Towards High-Fidelity Gaussian Splatting with Queried-Convolution Neural Networks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kumar, Abhinav, Aumentado-Armstrong, Tristan, Valkov, Lazar, Sharma, Gopal, Levinshtein, Alex, Grzeszczuk, Radek, Kumar, Suren
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918387595083776
author Kumar, Abhinav
Aumentado-Armstrong, Tristan
Valkov, Lazar
Sharma, Gopal
Levinshtein, Alex
Grzeszczuk, Radek
Kumar, Suren
author_facet Kumar, Abhinav
Aumentado-Armstrong, Tristan
Valkov, Lazar
Sharma, Gopal
Levinshtein, Alex
Grzeszczuk, Radek
Kumar, Suren
contents Gaussian Splatting has revolutionized the field of Novel View Synthesis (NVS) with faster training and real-time rendering. However, its reconstruction fidelity still trails behind the powerful radiance models such as Zip-NeRF. Motivated by our theoretical result that both queries (such as coordinates) and neighborhood are important to learn high-fidelity signals, this paper proposes Queried-Convolutions (Qonvolutions), a simple yet powerful modification using the neighborhood properties of convolution. Qonvolutions convolve a low-fidelity signal with queries to output residual and achieve high-fidelity reconstruction. We empirically demonstrate that combining Gaussian splatting with Qonvolution neural networks (QNNs) results in state-of-the-art NVS on real-world scenes, even outperforming Zip-NeRF on image fidelity. QNNs also enhance performance of 1D regression, 2D regression and 2D super-resolution tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2512_12898
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Towards High-Fidelity Gaussian Splatting with Queried-Convolution Neural Networks
Kumar, Abhinav
Aumentado-Armstrong, Tristan
Valkov, Lazar
Sharma, Gopal
Levinshtein, Alex
Grzeszczuk, Radek
Kumar, Suren
Computer Vision and Pattern Recognition
Graphics
Machine Learning
Gaussian Splatting has revolutionized the field of Novel View Synthesis (NVS) with faster training and real-time rendering. However, its reconstruction fidelity still trails behind the powerful radiance models such as Zip-NeRF. Motivated by our theoretical result that both queries (such as coordinates) and neighborhood are important to learn high-fidelity signals, this paper proposes Queried-Convolutions (Qonvolutions), a simple yet powerful modification using the neighborhood properties of convolution. Qonvolutions convolve a low-fidelity signal with queries to output residual and achieve high-fidelity reconstruction. We empirically demonstrate that combining Gaussian splatting with Qonvolution neural networks (QNNs) results in state-of-the-art NVS on real-world scenes, even outperforming Zip-NeRF on image fidelity. QNNs also enhance performance of 1D regression, 2D regression and 2D super-resolution tasks.
title Towards High-Fidelity Gaussian Splatting with Queried-Convolution Neural Networks
topic Computer Vision and Pattern Recognition
Graphics
Machine Learning
url https://arxiv.org/abs/2512.12898