As regular readers know, we’ve been trialling QR codes and a little while back rolled them on a small selection of object labels in a Japanese fashion display.
I’ve been keep an eye on their usage and some of the continuing problems around lighting, shadows, and low-resolution mobile phone cameras like the current iPhone 3G. So far usage has been, as expected, low. Firstly, the target audience for the exhibition content has, not surprisingly, not been very tech-savvy. Secondly, the ‘carrot’ isn’t clear enough to cause the audience to respond to the call to action.
More critically, one thing we still haven’t quite gotten right is the image size and error correction.
Shortly after the last post we upped the error correction in the codes to 30% (meaning that up to about 30% of the image can be obscured and it still scans – although it is isn’t evenly spread). This alone wasn’t enough.
With the long URLs encoded in the codes plus the error correction the resulting QR codes were even more ‘dense’ and hard to scan with 2 megapixel cameras. We’ve now done another set of codes with our own version of TinyURLs that generate locally. This has reduced the encoded characters from nearly 70 to around 25 characters – thus a far less dense code.
Even so, 2 megapixel cameras have patchy results when obscured by lens flare or shadow so our current thinking is that in the future the codes may need to be as much as 50% bigger.