Research map / 3D vision / 3D generation
latentSplat: Autoencoding Variational Gaussians for Fast Generalizable 3D Reconstruction
Christopher Wewer, Kevin Raj, Eddy Ilg, Bernt Schiele, Jan E. Lenssen
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 generative variational formulation and GAN loss (vs deterministic) (ablation) | latentSplat · CO3D hydrants extrapolation | FID: 69.33 -> 48.88 | Improved |
| add variational formulation and GAN loss (vs deterministic) (ablation) | latentSplat · CO3D hydrants extrapolation | PSNR: 16.50 -> 15.61 | Hurt |
| add multi-scale latent skip connections to decoder (ablation) | latentSplat · CO3D hydrants extrapolation | FID: 50.31 -> 48.88 | No clear effect |
| add RGB skip input to decoder (ablation) | latentSplat · CO3D hydrants extrapolation | FID: 48.48 -> 48.88 | No clear effect |
| replace pixelSplat's regression with latentSplat variational Gaussians and decoder | pixelSplat · RealEstate10k extrapolation | FID: 5.78 -> 2.79 | Improved |
| replace pixelSplat's regression with latentSplat on CO3D hydrants | pixelSplat · CO3D hydrants interpolation | PSNR: 18.05 -> 18.01 | No clear effect |
| use latentSplat output images for mesh reconstruction instead of pixelSplat | 2D Gaussian Splatting mesh · CO3D hydrants | Chamfer distance (x10^-8): 1.815 -> 1.535 | Improved |
| use generative decoder instead of GeNVS diffusion sampling | novel view synthesis · efficiency | Render time (s): 6.423 -> 0.003 | Improved |
| add generative decoder on top of Gaussian splatting | pixelSplat · render time | Render time (s): 0.002 -> 0.003 | Hurt |
Later papers that compared with this method
6 later accepted papers report LatentSplat as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.
- Sparse auto-regressive modeling for scene generation from multi-view images (ECCV 2026 · better on 8 of 8 shared results)
- ReconSplat: Generalizable 3D Scene Reconstruction Beyond Observed Views (ECCV 2026 · better on 5 of 6 shared results)
- One-Shot Refiner: Boosting Feed-forward Novel View Synthesis via One-Step Diffusion (AAAI 2026 · better on 1 of 1 shared result)
- Bolt3D: Generating 3D Scenes in Seconds (ICCV 2025 · better on 3 of 3 shared results)
- Uncertainty-Aware Diffusion-Guided Refinement of 3D Scenes (ICCV 2025 · better on 1 of 1 shared result)
- MVSplat360: Feed-Forward 360 Scene Synthesis from Sparse Views (NeurIPS 2024 · better on 10 of 10 shared results)
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.
- pixelSplat — pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction (CVPR 2024) · better on 9 of 11 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
- HiSplat: Hierarchical 3D Gaussian Splatting for Generalizable Sparse-View Reconstruction (ICLR 2025 · similarity 0.89)
- GaussianBlock: Building Part-Aware Compositional and Editable 3D Scene by Primitives and Gaussians (ICLR 2025 · similarity 0.89)
- pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction (CVPR 2024 · similarity 0.88)
- Generative Densification: Learning to Densify Gaussians for High-Fidelity Generalizable 3D Reconstruction (CVPR 2025 · similarity 0.88)
- HumanSplat: Generalizable Single-Image Human Gaussian Splatting with Structure Priors (NeurIPS 2024 · similarity 0.88)
- CodecSplat: Ultra-Compact Latent Coding for Feed-Forward 3D Gaussian Splatting (NeurIPS 2026 · 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.
