Research map / 3D vision / Gaussian splatting
Free-Range Gaussians: Non-Grid-Aligned Generative 3D Gaussian Reconstruction
Akhmedkhan Shabanov, Peter Hedman, Ethan Weber, Zhengqin Li, Denys Rozumnyi, Gael Le Lan, Naina Dhingra, Lei Luo, Andrea Vedaldi, Christian Richardt, Andrea Tagliasacchi, Bo Zhu and 1 more
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 flow matching generative formulation (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 26.30 -> 28.08 partial; 28.40 -> 31.49 full | Improved |
| add timestep-weighted multi-view rendering loss (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 26.87 -> 28.08 partial; 29.66 -> 31.49 full | Improved |
| add LoD coarse-to-fine curriculum training (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 26.42 -> 28.08 partial; 30.00 -> 31.49 full | Improved |
| add hierarchical patching over 8K Gaussians instead of no patching (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 26.75 -> 28.08 partial; 30.54 -> 31.49 full | Improved |
| add hierarchical patching over using 4K Gaussians unpatched (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 27.22 -> 28.08 partial; 30.47 -> 31.49 full | Improved |
| use LoD tree patching instead of random patching (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 27.16 -> 28.08 partial; 30.84 -> 31.49 full | Improved |
| add rendering-gradient concatenation to noisy inputs (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 27.51 -> 28.08 partial; 30.89 -> 31.49 full | Improved |
| add classifier-free guidance at inference (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 27.50 -> 28.08 partial; 31.30 -> 31.49 full | Improved |
| add photometric gradient guidance at inference (ablation) | Free-Range Gaussians · GSO (4 views, partial observation) | PSNR: 27.90 -> 28.08 partial; 31.10 -> 31.49 full | Improved |
| increase input views from 2 to 4 (ablation) | Free-Range Gaussians · GSO full observation | PSNR: 28.22 -> 31.49 | Improved |
| increase input views from 4 to 6 (ablation) | Free-Range Gaussians · GSO full observation | PSNR: 31.49 -> 32.04 | Improved |
| increase input views from 4 to 9 (ablation) | Free-Range Gaussians · GSO full observation | PSNR: 31.49 -> 32.39 | 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.
- ReconViaGen — ReconViaGen: Towards Accurate Multi-view 3D Object Reconstruction via Generation (ICLR 2026) · better on 5 of 6 shared results
- LaRa — LaRa: Efficient Large-Baseline Radiance Fields (ECCV 2024) · better on 5 of 6 shared results
- GS-LRM — GS-LRM: Large Reconstruction Model for 3D Gaussian Splatting (ECCV 2024) · better on 6 of 6 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
- FreeSplat: Generalizable 3D Gaussian Splatting Towards Free View Synthesis of Indoor Scenes (NeurIPS 2024 · similarity 0.89)
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.89)
- MaskGaussian: Adaptive 3D Gaussian Representation from Probabilistic Masks (CVPR 2025 · similarity 0.89)
- No Pose, No Problem: Surprisingly Simple 3D Gaussian Splats from Sparse Unposed Images (ICLR 2025 · similarity 0.89)
- High-fidelity 3D Object Generation from Single Image with RGBN-Volume Gaussian Reconstruction Model (CVPR 2025 · similarity 0.89)
- GaussianPro: 3D Gaussian Splatting with Progressive Propagation (ICML 2024 · 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.
