A 3D city model is digital model of urban areas that represent terrain surfaces, sites, buildings, vegetation, infrastructure and landscape elements in three-dimensional scale as well as related objects (e.g., city furniture) belonging to urban areas. Their components are described and represented by corresponding two- and three-dimensional spatial data and geo-referenced data. 3D city models support presentation, exploration, analysis, and management tasks in a large number of different application domains. In particular, 3D city models allow "for visually integrating heterogeneous geoinformation within a single framework and, therefore, create and manage complex urban information spaces."
Storage To store 3D city models, both file-based and database approaches are used. There is no single, unique representation schema due to the heterogeneity and diversity of 3d city model contents.
Encoding of components The Components of 3D city models are encoded by common file and exchange formats for 2D raster-based GIS data (e.g., GeoTIFF), 2D vector-based GIS data (e.g., AutoCAD DXF), 3D models (e.g., .3DS, .OBJ), and 3D scenes (e.g., Collada, Keyhole Markup Language) such as supported by CAD, GIS, and computer graphics tools and systems. All components of a 3D city model have to be transformed into a common geographic coordinate system.
Databases A database for 3D city models stores its components in a hierarchically structured, multi-scale way, which allows for a stable and reliable data management and facilitates complex GIS modeling and analysis tasks. For example, the 3D City Database is a free 3D geo database to store, represent, and manage virtual 3D city models on top of a standard spatial relational database. A database is required if 3D city models have to be continuously managed. 3D city model databases form a key element in 3D spatial data infrastructures that require support for storing, managing, maintenance, and distribution of 3D city model contents. Their implementation requires support of a multitude of formats (e.g., based on FME multi formats). As common application, geodata download portals can be set up for 3D city model contents (e.g., virtualcityWarehouse).
CityGML The Open Geospatial Consortium (OGC) defines an explicit XML-based exchange format for 3D city models, CityGML, which supports not only geometric descriptions of 3D city model components but also the specification of semantics and topology information.T. H. Kolbe: Representing and Exchanging 3D City Models with CityGML. 3D Geo-Information Sciences, J. Lee, S. Zlatanova, W. Cartwright, G. Gartner, L. Meng, and M. P. Peterson, Eds. Springer Berlin Heidelberg, 2009, pp. 15–31
CityJSON
CityJSON is a JSON-based format for storing 3D city models. It mostly follows the CityGML data model, but aims to be developer- and user-friendly by avoiding most of the complexities of its usual GML encoding. Due to its simple encoding and the use of JSON, it is also suitable for web applications.Ledoux, H., Arroyo Ohori, K., Kumar, K. et al. CityJSON: a compact and easy-to-use encoding of the CityGML data model. Open geospatial data, softw. stand. 4, 4 (2019). https://doi.org/10.1186/s40965-019-0064-0
Construction
Level of detail 3D city models are typically constructed at various levels of detail (LOD) to provide notions of multiple resolutions and at different levels of abstraction. Other metrics such as the level of spatio-semantic coherence and resolution of the texture can be considered a part of the LOD. For example, CityGML defines five LODs for building models: LOD 0: 2.5D footprints LOD 1: Buildings represented by block models (usually extruded footprints) LOD 2: Building models with standard roof structures LOD 3: Detailed (architectural) building models LOD 4: LOD 3 building models supplemented with interior features.
There exist also approaches to generalize a given detailed 3D city model by means of automated generalization.T. Glander, J. Döllner: Techniques for Generalizing Building Geometry of Complex Virtual 3D City Models. Advances in 3D Geoinformation Systems, (Peter van Oosterom and Sisi Zlatanova and Friso Penninga and Elfriede M. Fendel, ed.), Lecture Notes in Geoinformation and Cartography, Springer, pp. 381–400, 2008. For example, a hierarchical road network (e.g., OpenStreetMap) can be used to group 3D city model components into "cells"; each cell is abstracted by aggregating and merging contained components.
GIS data GIS data provide the base information to build a 3D city model such as by digital terrain models, road networks, land use maps, and related geo-referenced data. GIS data also includes cadastral data that can be converted into simple 3D models as, for example, in the case of extruded building footprints. Core components of 3D city models form digital terrain models (DTM) represented, for example, by TINs or grids.
CAD data Typical sources of data for 3D city model also include CAD models of buildings, sites, and infrastructure elements. They provide a high level of detail, possible not required by 3D city model applications, but can be incorporated either by exporting their geometry or as encapsulated objects.
BIM data Building information models represent another category of geo-spatial data that can be integrated into a 3D city model providing the highest level of detail for building components.
Integration at visualization level Complex 3D city models typically are based on different sources of geodata such as geodata from GIS, building and site models from CAD and BIM. It is one of their core properties to establish a common reference frame for heterogeneous geo-spatial and geo-referenced data, i.e., the data need not to be merged or fused based on one common data model or schema. The integration is possible by sharing a common geo-coordinate system at the visualization level.J. Döllner, B. Hagedorn: Integrating Urban GIS, CAD, and BIM Data By Service-Based Virtual 3D City-Models. Urban and Regional Data Management: UDMS 2007 Annual, (Massimo Rumor and Volker Coors and Elfriede M. Fendel and Sisi Zlatanova, ed.), Taylor & Francis Ltd., Stuttgart, Germany, pp. 157–170, 2007.
