Narrowly: the reproduction of Scotland’s built environment as a full scale, high quality 3D model for experiencing in virtual reality.
- analysis and study of Scotland’s built environment, culminating in the production of a diverse range of representations and reconstructions;
- educating, empowering, and enfranchising citizens and community groups in the autonomous application of these production methods, and encouraging autonomous experimentation in the research and development of further modes of production;
- incentivising the exploration and study of, and further innovative engagement with, Scotland’s built environment;
- facilitating citizens’ and community groups’ access to the digital representations and reconstructions for their use in individual and/or communal forms of self-expression and/or storytelling.
The explicit goal of the project is a full scale VR Scotland model, which, upon completion, will be a technical achievement of international significance. However, a vast quantity and range of assets will be generated en route to that goal. Those assets, and what will hopefully be the collaborative nature of their production, will form the greater part of the project.
The project’s methodology for reproducing Scotland’s built environment as a VR experience includes: Surveying (3D Scanning), Analysis, and 3D Modelling (through Computer Assisted Design, CAD). The subject is reproduced as a photogrammetry model, from which are derived schematics (town plans and architectural elevations).
The subject is then analysed, for buildings this includes an analysis of the presence of named architectural features. One of the overarching project goals is the production of a searchable database of historical Scottish architectural design, supplemented by a comprehensive collection of digital media (photos and models) of the buildings in question. By example: the main building of Dundee High School has Greek Doric columns, as does the Royal Scottish Academy in Edinburgh. The ground floor of the central offices of the Church of Scotland on Edinburgh’s George Street also has Greek Doric columns, but these latter are the Delos style (subcategory of Greek Doric) whereas the two aforementioned are Paestum. How many types of Greek Doric columns are there in Scotland, and where? The First Minister’s residence in Charlotte Square in Edinburgh, and more generally the ‘New Town’ style, includes rustication (quarried masonry, ashlar, with emphasised joints) on the ground floor of the type V-chamfered, both horizontal and vertical joints emphasised, and flat faced. Meanwhile in Dundee, both the Customs House and 3 High Street (RBS) have the same type of rustication but with only horizontal joints emphasised (which can be termed ‘banded’). How many types of rustication are there in Scotland, and where?
The analysis of the subject, coupled with the schematics derived from the photogrammetry model, then inform the manual (CAD) modelling of the VR environment.
Any virtual 3D model, irrespective of how it was produced (e.g. algorithmically, or CAD, or manually sculpted, or scanned), can be used as a locus for the generation of multiple derived assets. These may be:
- digital or physical (2D prints; 3D prints or moulds);
- 2D (any imaginable type of projection, including any simulation of a real world camera+lens) or 3D or used in a 4D production (animation/film or interactive experience);
- realistic or stylised;
- whole, parts, or any composites of wholes and/or parts.
Thus the production of a 3D scanned model opens the door to the production of further derived assets that can be used in any number of contexts (maps, prints, illustrations for use in diverse fiction or fact publications, virtual 3D models for use in animations or simulations or interactive VR and/or AR applications). The manual/CAD 3D model can be similarly utilised, whether the form of that model is realistic, or reproduced according to a specified art style, or even as historically accurate for a given period.
So, while the intentionally headline-grabbing goal of the project is a “full scale VR Scotland”, more pragmatically the aim would be to act as a catalyst in stimulating citizens and community groups across Scotland to engage in new forms of artistic self expression. Using nearby subjects – the built environment of the local town centre and associated buildings of civic significance – the project would encourage citizens to become autonomous in experimenting with photography, photogrammetry, and other forms of digital asset production, as media for facilitating their personal or communal artistic expression and storytelling.
I have no formal training in any of the production methods that I use. Everything has been learned from online and offline manuals and tutorials, and a very large amount of time spent mucking around with stuff trying to get it to work and give the desired results (or to use the formal title for that type of activity, “Research and Development”). Then, upon establishing a consistently replicable production method
In the early days (2013-2016) there were a lot of false starts, plenty of frustrations, and months-long periods when I did no work at all on what would eventually become this project. I am very proud, never ask for help, and often approach a problem with no prior knowledge on the off chance that that naivety will yield a new and advantageous solution that would otherwise be ignored were dogma to be respected. That is a sword that cuts both ways and you sometimes end up re-inventing the wheel, but I keep quiet about those times and file them away, alongside the big swathes of time spent on stuff that just does not work, under “learning experiences”.
In July 2013 I captured some reference photos of Horse Wynd in Dundee, then that evening downloaded the Unreal Development Kit game engine (UDK; Unreal Engine 3) and set about figuring out how one might reproduce that street as a virtual environment. A few months later I produced a very simple model of Dundee’s City Square using the UDK’s integrated modelling tools.
At the beginning of 2014 I turned my attention to my local town centre, Cupar, and produced a simple 3D model using Autodesk 3DS software which was imported and rendered in the UDK. However by June 2014 I made the conscious decision to switch my modelling software from Autodesk 3DS (free to students, otherwise costing ~£4K to the average citizen) across to Blender (Free and Open Source, can run on low spec public library computers) so that my production method would in principle be replicable by any citizen or community group irrespective of their financial situation. That next version of the virtual Cupar town centre – modelled in Blender, and now rendered in the Unreal Engine 4 which was released earlier that year – included buildings with a touch more detail, namely protrusions and recesses on the facades where the windows and doors should be.
Then I wondered “how might I add some accurate landscape to the background of the town model” so I spent ~250 hours developing a working method for generating 3D terrain models from public domain elevation data (including 100 hours of failure, and 50 hours of heightmap renders that had to be thrown out because the Color Space was set to sRGB when it should have been set to Raw). This yielded a reliable production method that I used e.g. a couple of years later in 2017 to generate a full scale bare terrain model of the Fife peninsula, 80km*50km, then recorded a video showing a player driving from St Andrews to Glenrothes in real time.
To inform the manual 3D modelling of a given building facade, in the early days the facade was captured in one photograph, then that photo was corrected for both lens and perspective distortion. The resulting image served as a reference ‘schematic’, imported to the 3D software and traced from, for cutting out the doorways and windows and adding window sills and other protrusions to the given facade. In mid 2016 I started experimenting with Hugin, a photo-stitching and panorama production software. One of its capacities was the production of ‘linear panoramas’, as illustrated in this album. Production duration was ~10 minutes per photo, incorporating both manual Control Point addition in Hugin, automated processing in Hugin, and manual compositing in GIMP. So e.g. a linear panorama of Princes Street building facades (920 metres of facades, produced at a resolution of 1px/1cm, thus 92,000 pixels wide) was a composite of ~250 photos and took ~35 hours to produce (full resolution ~250MB download):
More recently, after experimenting with photogrammetry software and its use in the production of ‘schematics’ (i.e. orthographic projections), as shown in this album,