FreeStreamGS: Online Feed-forward 3D Gaussian Splatting from Unposed Streaming Inputs

2026-06-02

AI Summary

This work addresses the rendering artifacts in existing online novel view synthesis methods caused by insufficient multi-view consistency. The authors propose an online feed-forward framework that operates without future frames and supports streaming image inputs without requiring known camera poses. Built upon 3D Gaussian Splatting (3DGS), the method introduces a decoupled intrinsic recovery head to eliminate cumulative errors in camera intrinsics and employs a dynamic point optimization strategy with offset adjustments to mitigate pose-depth coupling drift. These designs collectively preserve geometric and rendering consistency over extended sequences. Experiments demonstrate that, using only current and past frames, the proposed approach achieves rendering quality comparable to state-of-the-art offline 3DGS methods and significantly outperforms existing online alternatives.

Abstract

Feed-forward 3D Gaussian Splatting (3DGS) allows efficient and high-fidelity novel view synthesis (NVS) from an offline recorded image sequence. However, achieving online NVS from streaming and unposed image inputs remains challenging. Although online feed-forward geometric estimation methods have been proposed for streaming depth and point cloud recovery, they cannot be adapted to NVS due to severe rendering artifacts. This is because NVS demands stricter multi-view consistency in Gaussian scales and pose-geometry alignment; even minor deviations would accumulate over time and visibly degrade rendering quality. To this end, we propose FreeStreamGS, a robust online feed-forward framework for efficient and high-quality NVS. We introduce two key mechanisms: a Decoupled Intrinsic Recovery Head that removes cumulative camera intrinsic bias and prevents scene scale jitter during long-term streaming, and a Dynamic Point Refinement Offset strategy that relaxes rigid unprojection to correct coupled pose-depth drift. Extensive experiments show that FreeStreamGS achieves rendering quality competitive with state-of-the-art offline feed-forward 3DGS methods, despite operating without access to future frames.

Problem

• online novel view synthesis
• unposed streaming inputs
• 3D Gaussian Splatting
• multi-view consistency
• rendering artifacts

Innovation

• Online 3D Gaussian Splatting
• Novel View Synthesis
• Streaming Input
• Pose-Depth Drift Correction
• Decoupled Intrinsic Recovery