SGD: Street View Synthesis with Gaussian Splatting and Diffusion Prior

TL;DR

This paper introduces a novel street view synthesis method that combines 3D Gaussian Splatting with a diffusion prior and multi-modal data to improve rendering quality at novel viewpoints in autonomous driving scenarios.

Contribution

It enhances 3D Gaussian Splatting by integrating a diffusion model and multi-modal data, enabling better generalization to unseen viewpoints in street scenes.

Findings

Outperforms current state-of-the-art models in street view synthesis.

Improves rendering quality at viewpoints deviating significantly from training views.

Leverages LiDAR and adjacent frame images to enhance spatial and contextual understanding.

Abstract

Novel View Synthesis (NVS) for street scenes play a critical role in the autonomous driving simulation. The current mainstream technique to achieve it is neural rendering, such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS). Although thrilling progress has been made, when handling street scenes, current methods struggle to maintain rendering quality at the viewpoint that deviates significantly from the training viewpoints. This issue stems from the sparse training views captured by a fixed camera on a moving vehicle. To tackle this problem, we propose a novel approach that enhances the capacity of 3DGS by leveraging prior from a Diffusion Model along with complementary multi-modal data. Specifically, we first fine-tune a Diffusion Model by adding images from adjacent frames as condition, meanwhile exploiting depth data from LiDAR point clouds to supply additional spatial information. Then we apply the Diffusion Model to regularize the 3DGS at unseen views during training. Experimental results validate the effectiveness of our method compared with current state-of-the-art models, and demonstrate its advance in rendering images from broader views.