Research map / 3D vision / Gaussian splatting
TWINGS: Thin Plate Splines Warp-aligned Initialization for Sparse-View Gaussian Splatting
Hyeseong Kim, Geonhui Son, Deukhee Lee, Dosik Hwang
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 |
|---|---|---|---|
| use multi-view instead of pairwise correspondences (ablation) | TWINGS · DTU (3-view) | PSNR: 21.32 -> 21.52 | Improved |
| use multi-view instead of pairwise correspondences (ablation) | TWINGS · LLFF (3-view) | PSNR: 21.29 -> 21.49 | Improved |
| use multi-view instead of pairwise correspondences (ablation) | TWINGS · Mip-NeRF360 (12-view) | PSNR: 19.99 -> 20.35 | Improved |
| add TWINGS-Init (ablation) | 3DGS · DTU (3-view), 3DGS | PSNR: 17.65 -> 20.21 | Improved |
| add TWINGS-Init (ablation) | FSGS · DTU (3-view), FSGS | PSNR: 17.24 -> 20.42 | Improved |
| add TWINGS-Init (ablation) | CoR-GS · DTU (3-view), CoR-GS | PSNR: 19.21 -> 21.27 | Improved |
| use TPS instead of Linear Scaling (ablation) | TWINGS · LLFF (3-view) | PSNR: 20.90 -> 21.49 | Improved |
| use TPS instead of Free Form Deformation (ablation) | TWINGS · LLFF (3-view) | PSNR: 20.96 -> 21.49 | Improved |
| use MASt3R matches instead of SIFT (ablation) | TWINGS · LLFF (3-view) | PSNR: 21.01 -> 21.49 | Improved |
| use TWINGS-Init instead of MASt3R-triangulated point cloud (ablation) | TWINGS · DTU (3-view) | PSNR: 21.05 -> 21.52 | Improved |
| use nearest instead of random view selection (ablation) | TWINGS · Mip-NeRF360 (12-view) | PSNR: 20.00 -> 20.35 | Improved |
| shrink sampling radius from S/8 to S/64 (ablation) | TWINGS · LLFF (3-view) | PSNR: 21.49 -> 21.07 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DropGaussian — DropGaussian: Structural Regularization for Sparse-view Gaussian Splatting (CVPR 2025) · better on 6 of 6 shared results
- CoR-GS — CoR-GS: Sparse-View 3D Gaussian Splatting via Co-Regularization (ECCV 2024) · better on 5 of 5 shared results
- FreeNeRF — FreeNeRF: Improving Few-Shot Neural Rendering With Free Frequency Regularization (CVPR 2023) · better on 1 of 1 shared result
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
- Confidence-Guided Multi-Scale Aggregation for Sparse-View High-Resolution 3D Gaussian Splatting (CVPR 2026 · similarity 0.91)
- Path Matters: Unveiling Geometric Implicit Bias via Curvature-Aware Sparse View Optimization (ICLR 2026 · similarity 0.90)
- FatesGS: Fast and Accurate Sparse-View Surface Reconstruction Using Gaussian Splatting with Depth-Feature Consistency (AAAI 2025 · similarity 0.90)
- EDGS: Eliminating Densification for Efficient Convergence of 3DGS (CVPR 2026 · similarity 0.90)
- Sparse2DGS: Geometry-Prioritized Gaussian Splatting for Surface Reconstruction from Sparse Views (CVPR 2025 · similarity 0.90)
- KeyGS: A Keyframe-Centric Gaussian Splatting Method for Monocular Image Sequences (AAAI 2025 · similarity 0.89)
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.
