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

• URL: http://arxiv.org/abs/2604.15284v2
• Date: Fri, 17 Apr 2026 07:38:26 GMT
• Title: GlobalSplat: Efficient Feed-Forward 3D Gaussian Splatting via Global Scene Tokens
• Abstract summary: We introduce GlobalSplat, a framework built on the principle of align first, decode later.<n>Our approach learns a compact, global, latent scene representation that encodes multi-view input and resolves cross-view correspondences.<n>On RealEstate10K and ACID, our model achieves competitive novel-view synthesis performance while utilizing as few as 16K Gaussians.
• Score: 41.96744884413287
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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/

Related papers

• Compact Feed-Forward 3D Gaussians via Saliency-Guided Primitive Merging [16.91050950048738]
  We present a structure-aware merging pipeline that takes per-pixel primitives from any feed-forward method and consolidates them into a compact, content-adaptive Gaussian set.<n>As a post-processing module, the pipeline is backbone-agnostic, leveraging the strengths of existing feed-forward methods.
  arXiv  Detail & Related papers  (2026-08-11T09:32:08Z)
• InfiniSplat: Implicit Gaussian Decoding for Large-Baseline Monocular View Synthesis [49.87457834078904]
  InfiniSplat is a feed-forward single-image 3DGS framework that moves from a pixel-aligned representation toward a surface-aligned representation.<n>InfiniSplat achieves state-of-the-art performance compared with single-image feed-forward baselines.
  arXiv  Detail & Related papers  (2026-08-03T16:17:39Z)
• PointSplat: Compact Gaussian Splatting via Human-Centric Prediction [70.45838533838104]
  We propose PointSplat, a novel human-centric approach to producing 3D human representations from input views on the fly.<n>The proposed method first estimates a coarse geometric proxy and performs ray casting to prune redundant points.<n>It then employs a Point-Image Transformer to fuse appearance and geometry features, predicting Gaussian attributes in a single forward pass.
  arXiv  Detail & Related papers  (2026-06-30T17:59:05Z)
• FLAT: Feedforward Latent Triangle Splatting for Geometrically Accurate Scene Generation [72.38727895659405]
  We introduce FLAT, which decodes triangle splats directly from video diffusion latents.<n>On standard benchmarks, FLAT achieves significantly better geometric accuracy compared to state-of-the-art feedforward baselines.<n>We show that a lightweight test-time refinement step converts the predicted triangle soup into a fully opaque, game-engine-ready representation.
  arXiv  Detail & Related papers  (2026-06-23T17:53:41Z)
• Surflo: Consistent 3D Surface Flow Model with Global State [57.57234680235885]
  Surflo decodes oriented 3D surface points by independently transporting them from noise onto the surface via flow matching.<n>To suppress the local inconsistencies inherent to independent per-point decoding, an inference-time guidance term correlates nearby points by injecting a photometric gradient.<n>Surflo matches or surpasses feed-forward baselines on surface metrics, runs an order of magnitude faster than optimization-based methods that require hundreds of views, and is the only feed-forward approach to combine a global latent with arbitrary-resolution decoding.
  arXiv  Detail & Related papers  (2026-06-11T17:48:38Z)
• Learning Structural Latent Points for Efficient Visual Representations in Robotic Manipulation [28.481044299307346]
  Implicit representations lack explicit structural cues, whereas explicit ones preserve geometry but suffer from resolution limits and weak generalization.<n>We propose a novel pretraining framework that learns a hybrid representation-structural latent points.<n>The resulting compact latent preserves coarse structural tendencies, which do not encode precise geometry but capture richer rough shape and semantic information.
  arXiv  Detail & Related papers  (2026-05-20T14:48:01Z)
• Scal3R: Scalable Test-Time Training for Large-Scale 3D Reconstruction [50.5449251266956]
  This paper addresses the task of large-scale 3D scene reconstruction from long video sequences.<n>Recent feed-forward reconstruction models have shown promising results by directly regressing 3D geometry from RGB images without explicit 3D priors or geometric constraints.<n>We propose a novel neural global context representation that efficiently compresses and retains long-range scene information.
