Date of Award


Degree Name

MS in Computer Science


Computer Science


College of Engineering


Jonathan Ventura

Advisor Department

Computer Science

Advisor College

College of Engineering


Over the past year, 3D Gaussian Splatting has become a widespread research topic of interest. It has proven itself against other popular Novel View Synthesis algorithms
like Neural Radiance Fields (NeRF) by replacing the approximation function with
a modified point cloud. This modified point cloud allows for the rendering of novel
views in real time. While this gives Gaussian Splatting an advantage over algorithms
like NeRF, it shares the same issue of requiring prior camera intrinsics such as a Point
Cloud and Camera Poses. Most Gaussian Splatting pipelines use Colmap Sparse, a
Structure-from-Motion (SfM) technology that efficiently registers the necessary point
clouds and camera poses. However, these features require a surplus number of images that are often unavailable. In this thesis, we investigate and compare other types of SfM and Multi-View-Stereo (MVS) technologies that use both traditional and deep-learning-based algorithms. The objective is to see if these methods’ increased point cloud density results in better Gaussian Splatting results, provided a low number of images. This evaluation is performed with quantitative metrics (PSNR, SSIM, Point
Cloud Size, Runtime), and qualitative assessment (rendering Gaussian Splats and
Point Clouds).