If you haven't read the first part of our interview with Stuart Nixon then don't delay: do so now.
The rest of you will already know that he's the founder of NearMap, but also the person largely behind ER Mapper, and the inventor of the cool sounding HyperPod aerial camera system and HyperVision supercomputer processing technology. He's also someone we've wanted to hear from for ages, so let's get back into it:
NearMap is linked up to OpenStreetMap, what were your thoughts behind this pairing and how important do you think this kind of community owned and user generated mapping site is?
I've always been a very strong proponent of opening up spatial data. I remember flying back and forth between San Diego in the USA and Canberra in Australia a dozen times a while ago, while I was on the spatial steering group that convinced the Australian Federal Government to open up much of Australian government data to free use.
OpenStreetMap is a concept we very strongly support. NearMap explicitly includes rights to derive street map data from our PhotoMaps under the CC-BY-SA license, and we continue to take an active interest in helping OpenStreetMap grow.
Companies need to tread carefully with open community efforts like OpenStreetMap. It is important to help, but not be seen to try to take over or dictate to community efforts.
It's really nice for users to see a site with no advertising! But how does the OpenStreetMap creative commons connection work with the revenue model for licensing the data to government agencies etc. or is it just not a problem because you're adding so much to the content with NearMap's fantastic images?
Our support for OSM is intended to assist people wanting good street maps, rather than for government users. Government clients license our PhotoMaps. They don't typically use OpenStreetMap data, as government agencies have internal access to their own government street and cadastral data.
Because our core revenue is from licensing our PhotoMaps to government agencies, we are able to provide our map portal to people for free without lots of annoying ads (which I hate too).
Over time, you will see NearMap being used for media market applications, however even there we will continue to give people a nice map portal without lots of annoying in-your-face advertisements.
How do you get all those maps to align from month to month?
The industry standard for PhotoMaps without ground point control is a Root Mean Square (RMS) error of 200cm (2 meters). NearMap PhotoMaps are currently processed to an RMS error of 50cm (0.5 meters). This means that, on average, a point on the PhotoMap will be within 50cm of where it should be.
This is a remarkable achievement, as you can see over these tennis courts (use the TimeView slider to go back over months).
Please note that shadows shift between days, so don’t use shadow as a reference when comparing the two dates.
The precision of NearMap’s PhotoMaps is an interesting subject in itself. Most existing aerial cameras use pre-calibration of lens and other distortions to achieve what is known as "metric cameras". So long as the distortions don’t change (for example the sensor is often temperature controlled to ensure it does not change in size), the metric camera approach delivers good results.
As early as 1961, researchers such as D. Brown from the United States Air Force recognised the power of a more recent approach, known as "self-calibration". Here, camera and lens distortions are simultaneously computed during bundle adjustment (which is the process used to orient everything to produce a PhotoMap). For the technically minded, this PDF paper is quite a good summary on the subject.
Self-calibration provides better results, often improving precision by a factor of 10 or so compared to pre-calibrated metric cameas. However, this comes at a price, because many more observations are needed on photos, and processing is considerably more complex and time consuming - especially on large PhotoMap surveys involving hundreds of thousands of photos. NearMap uses full self-calibration to obtain the high precision we require. We developed our own bundle adjustment technology from the ground up, and draw on the multi-resolution capabilities of the HyperPod aerial camera system to make this possible.
As mentioned above, NearMap currently achieves 50cm RMS even in adverse conditions, without requiring any form of ground control data. Precision is important to ensure PhotoMaps align when swapping between dates in TimeView and for our future 3D capabilities. We expect to continue to improve precision, beyond the 20cm or so RMS that is normally achieved with ground control data, as our models show a theoretical capability of 7cm RMS accuracy should be possible from our HyperPods, without ground control points.
Going back to the cameras you use to capture the images – is it these or the way that the site is developed that you think is the key factor to NearMap's success in the future?
Right now, our technology gives us an edge. But it would be foolish to assume that any company will keep a unique technology edge, especially given the value that online maps offer to media companies.
Over time, you will see many more capabilities being added to our web portal, which will give people a clearer idea of where we are heading. Right now I'd say people are only seeing 10% or so of our planned portal capabilities.
NearMap's imagery is impressively clear – have you had any Google Street View type complaints from people accidentally photographed in compromising positions? Do you have any way of screening this sort of thing or is it just luck? Is there anything unusual you want that's been reported that you want to let us in on?
Aerial photos are a bit different to Street View. For a start, the resolution is lower, and secondly aerial images have been around for over 50 years. So there is a well established understanding of the capabilities and restrictions related to our types of PhotoMaps.
We have worked hard to ensure we address privacy concerns. We could easily produce higher resolution PhotoMaps, but one reason we don't at present to get a feel for people's comfort level for privacy versus convenience of online maps.
You've obviously had A LOT of experience at the point where technology and cartography converge – do you have any thoughts on what you think the future might hold and what you hope the future holds – without giving away any of your ideas of course!
In one word: 3D
The NearMap technology (cameras, processing, etc) was all designed with the ability to extract true 3D information from the raw photos. This opens up a whole slew of exiting applications and opportunities.
Stepping back from NearMap and looking at the global changes, I see two very interesting trends.
The first is community involvement. You may recall that in the early years Wikipedia was sneered at by "professional" encyclopedia users and creators. Yet this very rapidly turned into respect, and today Wikipedia is widely appreciated, notwithstanding the challenges it has during its ongoing evolution.
So Wikipedia, Facebook, photo sharing web sites, blogs, and now OpenStreetMap all are hints at an ever increasing community involvement, creation and management of knowledge. The fascinating thing is how quickly they become major parts of people's lives. Cynics might suggest that communities creating knowledge are simply doing what governments should be doing in the first place. Yet this understates the importance of these changes. Because this knowledge is created by and for communities, it leads future changes rather than follows. Another differentiator is that governments often restrict access to knowledge, by charging draconian pricing or licencing regimes that inhibit or even discourage wide spread use.
Once knowledge settles more firmly in the hands of the community, a natural progression is it becomes the source for policy and decision making, and communities by extension will play a more active part in setting these policies.
The second trend is the impact media companies are having in the spatial market. Historically governments created spatial information for their own needs, and generally begrudgingly or not all distributed the knowledge to wider circles of users.
Media companies, on the other hand, want to give something of value to people, to encourage those people to use their services, thereby opening up revenue opportunities for advertising.
There are many examples of media companies giving value away for free (or mostly free) to obtain a presence which can be used for advertising. Examples include newspapers giving away interesting articles on news, search engine companies offering free search capabilities, directory services, and now of course location based map services.
The sheer value of media markets can be startling to people outside the media world. For example, in Australia alone (with just 21 million people), the media market will be $30 billion this year. As this newspaper article notes, the internet media market in Australia is expected in 2014 to overtake free-to-air TV and newspapers to become Australia's biggest advertising medium at $3.8 billion.
The value of these sorts of markets means that media companies can afford to pour tremendous efforts into media applications, with online maps being a notable area that new media companies are busy expanding.
I'm please that NearMap is playing a part in this very exciting process. It will be fun to see how online maps change and grow over coming years.
Thanks so much for answering our questions about NearMap - it's certainly a site that the Nestoria team was very happy to find out about andish you every success with it.