scanning the family photo archives

I picked up a scanner a couple of months ago, and have been slowly scanning in old photos when I get a chance. A few batches in, and I’ve already done 451 photos. I’m viewing the activity as potentially rescuing family history from fading pieces of paper. I have no idea if JPEG files will still be readable in 100 years, but it’s worth a shot to try to preserve photos going back well over 100 years (the oldest photo is from before 1893).

Family archives

As I scan, I try to add as much metadata as I have – often it’s just a scrawl of a name on the back of the photo print. I’m struck by how metadata poor many of these photos are. I can’t even guess at what decade some of them were taken in. Some have a name or two provided, some a year, and some have a wealth of hand-written documentation. And the quality of the photos themselves – there are a handful that I’d describe as not bad. The vast majority are absolute crap, technically. Blurry, poorly exposed, and small – several of the older photos are only available as 1-square-inch prints – I’m guessing the cost of printing photos back in the 1920s-1940s made it prohibitive, but even when cranking up the resolution of the scans to 1200dpi, there’s just not a lot of image to work with.

Family archives 2

Some of the oldest photos, of my grandparents and my dad in his childhood, kind of blow me away. Especially, knowing that they’d likely be lost forever without being rescued. The only metadata available for many of the photos is inferred – either through association with other photos that have rudimentary data scribed on the back – or via automated tools like face recognition. It’s interesting to see names get pulled out of group photos, and to see photos of individual family members spanning over several decades.

Leonard norman photos

Contrast that with the almost 30,000 photos in my own family archive (with a few hundred more ready to be scanned in a couple of banker’s boxes in the basement). Most of those photos have at least some metadata, and everything in the last few years has buckets of automated data – location, time, etc… – assuming Aperture library and image files will be somehow readable in a few decades…