# GlobalSplat: Efficient Feed-Forward 3D Gaussian Splatting via Global Scene Tokens

arXiv: https://arxiv.org/abs/2604.15284 · HTML: https://mlsanity.com/paper/2604.15284/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Builds on 39 of its 44 references that are in the map.
- First posted on arXiv 16 Apr 2026.

## Builds on (most cited first, 30 of 39)

- [3D Gaussian Splatting for Real-Time Radiance Field Rendering](https://mlsanity.com/paper/2308.04079/) (2023)
- [pixelNeRF: Neural Radiance Fields from One or Few Images](https://mlsanity.com/paper/2012.02190/) (2020)
- [DUSt3R: Geometric 3D Vision Made Easy](https://mlsanity.com/paper/2312.14132/) (2023)
- [IBRNet: Learning Multi-View Image-Based Rendering](https://mlsanity.com/paper/2102.13090/) (2021)
- [Depth Anything 3: Recovering the Visual Space from Any Views](https://mlsanity.com/paper/2511.10647/) (2025)
- [pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction](https://mlsanity.com/paper/2312.12337/) (2023)
- [MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images](https://mlsanity.com/paper/2403.14627/) (2024)
- [Continuous 3D Perception Model with Persistent State](https://mlsanity.com/paper/2501.12387/) (2025)
- [Fast3R: Towards 3D Reconstruction of 1000+ Images in One Forward Pass](https://mlsanity.com/paper/2501.13928/) (2025)
- [SynSin: End-to-end View Synthesis from a Single Image](https://mlsanity.com/paper/1912.08804/) (2019)
- [GS-LRM: Large Reconstruction Model for 3D Gaussian Splatting](https://mlsanity.com/paper/2404.19702/) (2024)
- [MapAnything: Universal Feed-Forward Metric 3D Reconstruction](https://mlsanity.com/paper/2509.13414/) (2025)
- [DepthSplat: Connecting Gaussian Splatting and Depth](https://mlsanity.com/paper/2410.13862/) (2024)
- [3D Reconstruction with Spatial Memory](https://mlsanity.com/paper/2408.16061/) (2024)
- [DeepView: View Synthesis with Learned Gradient Descent](https://mlsanity.com/paper/1906.07316/) (2019)
- [No Pose, No Problem: Surprisingly Simple 3D Gaussian Splats from Sparse Unposed Images](https://mlsanity.com/paper/2410.24207/) (2024)
- [Infinite Nature: Perpetual View Generation of Natural Scenes from a Single Image](https://mlsanity.com/paper/2012.09855/) (2020)
- [AnySplat: Feed-forward 3D Gaussian Splatting from Unconstrained Views](https://mlsanity.com/paper/2505.23716/) (2025)
- [Compressed 3D Gaussian Splatting for Accelerated Novel View Synthesis](https://mlsanity.com/paper/2401.02436/) (2023)
- [LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias](https://mlsanity.com/paper/2410.17242/) (2024)
- [Streaming 4D Visual Geometry Transformer](https://mlsanity.com/paper/2507.11539/) (2025)
- [Learning-Based View Synthesis for Light Field Cameras](https://mlsanity.com/paper/1609.02974/) (2016)
- [MUSt3R: Multi-view Network for Stereo 3D Reconstruction](https://mlsanity.com/paper/2503.01661/) (2025)
- [TTT3R: 3D Reconstruction as Test-Time Training](https://mlsanity.com/paper/2509.26645/) (2025)
- [ContextGS: Compact 3D Gaussian Splatting with Anchor Level Context Model](https://mlsanity.com/paper/2405.20721/) (2024)
- [FreeSplat: Generalizable 3D Gaussian Splatting Towards Free-View Synthesis of Indoor Scenes](https://mlsanity.com/paper/2405.17958/) (2024)
- [MuRF: Multi-Baseline Radiance Fields](https://mlsanity.com/paper/2312.04565/) (2023)
- [HAC++: Towards 100X Compression of 3D Gaussian Splatting](https://mlsanity.com/paper/2501.12255/) (2025)
- [VolSplat: Rethinking Feed-Forward 3D Gaussian Splatting with Voxel-Aligned Prediction](https://mlsanity.com/paper/2509.19297/) (2025)
- [YoNoSplat: You Only Need One Model for Feedforward 3D Gaussian Splatting](https://mlsanity.com/paper/2511.07321/) (2025)

