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

arXiv cs.CV / 4/1/2026

📰 NewsSignals & Early TrendsIdeas & Deep AnalysisModels & Research

Key Points

  • The paper introduces AA-Splat, a new feed-forward 3D Gaussian Splatting (FF-3DGS) approach designed to fix artifact issues caused by incorrect screen-space dilation filters, especially under out-of-distribution sampling resolutions.
  • AA-Splat uses an opacity-balanced band-limiting (OBBL) pipeline, combining a 3D band-limiting post-filter that integrates multi-view maximal frequency bounds with an opacity balancing step that improves seamless integration of pixel-aligned Gaussians during rendering.
  • By band-limiting the reconstructed scene representations and eliminating degenerate Gaussians, AA-Splat targets robust anti-aliased rendering at any resolution.
  • Experiments report average 5.4–7.5 dB PSNR gains on novel view synthesis (NVS) compared with the DepthSplat baseline across resolutions ranging from 4× to 1/4×.
  • The authors state they will make the code publicly available, which should accelerate adoption and benchmarking by the research community.

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) design, which combines two components: a 3D band-limiting post-filter integrates multi-view maximal frequency bounds into the feed-forward reconstruction pipeline, effectively band-limiting the resulting 3D scene representations and eliminating degenerate Gaussians; an Opacity Balancing (OB) to seamlessly integrate all pixel-aligned Gaussian primitives into the rendering process, compensating for the increased overlap between expanded Gaussian primitives. AA-Splat demonstrates drastic improvements with average 5.4\sim7.5dB PSNR gains on NVS performance over a state-of-the-art (SOTA) baseline, DepthSplat, at all resolutions, between 4\times and 1/4\times. Code will be made available.