RoSplat: Robust Feed-Forward Pixel-wise Gaussian Splatting for Varying Input Views and High-Resolution Rendering

2026-05-13

AI Summary

This work addresses the limitations of existing pixel-wise feedforward Gaussian splatting methods, which often produce overly bright renderings when the number of input views varies and suffer from hole-like artifacts at high resolutions due to inaccurate Gaussian scale estimation. To overcome these issues, the authors propose a robust feedforward pixel-wise Gaussian splatting approach that enforces brightness consistency across varying view counts through alpha normalization and improves the accuracy of Gaussian scale estimation by introducing a 3D-sampling-based regularization term. Evaluated on standard benchmarks, the proposed method significantly outperforms current baselines, achieving more stable and higher-quality novel view synthesis under both variable input view counts and high-resolution rendering conditions.

Abstract

Generalizable 3D Gaussian Splatting has recently emerged as an efficient approach for novel-view synthesis, enabling feed-forward synthesis from only a few input views. However, existing pixel-wise feed-forward methods suffer from over-bright renderings when the number of input views varies during inference, as well as insufficient supervision for accurate Gaussian scale estimation, which leads to hole artifacts, particularly in high-resolution renderings. To address these issues, we identify that the over-brightness is caused by the varying number of overlapping Gaussians and propose a simple alpha normalization strategy to maintain brightness consistency across different number of input views. In addition, we introduce an auxiliary 3D sampling-based regularizer to improve Gaussian scale estimation, thereby mitigating hole artifacts in high-resolution rendering. Experiments on benchmark datasets demonstrate that our method significantly improves baseline models under varying input-view and high-resolution rendering settings.

Problem

• Gaussian Splatting
• novel-view synthesis
• high-resolution rendering
• hole artifacts
• brightness inconsistency

Innovation

• Gaussian Splatting
• novel-view synthesis
• alpha normalization
• scale regularization
• high-resolution rendering