Facebook uses AI technology to bring 3D photos to people who can only take photos on regular 2D phones,media reported. By training its AI system on existing 3D photos, the social networking company can figure out how to make 3D photos — even if the user’s camera can’t. Facebook announced the 3D photo project in 2018, but was originally only available on the iPhone and only for photos taken in portrait mode.
But it later expanded the feature to other dual-camera phones such as the Samsung Galaxy series. Now Facebook has used AI technology to create a depth map.
“Advances in this technology have made it easy for millions of people using a single-lens camera phone or tablet to use 3D photo technology for the first time,” Facebook researchers wrote in a blog post explaining 3D technology published Friday. “Another benefit is that people can take 3D photos with a front-facing camera.
Facebook points out that while the work won’t work on all smartphones, it should run on iPhone 7 or newer phones, or on the nearest mid-range or better Android device.
Facebook can now create 3D photos from many phones instead of a dual-camera mode like the new iPhone.
This feature shows that digital camera technology, known as computational photography, is becoming increasingly complex. The technology uses increasingly sophisticated processing techniques to do things like correct ingress photos, add portraits, beautify faces and even shoot stars on your phone at night.
Making 2D photos into 3D photos is a great feature for companies like Facebook and Google, which use AI to calculate depth so they can blur the background of a portrait. But 3D depth maps have other and more important uses. Tesla, for example, uses AI-driven depth judgments to drive its Autopilot and Full Self-Driving (fully autonomous driving) capabilities without the need for an expensive 3D laser scanner system called lidar. Facebook plans to improve 3D technology with new features, such as calculating the geometric orientation of the surface and applying it to augmented reality, and it hopes others can use their methods to tackle challenges such as creating robots that can navigate the real world.