Until recently, the software available for 3D city modelling focused more on the 3D part of things over specific city modelling needs. But as city models get more complex, boundaries are blurring between what is clearly BIM, CAD, GIS and visualisation. By Dan Campbell
Over the past few years there have been substantial changes in city modelling. What and how much one can model and the tools available to tackle these tasks are evolving at a rapid rate. As their scale and scope expand, so do the expectations regarding what a city model can reveal, explain and predict. There are two fundamental decisions one faces when developing a city model; what to model, and how to model it. Traditionally, the first question of what to model has been easy to answer; by definition, a city model was a collection of buildings. This focus on urban form comes from the primary city model users who were architects and urban designers. The early digital city models replicated the simple massing of the established tools of wooden context models, and played a similar role in providing the basic visual analysis. Another part of the ‘what’ question is how much to model. That was also easy to answer. Hardware and software limits kept the scale of what could be modelled fairly modest. For the most part, ‘city model’ has been overly generous title; most models rarely tackled areas beyond the neighbourhood scale.
Until recently, the software available for 3D city modelling has focused more on the 3D part of things over specific city modelling needs. General purpose tools such 3ds Max, SketchUp or Maya are very flexible and powerful, and they can be readily applied to create visually robust city models. The common practice has been to model the buildings using general purpose 3D software. Once completed, the model could be interrogated to provide information about shadowing, views, massing and other visual aspects based on geometric properties. By applying photo textures to the buildings and terrain, engaging photo-real output could be produced, as long as one were willing to commit their PC to countless hours of rendering time.
This process, within limitations, has worked well, and has met basic urban design needs. At their most basic level, city models created for game environments, movie CGI or city government use and share basic content elements. However, it is when there is a need for something beyond visual fidelity that these multipurpose 3D tools reveal their shortcomings.
While useful, these types of models are limited in their analysis capabilities. Essentially, they are “pretty but dumb”. In our world where data is all important, a traditional 3D model is “what you see is what you get”, without any attributes attached to the buildings to enable any analysis beyond basic visual review. Equally important is the fact that most city governments lack the resources to do this detailed modelling at the full city scale, and to ensure that the model is kept current. Let us assume that it takes a minimum of 20 minutes to construct a very simple digital representation of building using a traditional modelling application. The 20 minutes required to model a single building grows to over 11 years to complete a modest sized city.
That is not a sensible, economical or sustainable approach. Another factor to consider is the skill set required to work efficiently with 3D modelling applications. These are demanding programmes with a steep learning curve and municipal governments may find it difficult to justify having employees with such specialised skill set. Many rely on staff with a general working knowledge of a range of applications. In addition, these types of programmes were not designed to work with city scale data. While it is possible to work efficiently at the neighbourhood level, as the scale expands there is too much geometry to move around.
The big change
The big change has been the introduction of purpose built city modelling applications like Esri’s CityEngine and Autodesk’s InfraWorks. These type of tools, unlike general purpose 3D modelling software, have been built from scratch to address the need to work efficiently at the city scale, take advantage of existing GIS and BIM data, offer analytic capabilities and provide social connections with the public and the development community. The essential skill for working with these tools shifts from a focus on geometry and its manipulation to one on data. As most city governments have staff familiar with data and GIS skills, these new generation city modelling tools enable an organisation to quickly develop a functional and intelligent model. Support and development of the city models can be shifted from a small group of experts to a large set of staff with general skills.
A set of building footprints with some basic building height attributes, a digital elevation model and existing GIS data are enough to get one started. Rather than someone painstakingly transferring information from plans, manipulating vertices and faces to create a building model from scratch, the software uses the data attributes to generate the geometry on the fly.
While currently a data derived model is not as visually rich as a traditional geometry derived model, the trade-off, especially when the role of model as a city management tool is considered, is well worth it. The analysis capabilities, opportunities to use existing GIS and CAD data, and perhaps most importantly, the ability to work at the full city scale more than compensate for the model being a little less pretty. More important than pretty is open and intelligent. In the era of open data and open standards, there is an expectation that any and all data, be it CityGML, KMZ, or IFC can be easily consumed, displayed, analysed, and shared.
The most current development in city modelling is the growing support for LiDAR and point cloud data. LiDAR has been playing a secondary role as a source for the generation of geometry, building footprints, and attributes for building height. Now however, with more powerful hardware and software, the point cloud data can be viewed, analysed and themed directly in the city modelling application. As higher resolution LiDAR becomes more readily available, and computing power increases, the need to translate point cloud data to geometry might diminish. The point cloud itself could become the foundation for the model. All three of city model components — detailed geometry, data-derived geometry, and point clouds — are complementary to each other. They can be used together in a city modelling application to create an environment that is both visually robust, and efficiently intelligent.
With all the changes to city modelling, one wonders how the discipline will continue to evolve and change. For many, 3D is becoming the new standard work environment, with 2D work relegated for specific, more abstract application. For a generation that has grown up with Google Earth, 3D gaming and 3D movies, this only makes sense. Perhaps now is an appropriate time to drop the 3D part from the name; the assumption being that all city modelling, is of course, in 3D.
The challenges
It is exciting and important that we are now modelling our infrastructure, and other hidden aspects of the city, but equally important is the modelling of the invisible and the abstract. City modelling is just now emerging from its initial, rather crude state of development. Cities are incredibly complex, organic entities. It doesn’t make much sense to suggest that one can even begin to digitally replicate a city with some simple geometry and photo-textures. The challenge will be how to model regulatory zones indistinct boundaries, the changing seasons, endless varieties of landscape, pedestrians strolling down a street in the evening, the different times of day, a rainy afternoon in February, the sounds of traffic, and the forever changing visual fabric.
The boundaries between what is clearly BIM, CAD, GIS and visualisation, are becoming less distinct, often making it difficult to choose the best tool to apply to a specific task. Perhaps the era of specialised tools for most users is reaching its end. Consider all the individual devices we used to carry with us only a few years ago. Having an individual MP3 player, PDA, camera, phone, video player, and portable gaming device was common. Now all this is handled by a single smartphone. Is it so difficult to imagine that in the near future one may only need a single application to efficiently deal with BIM, CAD, GIS and visualisation?
