'There's a category of machine learning models in imaging that are known as generative networks,' Reynolds explained. If you haven't, those sites include the likes of Dall-E and Midjourney – and beware, they're a hugely dangerous weekend time-sink. When asked about Photo Unblur and Face Unblur, Reynolds said: 'I think you've probably seen in the news all these websites now where you can type in a sentence and it produces an image, right?'. The complex process of building noise maps and applying 'polyblur' techniques was explained recently by Isaac Reynolds, Google's Product Manager for the Pixel camera, in a chat on the company's new 'Made by Google' podcast (opens in new tab).

Another related feature, Face Unblur, is so demanding that it can only run on Google's custom Tensor processor.īoth modes do way more than crank up a sharpening slider. Photo Unblur and Magic Eraser may look simple, but they're built on some powerful machine learning. But the big change is that Google's tools are both automatic and live in the places where non-photographers shoot and store their photos.įor the average user of Google's phones or its cloud photo apps, that'll likely mean never having to dabble with a program like Photoshop again. Adobe also has enough of its own machine-learning wizardry to ensure its apps will remain essential among professionals for many years to come. Collectively, these modes aren't yet as revolutionary as the original Photoshop.

Adobe photoshop elements for macs