You don't need portrait mode for Facebook's 3D photos anymore

Facebook is using convolutional neural nets to make (almost) any image a moving image.

Facebook just (kind of) equalized the playing field for 3D photos. In an announcement on Friday, the company stated that it would let anyone create 3D photos even if they don't have access to a portrait mode on their camera.

When Facebook initially introduced this feature in 2018, portrait mode was the only way to turn a 2D image into a 3D masterpiece. Slowly and steadily, things are changing.

How does it work? — Portrait modes are normally present in phones with dual-camera setups. But not everyone has those. With the help of convolutional neural networks, Facebook says it is now able to inspect and render the spatial resolution of 2D photos taken through your regular camera mode into a 3D format. Facebook says it trained neural networks to "estimate a distance from the camera for each pixel," which basically configures the photo's visual depth.

It sounds a little complicated but the company demonstrated how its neural networks operate in nifty little 3D renderings of puppy photos, statues, and the ever-intriguing Mona Lisa. Facebook also said that it would apply its AI research on videos, which means we might see new developments in its virtual and augmented reality plans.

There's a catch — Of course, the good news arrives with a tiny bit of bad news. The feature isn't available to just about everyone, as you might have thought. The company notes:

Anyone with an iPhone 7 or higher, or a recent midrange or better Android device, can now try these options in the Facebook app.

We'll take it. Half a loaf, as they say, is better than no bread.