NoDrift3R Raymap-Guided Coupling for Drift-Robust Unposed Feed-Forward 3D Reconstruction

Jul 16, 2026

Pose-Free Feed-forward 3D Gaussian Splatting (3DGS) enables fast scene reconstruction from unposed image sequences, but quality drops on long sequences because cumulative camera pose drift propagates errors into geometry and limits rendering fidelity. The work identifies pose drift as the main bottleneck and notes instability from entangled geometry–pose optimization under rendering supervision. It proposes a Raymap-Guided Coupling Module (RGC) that explicitly couples geometry and appearance with joint RGB, raymap consistency, and camera regularization, plus Dual-Frequency Viewpoint Scheduling to stabilize training over long temporal ranges.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages19

Reading Time48 min

Abstract

Introduction

Pose drift in pose-free feed-forward 3DGS

Training instability in pose-free paradigms

Dual-Frequency Viewpoint Scheduling

Raymap-Guided Coupling (RGC) framework

arXiv :2607 .07 168v 1 [ cs .CV] 8 Jul 2026 NoDrift3R: Raymap-Guided Coupling for Drift-Robust Unposed Feed-Forward 3D Reconstruction Xiangyu Sun 1 ⋆, Liu Liu2 B , Seungkwon Yang3 , Jingbing Han2 , Seungtae Nam3 , Zhizhong Su2, and Eunbyung Park3 B 1 Sungkyunkwan University, South Korea 2 Horizon Robotics, China 3 Yonsei University, South Korea Abstract. Pose-Free Feed-forward 3D Gaussian Splatting (3DGS) has recently emerged as a powerful paradigm for fast scene reconstruction. However, its performance degrades significantly in long image sequences due to cumulative camera pose estimation drift, which propagates errors into geometric modeling and severely limits rendering fidelity. In this work, we revisit the long-sequence bottleneck and identify pose drift as the primary factor restricting reconstruction quality. Furthermore, while SfM-based pseudo ground-truth poses introduce sensor noise, purely rendering-based supervision often leads to optimization instability and local minima due to the entangled optimization of geometry and pose. To address the challenges, we propose a synergistic posefree framework that explicitly couples geometry and appearance via a Raymap-Guided Coupling Module (RGC) . Concretely, we anchor Gaussian centers to raymap-induced geometry and jointly optimize RGB reconstruction, raymap consistency, and camera regularization under a unified objective, yielding a bidirectional feedback loop: stronger geometry improves rendering, and appearance supervision in turn refines geometry and pose. To further stabilize learning across wide temporal ranges, we introduce a Dual-Frequency Viewpoint Scheduling strategy that combines easy-to-hard interval expansion with replay of short-interval pairs. Extensive experiments across in-domain and cross-domain datasets show consistent gains in both rendering and pose estimation, with notably improved robustness on long sequences. Ablation studies validate our central insight: explicitly designed geometry-appearance synergy is the key to scalable and drift-robust pose-free feed-forward 3D reconstruction. Keywords: Pose Drift · Dual-Frequency schedule · Gaussian Splatting 1 Introduction Feed-forward 3D Gaussian Splatting (3DGS) has recently emerged as a powerful paradigm for rapid 3D reconstruction and novel view synthesis by predicting ⋆ Intern at Horizon Robotics B Corresponding authors. 2 X. Sun et al. Fig. 1: Overview of our synergistic pose-free framework for feed-forward 3D reconstruction. Our method effectively suppresses the pose drift problem, especially in longsequence settings. Left: representative failure cases of existing methods under pose drift. Right: our pipeline and outputs (camera poses and 3D Gaussians) . scene representations directly from multi-view images. Compared to conventional per-scene optimization methods [20, 26], feed-forward approaches offer significantly faster reconstruction times, opening the door to applications that require fast 3D reconstruction in various practical scenarios. Early feed-forward pipelines typically assume known camera poses and focus on predicting Gaussian representations conditioned on calibrated multi-view observations [1, 3 , 16 , 27 , 41] . More recently, pose-free feed-forward models have been proposed to jointly infer camera geometry and scene representation directly from unposed image sequences [14, 18 , 33 , 42 , 43], thereby expanding the applicability of feed-forward 3DGS to settings where camera poses are not available in inference. Despite encouraging progress, pose-free feed-forward 3DGS remains challenging, particularly for long sequences or wide-baseline viewpoints. A central difficulty arises from the strong coupling between predicted camera parameters and scene representation: errors in pose estimation can directly distort the reconstructed representation, while inaccuracies in scene structure can further degrade pose predictions. This feedback loop often leads to accum

FAQs

How does the Raymap-Guided Coupling Module (RGC) improve reconstruction?

RGC explicitly couples geometry and appearance by anchoring Gaussian centers to raymap-induced geometry and jointly optimizing RGB reconstruction, raymap consistency, and camera regularization in a unified objective.

Why introduce Dual-Frequency Viewpoint Scheduling?

It stabilizes learning over wide temporal ranges by expanding from easy (high-overlap) to hard (low-overlap) intervals while replaying short-interval pairs to preserve local geometric consistency and suppress drift.

Jul 16, 2026

Upload by Theodore