Through the Lens of @FFD8FFDB: Art by Security Cameras

If you hang around near unsecured security cameras, you might accidentally appear on @FFD8FFDB, an automated Twitter art project run by developer Derek Arnold. The bot is connected to a range of unsuspecting cameras across the U.S. and tweets a screenshot from a random one every 20 minutes.

On the surface, this doesn’t sound particularly appealing. In fact, one of Arnold’s goals was to get any response at all, even a disinterested reaction. The project isn’t supposed to be creepy or menacing โ€” which is often the aesthetic of a security camera. Instead, the images are framed as “beautiful, rather than filthy”, he writes in an article explaining why he chose to start the project.


The project’s name comes from the file signature FFD8 โ€” meaning a .jpg file โ€” and FFDB, which (as far as I can tell) signifies a quantization table inside the code of a .jpg.

The cameras are usually trained at static objects like transmission towers and complexes. Arnold theorizes they were set up for the company that owns that property to simply monitor that the objects are still there. Its strange appeal (mundane acts and places being committed to tape permanently)ย is similar to Google Maps Street View, but there’s something much more mystical about the feeling of being let into someone else’s security footage. Unlike Street View, no one from the outside was supposed to see this footage.


So not to confuse the project with one of voyeurism, Arnold connected FFD8FFDB only to cameras with business IPs. The bot crops out identifying information in the margins as to protect the business’ privacy, but sometimes leaves artifacts like branding, road names, and text that you could use to trace where the images are coming from. That said, some brands of cameras like Foscam ship with questionable preset security settings, making them popular sources of internet spying.

What sets @FFD8FFDB apart from sites like Insecam,ย which aggregates a list of insecure cameras worldwide? To me, the images don’t feel like a trashy, stolen glance into a scene I shouldn’t really be looking at. There’s something a bit Chatroulette-ish about connecting to a stranger’s webcam and watching live, but when stills are snapped at random from cameras that don’t belong to individuals, it forces the images into a different context.


Unlike a human, the bot repeatedly takes images from the same inactive cameras or shoots at night when you can’t see anything. Arnold says that fact is part of why FFD8FFDB is generative art; he hands over the curative control to a machine and lets the project take on a mind of its own.

If you browse through even a month’s worth of images (around 2000), you’ll notice repeat shots from cameras you’ve seen before, with slight differences. Like many great projects, you get more out of it the more you explore. The sequential images form a narrative of random landscapes morphing over time. An obsessive viewer could easily track the changing landscape of a scene over time, through summer into winter.

29th Decemberย 2016

26th June 2017

With its nonsensical captions and dreamlike qualities, @FFD8FFDB transforms some of the most mundane imagery imaginable into something deeply intriguing.

Go to @FFD8FFDB on Twitter to be a part of the intrigue, and make sure to follow @_secretcave for the latest updates, podcasts, and videos.

Leave a Reply

Your email address will not be published. Required fields are marked *