Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TokenSplat: Token-aligned 3D Gaussian Splatting for Feed-forward Pose-free Reconstruction

Yihui Li, Chengxin Lv, Zichen Tang, Hongyu Yang, Di Huang

TL;DR

TokenSplat introduces a Token-aligned Gaussian Prediction module that aligns semantically corresponding information across views directly in the feature space, enabling coherent reconstruction and stable pose estimation without iterative refinement.

Abstract

We present TokenSplat, a feed-forward framework for joint 3D Gaussian reconstruction and camera pose estimation from unposed multi-view images. At its core, TokenSplat introduces a Token-aligned Gaussian Prediction module that aligns semantically corresponding information across views directly in the feature space. Guided by coarse token positions and fusion confidence, it aggregates multi-scale contextual features to enable long-range cross-view reasoning and reduce redundancy from overlapping Gaussians. To further enhance pose robustness and disentangle viewpoint cues from scene semantics, TokenSplat employs learnable camera tokens and an Asymmetric Dual-Flow Decoder (ADF-Decoder) that enforces directionally constrained communication between camera and image tokens. This maintains clean factorization within a feed-forward architecture, enabling coherent reconstruction and stable pose estimation without iterative refinement. Extensive experiments demonstrate that TokenSplat achieves higher reconstruction fidelity and novel-view synthesis quality in pose-free settings, and significantly improves pose estimation accuracy compared to prior pose-free methods. Project page: https://kidleyh.github.io/tokensplat/.

TokenSplat: Token-aligned 3D Gaussian Splatting for Feed-forward Pose-free Reconstruction

TL;DR

TokenSplat introduces a Token-aligned Gaussian Prediction module that aligns semantically corresponding information across views directly in the feature space, enabling coherent reconstruction and stable pose estimation without iterative refinement.

Abstract

We present TokenSplat, a feed-forward framework for joint 3D Gaussian reconstruction and camera pose estimation from unposed multi-view images. At its core, TokenSplat introduces a Token-aligned Gaussian Prediction module that aligns semantically corresponding information across views directly in the feature space. Guided by coarse token positions and fusion confidence, it aggregates multi-scale contextual features to enable long-range cross-view reasoning and reduce redundancy from overlapping Gaussians. To further enhance pose robustness and disentangle viewpoint cues from scene semantics, TokenSplat employs learnable camera tokens and an Asymmetric Dual-Flow Decoder (ADF-Decoder) that enforces directionally constrained communication between camera and image tokens. This maintains clean factorization within a feed-forward architecture, enabling coherent reconstruction and stable pose estimation without iterative refinement. Extensive experiments demonstrate that TokenSplat achieves higher reconstruction fidelity and novel-view synthesis quality in pose-free settings, and significantly improves pose estimation accuracy compared to prior pose-free methods. Project page: https://kidleyh.github.io/tokensplat/.
Paper Structure (25 sections, 13 equations, 14 figures, 5 tables)

This paper contains 25 sections, 13 equations, 14 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Feed-forward 3D Reconstruction
  4. Pose-Free 3D Scene Reconstruction
  5. Method
  6. Problem Formulation
  7. Architecture
  8. Asymmetric Dual-Flow Decoder
  9. Token Fusion for Scene Reconstruction
  10. Camera Pose Estimation Head
  11. Loss Functions
  12. Experiment
  13. Experimental Settings
  14. Experimental Results
  15. Ablation Analysis
  16. ...and 10 more sections

Figures (14)

  • Figure 1: Overview of TokenSplat. TokenSplat performs feed-forward 3D Gaussian reconstruction and camera pose estimation from unposed images. A shared ViT encoder extracts image tokens, which are processed by the Canonical Scene Decoder and the Asymmetric Dual-Flow Decoder (ADF-Decoder). The fused tokens are then used by the Token-aligned Gaussian Prediction module and the camera pose head to generate dense 3D Gaussians and accurate poses.
  • Figure 2: Qualitative comparison on RE10K and ScanNet under varying numbers of reference views.
  • Figure 3: Cross-dataset generalization from RE10K to ScanNet.
  • Figure 4: Scene-level visualizations and multiple novel viewpoints renderings.
  • Figure A1: Structure of our Gaussian Prediction Head.
  • ...and 9 more figures