When it comes to mobile photography, Google proved two important things. First, megapixel count isn’t the end all and be all of quality. The second is that software can sometimes achieve what sometimes requires additional or more advanced software. Since then, it has become en vogue among smartphone makers to use AI, machine learning, and algorithms to produce certain effects using smartphone cameras. Now a group of MIT Researchers has come up with an algorithm to fix yet another common complaint about smartphone cameras: distorted wide-angle photos.
Wide-angle lenses are becoming more and more popular on smartphones these days. They’re great for taking photos of sweeping landscapes, allowing you to take in as much of the scene as you can. But wide-angle lenses come with a cost that is made all apparent when people are part of the scene.
The lens often distorts objects and faces that are at the sides and corners, making them look squished and stretched. There are various methods to correct these ranging from manual editing to automated procedures. The latter often corrects faces but then distorts the rest of the photo instead.
Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory or CSAIL have developed an algorithm that prevents that from happening. Instead of applying the stereographic projection only to faces while using perspective projection to the background.
What makes the research algorithm quite ideal is that it’s fully automatic and interactive. In other words, it’s the type of software fix you can use on smartphones to immediately correct distorted wide-angle photos.