Just a few days ago, one of the "fathers of the Internet", Vint Cerf, warned about the risk of an upcoming Digital Dark Age, one in which a lot of our precious memories will be lost, or will become inaccessible, due to the obsolescence of digital formats and storage devices. This is something that is already happening: just think of all the documents stored on floppy disks, or of all the games the used to run on cartridges. While the disks themselves could be in good conditions, it would be difficult today to find a reader to browse them.
Besides advocating the creation of a "digital vellum", that would allow the recovery of out-of-date files, Cerf plainly suggested users to make a physical copy of their digital assets. “If there are pictures that you really really care about then creating a physical instance is probably a good idea. Print them out, literally," he reportedly said.
If only it were so simple! Thanks to the ease of digital reproduction, we now produce so much content that it's hard just to sift through it and find the things that are really worth saving and, in this case, printing. That's especially true for images: while in the past we valued every single picture, and thought hard before taking one, we now often adopt a 'shoot now, think later' approach, ending up with tons of worthless material in our hard disks.
That's why I find interesting, in perspective, the deep-learning, image recognition software that's being developed by Dutch startup Resnap, which has just closed $450.000 in seed funding.
For now, it works only with the pictures you have uploaded on Facebook or on Instagram. Your connect your account, choose which period of time you would like to cover (the easiest thing is to create a "yearbook", but you can also select specific dates), decide how many pages you want in your book, and there you go.
In less than one minute, the system chooses your 'best' pictures and creates a photo book that can be saved and shared online for free, or printed on demand. The startup of course doesn't want to reveal the secret sauce behind its deep learning algorithm, except for a few hints.
One factor is popularity: friends' likes and comments are taken into account, but it's not the only one. "We select the pictures that are most relevant for a user, based on what we know they like. So we look at what's in a photo. We also filter on quality, resolution an other factors," one of Resnap's founders, Thomas Beguin, tells me by email.
There are of course many other services that allow users to create photobooks, like Blurb, Montage and Mosaic, to name just a few, but it's in the speed, accuracy and ease of the pictures' selection process that Resnap seems to have and edge over most of the competition.
"In the future we will add other services, like the pictures on your computer this summer," Beguin says. The fun, it seems, it's just about to start.
