A little bit of context – we have been working with 3D digital city models for a number of years now. Our clients, typically large cities dotted across main-land Europe, have identified these models as a vital part of their management strategy for the urban environment, and they have invested heavily in the data, software and staff resources to make them happen. Their use is so prevalent that you could say that it is almost a legal prerequisite for any significant development project…..
So when I heard this statement from John Carpenter I was slightly put out. I hasten to add that I don’t think that this is his personal view, rather the sentiment carried in the statement is born out of a 3 month market perception study, that we participated with, done by the OS at the beginning of the year. Indeed the OS have been doing extensive research in how they can incorporate 3D in to their product set for a number of years now. What this does reveal is the lack of general understanding of where the value of 3D lies. But I also think that it glosses over the problems that are still to be addressed with the handling of 3D data – things that we perhaps take for granted when working with 2D. With the ever wider adoption of 3D to present information across a range of applications, the issues of accuracy of ‘representation’ have become more apparent – yet have not had the level of investigation applied that is common within conventional 2D cartography.
The world we live is in a complex, three dimensional place - yet
yet representation of that world has to date predominately been a process of abstraction into a 2D format.
The output of this process requires interpretation by the user, which introduces the potential for miscommunication
For applications where there is a need to understand our interaction within a place, such as an urban development project, accurate presentation of the actual (and planned) environment is very important For such projects architects, designers and engineers have always used 3D physical models, realising that drawings are the preserve of the professional and actively disinterest the wider stakeholder community
But we are now seeing a greater adoption of virtual (digital) 3D city models because of their ability to convey information more effectively than traditional methods. They offer the potential of more realism (simulation of shifting light and changing seasons), more versatility (respond to and incorporate change), but more than that, they hold the potential to be a map. That is to say, that this is not simply a picture of a street in Linz, but is a base data layer that can provide added understanding to a range of abstract data sets – noise modelling being a good example as it is significantly enhanced with the third dimension.
As pressure on our cities develops, there is greater need to consult with an ever wider group of interested parties on the effective use of our diminishing resources. Urban planning is now a multi-disciplinary pursuit – with each set of professionals bringing to the table their own set of understandings. Effective communication between these groups is vital. Repeated case studies (and I can add our own anecdotal experience to this) have shown that these professionals gain real insight from information presented in 3D There are sound business drivers behind their adoption: productivity and efficiency are enhanced though the entire lifecycle from design and build to operate and manage.
And this “discussion” must be held with the entire stake-holder community – the public are now an integral part of almost any process, but in particular planning. They are now empowered by law, technology and expectation, as never before to influence development. It is therefore vital to engage with them early to ensure their understanding of the proposal, thereby reducing potential conflict.
So if 3D is obviously the answer to all our problems…what is the point that John is making? Or to put it another way – just what are all our clients doing with their 3D city models? Are these models simply nothing more than expensive computer games purchased to impress a Mayor?
Well – there is a real danger that this is exactly what they could become – the reasons for modelling need to be kept in mind - There is a tendency to be led by the visual aspects, which I believe interferes with the true value a model could offer: The main aim has to be to save money – be it by ‘real savings’ in the design process or long-term management of an asset (be that a city, a sub-way network or a single building), or through improved decision making. So what are they doing with the models? Obviously there is a range of applications – but I’ve broadly grouped them in to these 4 categories.
Here the model is being used as a record of the built environment. There is no better means of recording the structure and fabric of an urban area than 3D. It’s More intuitive than a map, certainly more interactive – and this means that it can be used as a baseline for other applications such as disaster management (fire brigade) These images show CityGRID being used live in an emergency command centre – this was an exercise carried out in the run up to the European Cup. Detailed information of buildings (above and below ground) were relayed to ground teams who otherwise would have been acting blind.
The volume of information that can be captured and retained of important buildings is immense. Often these culturally significant buildings have no structural records making the management of them harder to say the least. Here we see a model constructed from a terrestrial laser scan survey being used – amongst other things – for the design of replacement masonry.
Urban Planning is obviously a ‘catch-all’ description that covers many issues and disciplines, but I thought I’d show you an extreme example of CityGRID being used actively to assist the decision making process In this case – it is shadow analysis. Not for a ‘right to light’ as we might expect in the UK, but for animal welfare – in particular for assessing the amount of time spent in the shade by Vienna's famous horse-drawn taxi’s. The yellow indicates the taxi-stands!
Clearly an important element of a 3D model is its visual impression – to be an effective communication tool, especially for non-technical audiences – it needs to convey complex information without over simplifying or distracting with the noise of portraying every detail of the real world. How the audience interacts with the model is an important consideration for its design. In this instance we’re working in a live public consultation meeting for OBB. A series of road-shows toured the route of a new high-speed line to the affected communities. Members of the public were given detailed information via the model, but with engineers on hand to address questions and concerns. The aim of the model therefore was realism.
… .where we need to interact with the wider community, for example via information booths (this one based on the ESRI ArcExplorer technology for the city of Oslo), the model is simplified to represent the key messages the client needs to convey. Clutter is removed.
Or we can go wider still via the WWW – in this case Google Earth….. While a useful tool for reaching a mass audience it is not a preferred option for consultation It is useful however for Tourism – and……
this is a rather interesting example from the City of Linz and Ars Electronica who have used CityGRID to develop a 3D projected public information terminal!
And just to confound – many of our clients are using digital models simply to generate hard copy output.
