If You Think Instagram Is Intimidating, Wait Until It’s In 3D

Facebook has rolled out limited 3D features, and a new app may push the concept even further

Eric Ravenscraft
OneZero
Published in
5 min readMar 6, 2020

--

Image: Facebook AI

IImagine sharing not just a photo from your next vacation, but an interactive model of the beach itself, one that other people can “walk” through and explore from every angle. That’s the future that Display.Land, a new Android and iOS app, wants to bring to your phone. And it’s one that social media giants like Facebook have been working toward for a while.

Facebook first dipped its toe into the 3D-sharing pond in late February, when it rolled out a feature that allows some users with select Samsung and Google phones to create 3D versions of their photos. It isn’t true 3D because the original photo doesn’t actually contain any depth information. To achieve a 3D-like effect, Facebook instead uses machine learning to estimate how far an object is from the camera and cuts the picture into “slices” that can move independently. It’s similar to how movies that are shot in 2D can be converted to 3D, only done automatically.

Facebook’s solution is a clever way to fake depth on images taken with a single camera sensor. But dual-camera phones — which are becoming more common — can improve the process. Algorithms can examine the small differences in how each of the two lenses captures a photo in order to generate depth data, similar to how human eyes work.

The result from either method is the kind of “3D” photo you might’ve seen on your Facebook feed, one that you can wiggle a bit with your cursor to distort in a way that makes it feel like you’re seeing it from a slightly different angle. It’s a funny novelty and it can make for some amusing memes, but it’s hardly a real 3D representation of an object or scene.

A more accurate re-creation of a scene would involve a process called photogrammetry, which involves taking a large number of photos of a physical object and converting data from those photos into a 3D model. With enough information — and it can take quite a lot of information — any object can be re-created almost perfectly. A variation of this method called videogrammetry does the same kind of analysis using frames of a video. Free software like Meshroom has allowed…

--

--

Eric Ravenscraft
OneZero

Eric Ravenscraft is a freelance writer from Atlanta covering tech, media, and geek culture for Medium, The New York Times, and more.