- Automatically knows when to take a photo.
- Unclear how well AI will work.
- The Google Clips camera uses artificial intelligence to know exactly when to take the right picture or video clip.
What Is It?
The Clips is a small, square camera that looks a bit like a GoPro thanks to a protruding, fixed-focus lens. It’s made of hard plastic and comes in a rubbery case with a clip on the back. The camera itself isn’t water resistant or rugged, but the rubbery case hides the USB-C power port and makes it bouncy.
I must make this clear: This is not a GoPro. I said it before, but it’s worth saying again: This is not an action camera. It’s a new, experimental category.
The idea is to plant the Clips in your house, or at a party—still, not moving—and it will function as a sort of event photographer. It’s especially good for owners of small children or dogs, for whom every day is an event with cute things unexpectedly happening.
Camera Quality and AI
The camera has a 12-megapixel sensor and a 130-degree f/2.4 lens with a 3 feet to infinity fixed-focus design. You power it on by turning the ring around the lens, and there’s a shutter button below that. But ideally, you’d never use the shutter button. Here’s where things get weird and interesting.
The Clips works over Wi-Fi Direct with a special app that runs on Pixel phones, the Samsung Galaxy S7 and S8, and the iPhone 6 or later. In the app, you set the Clips to take few, some, or many photos. Then you turn it on, and you put it somewhere, and the battery at that point lasts about three hours.
We’ve had AI video editing software that figures out when things are interesting for more than a decade; the Clips puts it in a tiny device and combines it with facial recognition.
The camera will automatically record five- to eight-second, 15fps video clips (without audio) when it detects someone you know doing something interesting. It isn’t just motion activated; it’s using facial recognition, too. Set it up by manually taking pictures of your kids and pets, and it’ll recognize them and figure out when they’re up to cute shenanigans.
You can save the video clips or pull stills from them. I’m not a fan of 15fps video (it looks jerky), so I’d probably pull stills. The camera can store hundreds of clips in its 16GB of memory.
Google showed me some canned demos of these clips, and they looked great: kids making faces, goofing around, and doing gymnastics. Whether this will work in real life remains to be seen.
Because this is Google, my mind immediately goes to creepy places. I can’t quite shake the feeling that it’s identifying your children so Google can later use that knowledge against you somehow, but that’s utterly unfounded, right?
The company says all of the recognition goes on within the camera itself, and images don’t get uploaded to anything unless you say they should. On-device facial recognition is handled by an Intel processor originally designed for self-driving cars. I don’t know if it works, but it’s interesting.
We’ll find out more when we get a Clips in for testing. No release date has been announced.