Building reconstruction The simplest form of building model construction consist in extruding the footprint polygons of buildings, e.g., taken from the cadaster, by pre-compute average heights. In practice, 3D models of buildings of urban regions are generated based on capturing and analyzing 3D point clouds (e.g., sampled by terrestrial or aerial laser scanning) or by photogrammetric approaches. To achieve a high percentage of geometrically and topologically correct 3D building models, digital terrain surfaces and 2D footprint polygons are required by automated building reconstruction tools such as BREC. One key challenge is to find building parts with their corresponding roof geometry. "Since fully automatic image understanding is very hard to solve, semi-automatic components are usually required to at least support the recognition of very complex buildings by a human operator."N. Haala, M. Kada: An update on automatic 3D building reconstruction. ISPRS Journal of Photogrammetry and Remote Sensing 65 (2010), 570–580. Statistical approaches are common for roof reconstruction based on airborne laser scanning point clouds. H. Huang, C. Brenner, M. Sester: 3D building roof reconstruction from point clouds via generative models. GIS 2011: 16-24.
Fully automated processes exist to generate LOD1 and LOD2 building models for large regions. For example, the Bavarian Office for Surveying and Spatial Information is responsible for about 8 million building models at LOD1 and LOD2.
Visualization The visualization of 3D city models represents a core functionality required for interactive applications and systems based on 3D city models.
Real-time rendering Providing high quality visualization of massive 3D city models in a scalable, fast, and cost efficient manner is still a challenging task due to the complexity in terms of 3D geometry and textures of 3D city models. Real-time rendering provides a large number of specialized 3D rendering techniques for 3D city models. Examples of specialized real-time 3D rendering include: Real-time 3D rendering of road networks on high resolution terrain models.M. Vaaraniemi, M. Treib, R. Westermann: High-Quality Cartographic Roads on High-Resolution DEMs. Journal of WSCG 19(2):41-48, 2011. Real-time 3D rendering of water surfaces with cartography-oriented design.A. Semmo et al.: Real-Time Rendering of Water Surfaces with Cartography-Oriented Design. Proceedings International Symposium on Computational Aesthetics in Graphics, Visualization, and Imaging (CAe), pp. 5–14, 2013. Real-time 3D rendering of day and night sky phenomena.D. Limberger et al.: Single-Pass Rendering of Day and Night Sky Phenomena. Proceedings of the Vision, Modeling, and Visualization Workshop 2012, Eurographics Association, pp. 55-62, 2012. Real-time 3D rendering of grid-based terrain models.F. Losasso, H. Hoppe: Geometry clipmaps: Terrain rendering using nested regular grids. ACM Trans. Graphics (SIGGRAPH), 23(3), 2004. Real-time 3D rendering using different levels of abstraction, ranging between 2D map views and 3D views.Archived at Ghostarchive and the Wayback Machine: Real-time 3D rendering of multiperspective views on 3D city models.Archived at Ghostarchive and the Wayback Machine: http://www.hpi.uni-potsdam.de/doellner/publications/year/2014/2390/PSTD2014.html Multiperspective Views for 3D City Models
Real-time rendering algorithms and data structures are listed by the virtual terrain project.
Service-based rendering Service-oriented architectures (SOA) for visualizing 3D city models offer a separation of concerns into management and rendering and their interactive provision by client applications. For SOA-based approaches, 3D portrayal services are required, whose main functionality represents the portrayal in the sense of 3D rendering and visualization.J. Klimke, J. Döllner: Service-oriented Visualization of Virtual 3D City Models. Directions Magazine, 2012. http://www.directionsmag.com/articles/service-oriented-visualization-of-virtual-3d-city-models/226560 SOA-based approaches can be distinguished into two main categories, currently discussed in the Open Geospatial Consortium:
Web 3D service (W3DS): This type of service handles geodata access and mapping to computer graphics primitives such as scene graphs with textured 3D geometry models as well as their delivery to the requesting client applications. The client applications are responsible for the 3D rendering of delivered scene graphs, i.e., they are responsible for the interactive display using their own 3D graphics hardware. Web view service (WVS): This type of service encapsulates the 3D rendering process for 3D city models at the server side. The server generates views of the 3D scene or intermediate, image-based representations (e.g., virtual panoramas or G-buffer cube mapsJ. Döllner, B. Hagedorn: Server-Based Rendering of Large 3D Scenes for Mobile Devices Using G-Buffer Cube Maps. Web3D '12 Proceedings of the 17th International Conference on 3D Web Technology, pp. 97-100, 2012.), which are streamed and uploaded to requesting client applications. The client applications are responsible for re-construction the 3D scene based on the intermediate representations. Client applications do not have to process 3D graphics data, but to provide management for loading, caching, and displaying the image-based representations of 3D scenes and do not have to process the original (and possibly large) 3D city model.
Map-based visualization