Ultra-wide-angle cameras are a fantastic addition to smartphones, allowing users to easily capture landscapes, skyscrapers, and a big group of people. The latter can be an issue at times though due to distortion warping faces near the edge of the snap, but a group of Google and MIT researchers have discovered a solution (h/t: PetaPixel).
The researchers crafted an algorithm that allows for distortion-free portraits when taking a snap with your smartphone’s ultra-wide shooter. This algorithm creates a “content-aware warping mesh” that goes into action when it detects faces in an ultra-wide snap. Check out the comparisons below.
The team said it achieved good results from photos with a field of view from 70 degrees to 120 degrees. Many ultra-wide smartphone cameras fall into this bracket (e.g. LG G8, Asus ROG Phone), but there are more than a few that have a slightly wider field of view (e.g. Huawei P30 series, Samsung Galaxy S10 family).
Furthermore, the researchers said the solution was fast enough to be used at “at an interactive rate” on a phone. The team says this algorithm takes roughly 920 milliseconds to get the job done on mobile, adding that the process is often finished before users are able to see the photos.
These researchers are Google employees, so it’s fair to assume that we could see this algorithm in Pixel phones. Google’s Pixel 3 smartphone, for example, features an ultra-wide selfie camera in addition to a standard selfie snapper. The upcoming Pixel 4 is also expected to offer at least two rear cameras for the first time in the Pixel family. It’s unclear whether the new device will offer an ultra-wide rear camera, but the algorithm would certainly help if this is the case.