Research map / 3D vision / Gaussian splatting
GIFSplat: Generative Prior-Guided Iterative Feed-Forward 3D Gaussian Splatting from Sparse Views
Tianyu Chen, Wei Xiang, Kang Han, Yu Lu, Di Wu, Gaowen Liu, Ramana Rao Kompella
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 iterative forward-only residual refinement (ablation) | GIFSplat · RealEstate10K 2-view (Average) | PSNR: 24.901 -> 26.559 | Improved |
| add generative prior cues from frozen Difix (ablation) | IFSplat · RealEstate10K 2-view (Average) | LPIPS: 0.145 -> 0.138 | Improved |
| add generative prior cues from frozen Difix (ablation) | IFSplat · RealEstate10K 2-view (Average) | PSNR: 26.291 -> 26.559 | Improved |
| add window attention in 3D Gaussian space (ablation) | GIFSplat · RealEstate10K 2-view (Average) | PSNR: 25.327 -> 26.559 | Improved |
| increase refinement steps from 3 to 4 (ablation) | GIFSplat · RealEstate10K 2-view (Average) | PSNR: 26.559 -> 26.561 | No clear effect |
| increase refinement steps from 0 to 3 (ablation) | IFSplat · RealEstate10K 2-view (Average) | PSNR: 24.901 -> 26.291 | Improved |
| use IFSplat iterative refinement instead of AnySplat one-shot prediction | feed-forward 3DGS · DTU 2-view (cross-domain) | PSNR: 18.122 -> 19.921 | Improved |
| add generative prior over IFSplat | IFSplat · DL3DV 8-view | PSNR: 24.69 -> 24.91 | Improved |
| iterative refinement vs AnySplat, small-overlap inputs | IFSplat · RealEstate10K 2-view (Small overlap) | PSNR: 22.703 -> 24.207 | Improved |
| iterative refinement vs AnySplat, large-overlap inputs | IFSplat · RealEstate10K 2-view (Large overlap) | PSNR: 27.414 -> 27.895 | Improved |
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.
- FLARE — FLARE: Feed-forward Geometry, Appearance and Camera Estimation from Uncalibrated Sparse Views (CVPR 2025) · better on 2 of 2 shared results
- PixelSplat — pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction (CVPR 2024) · better on 1 of 1 shared result
- MVSplat — MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images (ECCV 2024) · 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
- Generative Sparse-View Gaussian Splatting (CVPR 2025 · similarity 0.92)
- CompactSplat: Spatially Adaptive Gaussian Distribution for Feedforward Scene Reconstruction (NeurIPS 2026 · similarity 0.92)
- pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction (CVPR 2024 · similarity 0.91)
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.91)
- DiffSplat: Repurposing Image Diffusion Models for Scalable Gaussian Splat Generation (ICLR 2025 · similarity 0.91)
- Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives (CVPR 2025 · similarity 0.91)
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.
