Danesfield Courier: 3D Models from Point Clouds for Efficient Transmission and Rendering

An important application for geospatial 3D models is fast transmission and rendering in web and mobile applications, especially for limited bandwidth connections.  Point cloud models are commonly used within the NGA and are derived from multiple sources including Lidar and aerial/satellite imagery and video.  However, point clouds are too large for efficient transmission and rendering.  Kitware's research, initiated on the IARPA CORE3D program, has demonstrated procedural building models from satellite-derived point clouds for more compact mesh representations with segmentation into layers of terrain, vegetation, and buildings. These models have not yet been adapted for fast transmission and rendering by converting to OGC standard representations like 3D Tiles. Although 3D Tiles is an open standard, existing tools to convert models to this format are proprietary and do not generate multiple levels of detail (LOD) to optimize transmission. \n \nKitware proposes Danesfield Courier, extending Danesfield, our CORE3D system, to generate output suitable for fast transmission and rendering using multiple levels of detail, texture optimization, and tiling. We will generalize our tools to a wider variety of point cloud sources including Lidar, aerial video, and satellite imagery. As determined in Phase 1, we will focus on 3D Tiles for use with Cesium and ATAK clients, but will also support other standards such as CDB.  We will develop these tools as open source software that is free of licensing fees and restrictions.  Our web application, Danesfield, will demonstrate end-to-end processing of multiple data sources and 3D visualization of results.  Integration of the tools into existing desktop applications, such as TeleSculptor and ParaView, will expose these capabilities to analysts already using those applications. Docker containers containing the command line utilities and APIs will enable these tools to plug directly into the NGA infrastructure.  These advances will enable remote users to rapidly access the quality 3D data that they need, when they need it, to make mission-critical decisions. \n