectified Flow Model for 3D Gaussian Splatting Synthesis (CVPR 2025 · similarity 0.88)
- DOF-GS: Adjustable Depth-of-Field 3D Gaussian Splatting for Post-Capture Refocusing, Defocus Rendering and Blur Removal (CVPR 2025 · similarity 0.88)
- GS-IR: 3D Gaussian Splatting for Inverse Rendering (CVPR 2024 · similarity 0.87)
- SplatFormer: Point Transformer for Robust 3D Gaussian Splatting (ICLR 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.
