Apple's latest AI-powered image editing features in Photos deliver real results, particularly the upgraded "Clean Up" tool that removes unwanted objects from images. The feature works by identifying and erasing distracting elements like photobombers or background clutter without requiring manual selection.
The Clean Up tool performed well during testing. Users can tap to identify what to remove, and the AI fills in the background realistically. It beats manual editing for speed and produces cleaner results than earlier versions. Most edits completed in seconds without noticeable artifacts or obvious AI-generated patches.
Other AI tools in the update show mixed results. Portrait Mode enhancements work reliably for headshots but struggle with complex compositions or extreme angles. The Glow filter produces flattering results on some images while appearing artificial on others. Smart search improvements help locate photos by object or scene, and that feature performs consistently.
These tools arrive as Apple rolls out Apple Intelligence, its broader AI initiative across iPhones, iPads, and Macs. The company positions these features as privacy-preserving since much processing happens on-device rather than in the cloud.
For everyday users, the Clean Up tool justifies the upgrade alone. Anyone who regularly deletes photos due to unwanted background elements or has spent time learning complicated Photoshop techniques will notice the difference. The tool reduces friction from photography, letting people save shots they would otherwise discard.
The remaining features work well enough for casual editing but don't replace dedicated photo apps like Snapseed or Adobe Lightroom. Users invested in professional photo editing won't find these tools sufficient for serious work.
Apple's approach here makes sense. The Clean Up tool solves a genuine problem that affects ordinary photos from ordinary people. The other features feel more experimental but don't clutter the interface. Users can ignore what doesn't work and lean on what does.