City of Zagreb –To replace a wood model that had ‘burnt down’
So we can see that accuracy defines the models suitability to its proposed application But when we discuss 3D accuracy we entre a grey area. Positional accuracy becomes superseded by issues of representation. But I feel that for a model to have real value, it can not simply focus on visual characteristics, but must also be semantically correct – and the reason for this is that for all of the examples just listed, none of those models was developed with the single application in mind. They are derived works from a base product to meet specific requirements. This is where a structured data model becomes vital.
So Accuracy is a factor of a number of criteria. The primary indicator is the positional accuracy of the features – this is generally specified in plan, with Z (where recorded) generally being significantly coarser. This is familiar territory to most here, but for many involved in the commissioning of city models, it is apparently of little interest. There is an assumption that the data will be ‘accurate’, and perhaps with the current technologies on offer this is a reasonable expectation. It is not uncommon however for us to have to introduce ‘error’ into the model in order to conform to cadastral records – truth being an inconvenience to the law.
Geometry is where we start defining the form and appearance of the buildings – and the vagaries begin. Where a 2D map is understood to be an abstraction, there is an expectation that a 3D model is a carbon copy of the actual city. But there are two issues with this: Cost of capture Utility of the model – at what point does ‘enhancement’ become ‘noise’?
Textures are applied to the surface of the model to add visual clarity – but can also be used to enhance the model through, for example, the identification of windows which is very useful for noise modelling. They offer therefore a very fast and cost effective way of adding real information to a model, without the need to actually model to that level of detail
I have an example of that from Glasgow. Here the client insisted on the physical capture of all recesses over 10cm deep. An expensive and time consuming requirement, but one that added nothing of value to the model for its intended purpose. Instead, applying bump images to a simplified façade provides an equally engaging street view – and that was what this model was required for: contextual placement of proposed developments. A detailed massing model
In this example, we see an impressive model of Vas in Norway. Generated using the very smart C3 technology, this is actually a draped surface – created incredibly cheaply and quickly it is a very good tool for providing an overview of the city, but….
… once we zoom in – we see that the model does not make logical sense. Here the tree is actually incorporated into the fabric of the building. This would render it unsuitable for nearly any analytical purpose. So how do we describe it’s accuracy in a helpful way?
That is something that standards should help address – and in this particular area of interest that falls to CityGML
Released by the OGC, it is actually supported by the 3DSIG of which we sit on various technical committees. CityGML uses GML (Geographic Mark-up Language) for the geometric descriptions of buildings. The significant element to CityGML however is that it adds semantics to the geometric data . This includes a concept of Levels of Detail (LoD) for buildings;
So lets examine these. There are 5 classes from 0 (a terrain surface) to 4 (interors) The LoD is an indication of the precision that the building has been modelled to independent of any positional issues or correspondence with reality. It is also to the discretion of the modeller as to what LoD they apply particular features to – thus the level of granularity varies between models. There are sound reasons for a generic standard to adopt a flexible approach – but it does help confuse the wider audience!
But there does appear to be some emerging acceptance of these levels….. There is an added ambiguity however, through the use of sub levels. These are not in anyway official and show that in general ‘par-lance’ the terms are used flexibly. So 1.5 is being used to described buildings which have a generalised roof-model. This means that the contractor is applying a best-fit from a library. This is typical of (semi) automated processes - especially when LiDAR data is being used as the raw input. The result is significantly cheaper, but visually indistinguishable from a true LoD2 model (where the roof has been modelled from real data) but clearly has a low utility for certain applications (for example solar panel placement – as we have done for EDF) The cynic might conclude this was potentially deliberate to elevate the offerings of one technical delivery over another.
CityGML also introduces other concepts of importance: There are different classes of model components for describing buildings, terrain, vegetation, city furniture, roads, water bodies and so on, And there is a hierarchy of parts allowing for aggregation into objects Predictably however many CityGML models will follow the standard for geometry & appearance so that the data is displayed correctly in CityGML viewers, but will disregard the semantic elements . Visually these models are acceptable, but as tools for analysing and managing the urban infrastructure they are flawed. So here we see the water being classed as land – and it’s not simply a case of updating the attributes to resolve this.
So to add further to this mix – there is always the issue of competing standards – and there is a question mark over the potential uptake of CityGML – the US is a key market to crack, but there an alternative approach is gaining momentum in the form of INDUSTRY FOUNDATION CLASSES from the School of BIM. Perhaps there is some potential for convergence here – as has happened so successfully with GIS and CAD, So to that end we’re currently working with Northumbria University on an R&D proposal to examine the interchange between the standards (ref. magazine!!!)
So in summary – it is buyer beware out there in the 3D market place. If you thought that mash-ups were highlighting a depressing general lack of understanding of data quality – be prepared to be depressed all over again.
3D as a primary form of information display will happen – but there are some huge hurdles to over come, and I’ve only been looking at one issue here.
June 2010 CityGRID 2009 Intelligent 3D data management
LOD: Scalable solutions according need Application: Data Source: Topographic mapping Aerial photogrammetry or airborne laser scanning LoD 1.5 Generalised Roof Model LoD 1 Block Model LoD 2 Roof Model LoD 3 Detailed Roof Model LoD 4 Interiors Aerial photogrammetry or airborne laser scanning Detailed aerial photogrammetry Façade photographs and 3D TLS of complex buildings Construction plans or surveying Noise protection, large area city planning Local area city planning, project development Project development, architectural documentation, cultural heritage. Homeland security, asset management