Adobe has been moving quickly to update its imaging software to work natively on Apple's new in-house processors for Macs, starting with the M1-based MacBook Pro and MacBook Air released late last year. After shipping native versions of Lightroom and Camera Raw, it's now releasing an Apple Silicon-optimized version of Photoshop, which delivers big performance gain versus the Intel version running on Apple's Rosetta 2 software emulation layer.
How much better? Per internal testing, Adobe says that users should see improvements of up to 1.5x faster performance on a number of different features offered by Photoshop versus the same tasks being done on previous-generation Macs. That's just the start, however, since Adobe says it's going to continue to coax additional performance improvements out of the software on Apple Silicon in collaboration with Apple over time. Some features are also still missing from the M1-friendly addition, including the "Invite to Edit Cloud Documents" and "Preset Syncing" options, but those will be ported over in future iterations as well.
In addition to the Apple Silicon version of Photoshop, Adobe is also releasing a new Super Resolution feature in the Camera Raw plugin (to be released for Lightroom later) that ships with the software. This is an image-enlarging feature that uses machine learning trained on a massive image data set to blow up pictures to larger sizes while still preserving details. Adobe has previously offered a super resolution option that combined multiple exposures to boost resolution, but this works from a single photo.
It's the classic "Computer, enhance" sci-fi feature made real, and it builds on work that Photoshop previously did to introduce its "Enhance details" feature. If you're not a strict Adobe loyalist, you might also be familiar with Pixelmator Pro's "ML Super Resolution" feature, which works in much the same way -- albeit using a different ML model and training data set.
Adobe's Super Resolution in action
The bottom line is that Adobe's Super Resolution will output an image with twice the horizontal and twice the vertical resolution -- meaning in total, it has 4x the number of pixels. It'll do that while preserving detail and sharpness, which adds up to allowing you to make larger prints from images that previously wouldn't stand up to that kind of enlargement. It's also great for cropping in on photos in your collection to capture tighter shots of elements that previously would've been rendered blurry and disappointing as a result.
This feature benefits greatly from GPUs that are optimized for machine learning jobs, including CoreML and Windows ML. That means that Apple's M1 chip is a perfect fit, since it includes a dedicated ML processing region called the Neural Engine. Likewise, Nvidia's RTX series of GPUs and their TensorCores are well-suited to the task.
Adobe also released some major updates for Photoshop for iPad, including version history for its Cloud Documents non-local storage. You can also now store versions of Cloud Documents offline and edit them locally on your device.