GaussianSR: 3D Gaussian Super-Resolution with 2D Diffusion Priors

TL;DR

GaussianSR introduces a novel 3D super-resolution method that leverages 2D diffusion priors and Gaussian splatting to efficiently generate high-resolution novel views from low-resolution inputs.

Contribution

The paper proposes a new 3D super-resolution approach combining 3D Gaussian splatting with 2D diffusion priors and introduces techniques to reduce stochastic disturbances during training.

Findings

Achieves high-quality high-resolution view synthesis from low-resolution inputs.

Outperforms previous methods in rendering speed and quality.

Effective on both synthetic and real-world datasets.

Abstract

Achieving high-resolution novel view synthesis (HRNVS) from low-resolution input views is a challenging task due to the lack of high-resolution data. Previous methods optimize high-resolution Neural Radiance Field (NeRF) from low-resolution input views but suffer from slow rendering speed. In this work, we base our method on 3D Gaussian Splatting (3DGS) due to its capability of producing high-quality images at a faster rendering speed. To alleviate the shortage of data for higher-resolution synthesis, we propose to leverage off-the-shelf 2D diffusion priors by distilling the 2D knowledge into 3D with Score Distillation Sampling (SDS). Nevertheless, applying SDS directly to Gaussian-based 3D super-resolution leads to undesirable and redundant 3D Gaussian primitives, due to the randomness brought by generative priors. To mitigate this issue, we introduce two simple yet effective techniques to reduce stochastic disturbances introduced by SDS. Specifically, we 1) shrink the range of diffusion timestep in SDS with an annealing strategy; 2) randomly discard redundant Gaussian primitives during densification. Extensive experiments have demonstrated that our proposed GaussainSR can attain high-quality results for HRNVS with only low-resolution inputs on both synthetic and real-world datasets. Project page: https://chchnii.github.io/GaussianSR/