When I started in the GIS game back in the late eighties, we were one of the first private consultancy firms doing GIS in Queensland. If you hit the Turbo button on the front of the PC, it went from 6 to 8 MHz.
Back then, 3D meant putting a drop shadow on your headings in a Word document. 3D geographic simulation and visualisation wasn't even on the drawing board, at least as far as private sector consultancies were concerned.
A few years later, when I started creating these amazing animated and interactive scenes on a high-end graphics supercomputer, it was the realm of fantasy and, not surprisingly, big budgets.
When we gave demonstrations of what we could do, it was awe-inspiring to the audience. It would get 'Oohhs' and 'Aahhs' a-plenty, but there was rarely an open chequebook at the end because, frankly, nobody knew how to put a dollar value on such a thing.
At best, interactive 3D geographic visualisation was considered a cool add-on to other more important things, such as digital elevation models or orthorectified imagery. These, however, were apparently worth paying for. Defence has been in the simulation and visualisation game for decades. The civilian community are only just now embracing it.
Why is that?
Back in the early 1990s, if you wanted to do a visualisation demo, you had to back up the Silicon Graphics truck and wheel in a half million dollar piece of kit on which you could wield your magic. By the end of the millennium, the hardware price had dropped to about $40,000. The computer could be carried by one person (the monitor by another).
Interestingly, Silicon Graphics had by then started to see some significant competition in their visualisation space from the more upmarket PC vendors. It was around the turn of the century that some serious stuff happened. Computer prices plummeted; computer graphics became a commodity. It was now well entrenched in the Win/tel camp. 3D engines for video games hit a crescendo, GIS matured; 3D data became more freely available.
It was then that public perception started to change. 3D visualisation was no longer part of science fiction. It no longer needed deep pockets. It was no longer the sole preserve of the defence community. It was affordable and we began to hear those magic words: 'Hey, I could use that!'
With the advent of mobile computing and graphics cards such as the nVidia G-Force 2Go, you could even do a reasonable visualisation on a laptop. Demos got a lot easier.
Even so, it took a while for the message to sink in. I started VizMAP in 2000 with the idea that geospatial visualisation was a goer. After developing a business plan, VizMAP secured a few contracts in those first months from the early adopters (justifying our initial enthusiasm). Then: nothing. We made a living doing international aid work, with the occasional vis/sim and digital mapping job to justify our existence.
It wasn't until 2004 that we started getting jobs from people looking at our website, from word of mouth and from our purpose-specific marketing. Things started to change. What had happened?
The major contributor to the change was probably the video gaming community. Gamers pushed PC-based visualisation into the value-added realm. They would never have had this impact, however, if it had not been for the ongoing convergence of the geographic sciences (particularly GIS, remote sensing and photogrammetry). The logarithmic hardware/software cost/performance equation also came into play, which has put the industry where it is today.
The following timeline should not be taken as gospel, but is based on personal experience...
In 1995, the main role of geospatial visualisation was for high-end flight simulators in the military, and then, it was often geo-typical, not geo-specific (and more often than not was only wireframe visualisation).
In 2000, there was a deepening awareness of vis/sim in the civilian community, but they didn't want to 'buy it';
In 2002, state governments became aware that that here was an important public consultation tool;
In 2004, the civilian community embraced vis/sim as a useful tool for community consultation;
In 2005, we saw the first reasonable dollar values being put in contracts for 3D geospatial visualisation;
In 2007, it is now commonplace. The dollar value of a vis/ sim database is now able to be determined; and ….
By 2010? Speculation suggests that vis/sim may well be legislated as a requirement for the justification for proposed engineering, infrastructure, mining and exploration projects
At this point we need to make a distinction between simulation and visualisation.
The 3D visualisation of geographic data doesn't necessarily constitute a simulation, although the dividing line is quite blurred. Typically, a simulation will involve the end user in some activity other than simply observing the visualisation. Such activities typically include flight, train and car simulation for training and recreational purposes.
In the Defence arena we can also include weapons and tactical training such as Tactical Exercises Without Troops (TEWTs). In fact, training for Defence's Command, Control, Communications, Computers, Intelligence, Surveillance and Reconnaissance genre (C4ISR), is simulation-based.
This considered, we should extend the civilian-based geospecific simulation community to include anybody using the 3D visualisation of geographic data that is able to make decisions based on this visualisation.
A good example is that some of the larger local authorities are now performing interactive 3D sun-shadow analyses on proposed high-rise buildings to help determine approvals (see separate story in this issue on urban modelling).
