FOSS4G-Asia 2023 Seoul

3D buildings layer is essential to implement geospatial digital twin environment. Considering content construction cost, having 3D buildings of LOD 1 is regarded cost-effective for location-based visualization and simulation. 3D buildings of LOD 1 layer without texture can be easily created by extruding the building heights from building footprint's planar geometry. 3D building models are processed to be a streamable tileset over the web. To apply enormous number of building features at nation-wide scale, a global tileset by merging external tilesets was implemented.

The author has implemented creating 3D buildings of LOD 1 and tiling it as a streamable tileset of 3D Tiles. Per each building feature, 3D COLLADA model with local coordinate was created from buliding footprint's geometry of shapefiles by FME Desktop. In order to manage each building's metadata, KML reference model per building is applied while 3D model is linked as a COLLADA file. At a building footprint's centroid location, a KML is created containing longitude, latitude, altitude and linking a corresponding COLLADA file. The metadata in KML is generated from shapefile's attributes by a custom bash shell script. When KML/COLLADA models are ready, Cesium ion 3D Tiling Pipeline is used to create a per-building tileset of 3D Tiles. As the number of building features grow up to nation-wide scale, managing a single tileset is not only difficult to create but also causing network payloads. To resolve these issues, tilesets are granulated at manageable level and merged as a global tileset by merging external tilesets using open source 3D Tiles Tools. The characteristics of this talk is in its streamlined process workflow rather than each part of the job.