Learning to Adaptively Allocate Gaussians for Arbitrary-Scale Image Super-Resolution

2026-06-28

AI Summary

This work addresses the limitation of existing super-resolution methods, which typically support only fixed scales and struggle to efficiently reconstruct image details at arbitrary scaling factors. The authors propose QuADA-GS, a novel model that introduces Gaussian splatting into a feed-forward arbitrary-scale super-resolution framework for the first time. By employing a neural routing mechanism to assess local content complexity, the model dynamically allocates Gaussian primitive density and leverages hierarchical pointer convolutions to enable efficient decoding over irregular topologies. This approach facilitates continuous-scale representation learning and O(1) spatial neighborhood lookup, achieving state-of-the-art performance in arbitrary-scale super-resolution while maintaining low latency and minimal memory consumption.

Abstract

In computer graphics, visual content is continuously warped, zoomed and resampled. This occurs when engines upscale frames, users zoom into 3D scenes, or foveated VR applies varying scaling. Handling these transformations requires Arbitrary-Scale Super-Resolution (ASR). Traditional models, designed for fixed scales, typically predict at a lower integer scale (e.g., x4) and rely on sub-optimal interpolation for continuous resolutions, compromising quality. Furthermore, most methods process pixels uniformly. Since fine details are sparse, this creates overhead; efficiency dictates concentrating resources only where structural complexity demands it. While implicit models and Gaussian Splatting (GS) enable continuous representation, GS is advantageous due to adaptive densification. However, transitioning GS into a feed-forward model for ASR is non-trivial. Standard GS optimization needs high-resolution gradients to drive primitive growth, which are unavailable during inference. Thus, the network must autonomously predict GS densification from low-resolution inputs. To solve this, we propose QuADA-GS. After encoding inputs into a latent space, a Neural Routing Architecture evaluates local complexity to distribute a global budget, assigning specific upsampling factors to features to avoid redundant processing. Features are dynamically densified based on these factors, forming an irregular topology decoded into 2D Gaussian primitives. To coordinate features before decoding, we introduce Hierarchical Pointer Convolution. This non-grid operator achieves O(1) neighbor lookup complexity, facilitating efficient spatial communication and bypassing dense bottlenecks. Experiments show QuADA-GS achieves state-of-the-art ASR performance, maintaining low latency and a lean memory footprint.

Problem

• Arbitrary-Scale Super-Resolution
• Gaussian Splatting
• adaptive resource allocation
• continuous resolution
• image upscaling

Innovation

• Arbitrary-Scale Super-Resolution
• Gaussian Splatting
• Neural Routing Architecture
• Adaptive Densification
• Hierarchical Pointer Convolution