## Built on by (most cited first, 12 of 12)

- [Instant NuRec: Feed-Forward 3D Gaussian Reconstruction for Driving Scene Simulation](https://mlsanity.com/paper/2607.14203/) (2026)
- [AdaptSplat: Adapting Vision Foundation Models for Feed-Forward 3D Gaussian Splatting](https://mlsanity.com/paper/2605.10239/) (2026)
- [PRISM: Feed-Forward Single-Image 3D Reconstruction via Geometric Warp-Residual Modeling](https://mlsanity.com/paper/2606.25430/) (2026)
- [ATSplat: Compact Feed-forward 3D Gaussian Splatting with Adaptive Token Expansion](https://mlsanity.com/paper/2607.20417/) (2026)
- [Imagine3D-LLM: Teaching MLLMs to Imagine 3D Scenes Before Answering](https://mlsanity.com/paper/2609.38177/) (2026)
- [Learning Stable Canonical Worlds for Novel View Synthesis and Beyond](https://mlsanity.com/paper/2606.23027/) (2026)
- [RoadVGGT: Road-Structure-Aware Feed-Forward Road Surface Reconstruction](https://mlsanity.com/paper/2607.23758/) (2026)
- [LocusGS: Spatially Grounded Tokens for Feed-Forward 3D Gaussian Splatting](https://mlsanity.com/paper/2608.12825/) (2026)
- [PhGS: Post-Hoc Pruning and Refinement of Single-View Feed-Forward 3D Gaussian Reconstructions](https://mlsanity.com/paper/2609.20623/) (2026)
- [FeCoSplat: Feedback-Guided Compression for Feed-Forward 3D Gaussian Splatting](https://mlsanity.com/paper/2609.33330/) (2026)
- [S2Tok: Streaming 3D Gaussian Reconstruction with Persistent Spatial Tokens](https://mlsanity.com/paper/2610.08978/) (2026)
- [CoTL: Compact Token Learning for 3D Scene Representation](https://mlsanity.com/paper/2610.18446/) (2026)

## Abstract (arXiv)

The efficient spatial allocation of primitives serves as the foundation of 3D Gaussian Splatting, as it directly dictates the synergy between representation compactness, reconstruction speed, and rendering fidelity. Previous solutions, whether based on iterative optimization or feed-forward inference, suffer from significant trade-offs between these goals, mainly due to the reliance on local, heuristic-driven allocation strategies that lack global scene awareness. Specifically, current feed-forward methods are largely pixel-aligned or voxel-aligned. By unprojecting pixels into dense, view-aligned primitives, they bake redundancy into the 3D asset. As more input views are added, the representation size increases and global consistency becomes fragile. To this end, we introduce GlobalSplat, a framework built on the principle of align first, decode later. Our approach learns a compact, global, latent scene representation that encodes multi-view input and resolves cross-view correspondences before decoding any explicit 3D geometry. Crucially, this formulation enables compact, globally consistent reconstructions without relying on pretrained pixel-prediction backbones or reusing latent features from dense baselines. Utilizing a coarse-to-fine training curriculum that gradually increases decoded capacity, GlobalSplat natively prevents representation bloat. On RealEstate10K and ACID, our model achieves competitive novel-view synthesis performance while utilizing as few as 16K Gaussians, significantly less than required by dense pipelines, obtaining a light 4MB footprint. Further, GlobalSplat enables significantly faster inference than the baselines, operating under 78 milliseconds in a single forward pass. Project page is available at https://r-itk.github.io/globalsplat/