Pixelmator Pro adds AI-powered background removal and advanced subject selection

Posted on |

Pixelmator Pro is adding advanced background removal and subject selection with today’s 2.3 update. Using machine learning and AI, Pixelmator users can separate the foreground from its background in one click.

The Remove Background command automatically detects the subject and erases the background. Version 2.3 also includes more fine-grained AI-enhanced selection options if you require more manual control.

To remove a background from an image in the new Pixelmator, you simply press the button and it figures out the rest. As well as using machine learning for precise selection cutouts, Pixelmator has a final stage called ‘Decontaminate Colors’ to help refine the colors at the edges of the objects. This step removes color fringing artifacts (traces of the previous background showing through) and helps integrate the superimposed layer much more realistically with a replaced background.

Of course, the background removal is accelerated using Apple’s Core ML framework which means it can take advantage of the Neural Engine found in Apple Silicon chips. A background removal on Intel machine could take about 30 seconds, whereas on a M1 computer it will be in the 2-5 second range.

Remove Background is also exposed through AppleScript and as a Shortcuts action.

If you want something more manual, you can ask Pixelmator to select the subject only and then refine the selection using the usual selection tools palette. They are also introducing a new Select and Mask tool to help make exact selections of traditionally-difficult areas like human hair or animal fur.

As always, Pixelmator Pro is available exclusively through the Mac App Store. Update for free to the latest version to try out the new features. For new buyers, the app is currently on sale — 50% off for a limited time.

FTC: We use income earning auto affiliate links. More.

Check out 9to5Mac on YouTube for more Apple news:

Leave a Reply

Your email address will not be published. Required fields are marked *