Reconstruction of underwater wreckage scenes plays a critical role in maritime archaeology. Understanding the context and details of these submerged sites is essential for both preservation and research. Accurately representing these scenes digitally allows for the continued exploration of these sites without the need for extraneous expeditions. The Naval History and Heritage Command (NHHC), as managers of the Department of the Navy’s sunken military craft, provided NIWC Pacific with hundreds of hours of deep-sea underwater video data of different World War II (WWII) shipwrecks collected by Vulcan, LLC. This data was released to us for the purposes of historical, archaeological, and educational research. Our team first pieced together individual video feeds using traditional photogrammetry techniques to create detailed three-dimensional representations of the wreck sites. We then successfully utilized two state-of-the-art frameworks for 3D reconstruction – Neural Radiance Fields (NeRFs) and Gaussian splatting – to create novel views for these comprehensive underwater scene reconstructions. In this paper we share qualitative and quantitative analysis of these techniques and explore the positives and negatives of each as they perform on our unstructured deep sea data.