Clemson University's Center for Geospatial Technologies has turned existing drone imagery of Memorial Stadium into a Gaussian splat with Esri's ArcGIS Drone2Map.
The center, known as CCGT, began the work in 2023, when Gingrich's predecessor tested whether drone imagery of Samuel J. Cadden Chapel could become a 3D model. The university's planning director then asked for the whole campus, and CCGT built a 3D mesh of the grounds from more than 70,000 drone images. That mesh had to be split into 10 segments to load without lag.
Drone2Map added a Gaussian splat product in version 2025.2, generated as a 3D tile layer using deep learning. Esri has also brought cloud splat generation to Site Scan for ArcGIS, and Gretna, Louisiana flew the whole city for Gaussian splats in ArcGIS Reality. Jared Gingrich, CCGT's drone and lidar specialist, reprocessed the stadium imagery through the 3D Products template, which outputs point clouds, textured meshes and Gaussian splat layers from the same camera alignment. No new flights were needed, and the stadium mesh and splat processed together in about 13 hours.
The Clemson splat covers Memorial Stadium and the surrounding area as a single segment and, according to Esri, loads much faster than the mesh. Light poles, flagpoles, the spires along the upper tier, goalposts, and tree trunks and branches all resolve in the splat. The mesh rendered the same trees as blocky shapes and gave the spires no definition at all.
Esri does not state the hardware behind the 13 hour run, the splat count, or the drone used for the original capture. Drone2Map documentation lists a separate deep learning package install and specific GPU requirements for splat generation, and splat layers share as a web layer with a .3tz package.
Stephanie Cooper, the university architect and director of planning, design and construction, is exploring the splat for land use management, placing proposed buildings into the existing scene. CCGT is now generating a splat of the complete main campus from its archive of 70,000 images and plans a public step by step workflow, guest lectures, and training sessions open to any department this fall.
The project is detailed in Esri's ArcWatch case study.