  arXiv  Detail & Related papers  (2026-04-09T17:59:50Z)
• S-VGGT: Structure-Aware Subscene Decomposition for Scalable 3D Foundation Models [15.408916900664783]
  Feed-forward 3D foundation models face a key challenge: the quadratic computational cost introduced by global attention.<n>We introduce textbfS-VGGT, a novel approach that addresses redundancy at the structural frame level.<n>S-VGGT is entirely to token-level acceleration methods, allowing the two to be seamlessly combined for compounded speedups.
  arXiv  Detail & Related papers  (2026-03-18T11:42:55Z)
• Global-Aware Edge Prioritization for Pose Graph Initialization [58.77851776918465]
  A core component of Structure-from-Motion (SfM) where images act as nodes and edges encode relative poses.<n>Since geometric verification is expensive, SfM pipelines restrict the pose graph to a sparse set of candidate edges.<n>We address this limitation through the concept of edge prioritization, ranking candidate edges by their utility for SfM.<n>Our approach has three components: (1) a GNN trained with SfM-derived supervision to predict globally consistent edge reliability; (2) multi-minimal-spanning-tree-based pose graph construction guided by these ranks; and (3) connectivity-aware score modulation that reinforces weak regions
  arXiv  Detail & Related papers  (2026-02-25T14:44:53Z)
• What matters for Representation Alignment: Global Information or Spatial Structure? [64.67092609921816]
  Representation alignment (REPA) guides generative training by distilling representations from a strong, pretrained vision encoder to intermediate diffusion features.<n>We investigate a fundamental question: what aspect of the target representation matters for generation, its textitglobal revisionsemantic information.<n>We replace the standard projection layer in REPA with a simple convolution layer and introduce a spatial normalization layer for the external representation.
  arXiv  Detail & Related papers  (2025-12-11T16:39:53Z)
• ${C}^{3}$-GS: Learning Context-aware, Cross-dimension, Cross-scale Feature for Generalizable Gaussian Splatting [16.868578618340262]
  Generalizable Gaussian Splatting aims to synthesize novel views for unseen scenes without per-scene optimization.<n>We propose $mathbfC3$-GS, a framework that enhances feature learning by incorporating context-aware, cross-dimension, and cross-scale constraints.<n>Our architecture integrates three lightweight modules into a unified rendering pipeline, improving feature fusion and enabling synthesis without requiring additional supervision.
  arXiv  Detail & Related papers  (2025-08-28T13:12:18Z)
• R3GS: Gaussian Splatting for Robust Reconstruction and Relocalization in Unconstrained Image Collections [9.633163304379861]
  R3GS is a robust reconstruction and relocalization framework tailored for unconstrained datasets.<n>To mitigate the adverse effects of transient objects on the reconstruction process, we ffne-tune a lightweight human detection network.<n>To address the challenges posed by sky regions in outdoor scenes, we propose an effective sky-handling technique that incorporates a depth prior as a constraint.
  arXiv  Detail & Related papers  (2025-05-21T09:25:22Z)
• CompGS++: Compressed Gaussian Splatting for Static and Dynamic Scene Representation [60.712165339762116]
  CompGS++ is a novel framework that leverages compact Gaussian primitives to achieve accurate 3D modeling.<n>Our design is based on the principle of eliminating redundancy both between and within primitives.<n>Our implementation will be made publicly available on GitHub to facilitate further research.
  arXiv  Detail & Related papers  (2025-04-17T15:33:01Z)
• MCGS: Multiview Consistency Enhancement for Sparse-View 3D Gaussian Radiance Fields [100.90743697473232]
  Radiance fields represented by 3D Gaussians excel at synthesizing novel views, offering both high training efficiency and fast rendering.<n>Existing methods often incorporate depth priors from dense estimation networks but overlook the inherent multi-view consistency in input images.<n>We propose a view synthesis framework based on 3D Gaussian Splatting, enabling scene reconstruction from sparse views.
  arXiv  Detail & Related papers  (2024-10-15T08:39:05Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.