There are five data components that contribute to a vis/sim database: terrain (typically a DEM/DTM); some form of nadirlooking imagery (usually orthorectified - aerial or satellite); 3D modelling of culture, (superstructure, substructure, vegetation, etc.); textures that can be draped over the 3D modelling (usually digital photography); and GIS data.
The data may or may not need the textures to be seen. Some GIS data may be used to simply enhance the DEM or allow interrogation points to access the raw GIS.
Of those five points, none are critical, but one or more (and usually five) are required. In my experience, a DEM is a good starting point.
We are still largely dependent on third party vendor formats for this data. Geoscience Australia distributes its Landsat 7 Mosaic of Australia in ERMapper's ECW format. Its nine second DEM of Australia is in ESRI's GridASCII format.
Nine seconds of arc equates to about 250 metres on the ground, so the nine second DEM has recently lost favour in comparison with the SRTM (Shuttle Radar Topographic Mission) DEM of almost the entire planet. This is good to about 90 metres on the ground. Interestingly, this is available for free download from NASA in the format of ER Mapper competitor, Mr.SID (although Mr.SID is not free when you want to view that data).
In relation to GIS formats, it has been my experience that most vis/sim database creation tools will read GIS data from some, if not all of the top CAD/GIS packages.
ESRI's Shape file format is a front-runner for customer supplied data. AutoCAD's DXF and MicroStation's DGN formats are riding high, both in the availability of customer and publicly accessible data.
DXF is the de facto standard for engineering agencies that may not have a GIS facility, but have a CAD department. They might not necessarily run AutoCAD, but you can bet that whatever they DO use will output to DXF.
The two remaining items in the list required for a vis/sim database are textures and 3D modelling. These are both very much third party software driven.
The texture data format is trivial in that JPEG, GIF, TIFF, etc are the preferred options, whereas the 3D modelling data format almost exclusively falls into one camp: 3D Studio Max's 3DS and AutoCAD's DXF - both from Autodesk.
Output of a 3D vis/sim database requires software/system/ format that can render it in real time, at 30+ frames per second. Until recently, there was only one option and that was OpenFlight.
OpenFlight was a format developed by Multigen in the heady Silicon Graphics days of the 80s and 90s to cater to the military flight simulation market. It is still very much in development and refinement today. Almost everything in the way of tools to create vis/sim databases will have some facility for outputting to OpenFlight.
Of the rest, every vis/sim toolbox vendor has their own 'proprietary' format, and most are very good at using their own vendor-based software tool to view the eventual database.
Many of these tools are free, so users can display their 3D databases without the constraints and overheads of an expensive software licence. This has been one of the hindrances associated with the exposure of the visualisation aspects of some of the grander GIS and remote sensing packages.
Another has been their inability to render, in real time, huge amounts of visual data. This is a fundamental aspect of what vis/sim tools do - horses for courses?
One underestimated component of the 'real' vis/sim community is the gaming fraternity, not the least important being Microsoft's Flight Simulator. With its global coverage of accurately rendered airports and the inclusion of centrelines of roads, railways, power lines, streams, and shorelines, the outline of parks, golf courses and cleared areas used for airports, the $100 package is a wealth of information.
The price also includes raster data consisting of elevation grids, aerial imagery, land and water classifications, and seasonal data. This, coupled with the fact that when it is used in its design environment (Flight Simulation) the experience is, indeed, very real.
In Australia, GeoSim Technologies has been able to harness the complexity of Microsoft Flight Simulator to include it in a commercially available helicopter, and in fixed-wing simulators. The simulators have received CASA accreditation. They can be used in a training environment, such that hours in the simulator actually count as hours in an aircraft.
Here in 2007, a number of new players have also made their presence felt in the software side of the 3D vis/sim marketplace. Not the least in impact is Google with its Google Earth product. It is noteworthy that Google doesn't provide the tools to create a vis/sim database in the strict sense, but it does allow the user to roam around the planet and investigate other users' input. What Google have very successfully done is to incorporate third party input, including (now) rendered buildings and geographic imagery overlays, not to mention the myriad of user-supplied spot information on various points of interest.
This is not 3D vis/sim as such, but is proving invaluable as an educational tool for the general public. It shows what is possible.
Graeme Brooke is the managing director of VizMap Ltd in Brisbane. He was previously the technical director of Virtuozo, the first practical low cost visualisation product available in Australia.
