Panel Details

Reality Mapping with 3D Gaussian Splats from Drone and Terrestrial Imagery

Tools, Workflows, Best Practices, and Applications

This workshop presents an end-to-end workflow for producing georeferenced 3D Gaussian Splats from 360-degree video and imagery captured with commodity cameras such as Insta360 and GoPro MAX, as well as aerial imagery collected with drones such as Antigravity. Participants will learn practical methods for capture planning, perspective-image generation, camera-pose estimation, 3DGS training and optimization, georeferencing, quality assessment, publishing, and application development. Supporting products—including point clouds, orthophotos, oriented imagery, and textured 3D meshes—will be discussed as complementary outputs for geospatial analysis, visualization, and comparison with 3D Gaussian Splats. Demonstrations will show how these workflows support infrastructure inspection, asset mapping, visualization, visual positioning, and automated detection of signs and nameplates. The workshop will conclude by identifying research opportunities for the geospatial science community.

Panel Moderator

Speaker

Morakot Pilouk

Morakot Pilouk, Ph.D., is a Senior Principal Software Development Engineer at Esri. Since joining the company in 1996, he has helped advance geospatial technologies spanning 3D, real-time GIS, indoor mapping, reality capture, 3D Gaussian Splatting, Visual SLAM, and positioning systems. He has led the development of major software products and capabilities, including 3D Analyst, Raster, Spatial Analyst, Geostatistical Analyst, ArcScan, and Map Generalization. He currently leads Esri’s 3D and Digital Twin Technology Center and is a key contributor to research at Esri Innovation Labs, with a focus on visual positioning systems and the generation and application of 3D Gaussian Splats from 360-degree videos and images.