boakes.org

For the last few weeks I’ve been testing out some geo-tagging software on my camera. The concept is fairly simple, whenever I take a photo, the built-in GPS works out where I am and records this metadata within the photo.

The camera is already set-up so that whenever it’s in range of my computer the photos are automatically uploaded, stored for posterity and (most importantly) available to be viewed and enjoyed. What’s more, the software we use for managing our ever increasing pile of digital memorabilia (Picasa) already includes a tool for adding geo-tags to photos after they’ve been taken. They can then be viewed within Google Earth or Google Maps: so almost unsurprisingly, when I got home after work the other day, the photos I’d taken appeared dotted around Google Earth without me having to do a thing.

Geo-tagging is a wonderful thing; but it’s not yet photography nirvana - there are several issues that still need to be tackled:

Battery

GPS units eat battery, so regular use of geo-tagging is not particularly useful if you’re not able to charge your camera several times per day. One strategy is to keep the GPS circuitry switched off until it’s needed, however, it can take several minutes to get a GPS fix, which is useless for photos that capture the moment. In these circumstances the software I’ve been using appears to retrospectively tag photos as soon as a fix is obtained (which may be several minutes and some distance later).

GPS Accuracy

GPS accuracy in urban areas tends to be haphazard because tall buildings with solid flat walls bounce the satellite signal around before it reaches the phone. Unlike automobile GPS units, it’s not feasible to use a roadmap to correct for such anomalies and predict where the unit might really be when signal is poor.

An approximate tag is still infinitely better than no tag at all, but it’s far from perfect. I took a few photos whilst walking between Waterloo and the City of London the other day, which serve as a good example - the pictures appear clustered around certain spots rather than being spread out along the riverside.

View Larger Map

Enriched GPS Detail

Recording latitude and longitude is a natural starting point, but there’s other information that would be similarly useful. Several of the photos in the example set were taken fourteen floors up a building, so altitude information would also be good. What would be really cool however (and I’ve said this before) would be if some kind of compass and attitude sensor could be combined so that the camera direction could also be captured with the photo.

What originally gave me this idea was wandering around the fields of Glastonbury festival after we’d lost our disposable camera (whilst crowd-surfing - there’s a lesson there). We were rather gutted that we’d lost our pictures and I found myself thinking that even though we’d lost our pictures, we’d certainly be in lots of other peoples photos. So to get a photographic record of our time at the festival we’d just have to solve the problem of (a) sharing the photos and (b) making the photo search tractable. If there were 50,000 cameras used during the weekend and each of took 20 photos, there would be one million potential images that we may have wandered through so the search would be impossible. Surely GPS could help! If we had a GPS trail log of where we’d been, and if every camera position and direction could be recorded, then the intersection of those two data sets would predict which photos we might feature in. Comparing GPS trails alone is not enough in heavily populated events, it’s necessary to know not just if you are near the camera at the time the photo is taken, but also if the camera is pointing at you. For example, person b, is nearer the camera in the diagram, but person a is in the image. It’s still a pipe-dream until someone comes up with a camera that can record pitch, roll and yaw against polar coordinates, but when that’s possible it’ll open up some really interesting new query possibilities!

There’s a happily-ever-after for this tale too. We returned to the pyramid stage after the crowds were gone and scoured the floor, eventually finding the camera trodden into the mud, just inside the security fence in front of the stage (we were able to identify it because we’d written our postcode on it), so we got out photos back!