AI & ML New Capability

Provides a robust solution for anti-aliasing in Feed-forward Gaussian Splatting, enabling high-fidelity rendering across varying sampling rates and resolutions.

April 1, 2026

Original Paper

AA-Splat: Anti-Aliased Feed-forward Gaussian Splatting

Taewoo Suh, Sungpyo Kim, Jongmin Park, Munchurl Kim

arXiv · 2603.29394

The Takeaway

Existing Gaussian Splatting methods suffer from severe artifacts when rendering at resolutions different from their training data. This band-limiting approach allows for 3D representations that remain visually stable and accurate across a 4x scale range without retraining.

From the abstract

Feed-forward 3D Gaussian Splatting (FF-3DGS) emerges as a fast and robust solution for sparse-view 3D reconstruction and novel view synthesis (NVS). However, existing FF-3DGS methods are built on incorrect screen-space dilation filters, causing severe rendering artifacts when rendering at out-of-distribution sampling rates. We firstly propose an FF-3DGS model, called AA-Splat, to enable robust anti-aliased rendering at any resolution. AA-Splat utilizes an opacity-balanced band-limiting (OBBL) de