Adobe has announced a new feature for its suite of photo-editing apps called Enhance Details that uses machine learning to artificially improve the resolution and picture quality of zoomed-in images. The results aren’t as dramatic as the “zoom and enhance” tropes of TV and film, but they do show how AI can help fill in details in digital pictures.
The new tool will be available in Adobe Camera Raw, Lightroom Classic CC, and Lightroom CC for Mac and Windows, and Adobe claims it can increase the resolution of certain images by up to 30 percent. However, the actual improvements are not as striking as that figure might make you think. You can see what I mean in the comparison GIF below.
Enhance Details works by introducing machine learning to a process called “demosaicing.” This is an algorithmic step that happens during the creation of a digital image, in which the camera adds color to the initial monochrome file. There is no single way to handle this process, and different companies and camera makers have their own special approaches.
With the new Enhance Details feature, Adobe is adding AI to the mix. The company used deep learning to train a demosaicing algorithm on a huge dataset of more than 1 billion sample images. From this data, the program learned the best way to color images, leading to a slight improvement in picture quality.
Adobe isn’t the only company to use AI in this way — many independent services offer similar machine learning upscaling tools — but it’s still significant that Adobe is building this service into its hugely popular products. It’s not the first AI tool the firm has experimented with, and it’s clear that Adobe thinks machine learning will be a big part of its future.
Anyone excited to test out Enhance Details will need to have a fairly up-to-date computer (macOS 10.13 or Windows 10 version 1809 or later) and will have to use RAW image format. You can find more details about the feature on Adobe’s blog.