Apple launched its ‘Clean Up‘ feature as part of the Apple Intelligence suite in iOS 18.1, offering users the ability to remove unwanted objects directly in the Photos app. The company positioned this tool as an intuitive, one-tap solution for erasing distractions from images, an answer to similar tools on rival platforms.
But the real-world experience tells a different story. Users report that Clean Up often leaves visible artifacts, handles backgrounds poorly, and misfires when dealing with prominent objects or linear features like power lines.
Don’t miss the best of The Mac Observer
Set us as a preferred source and our Apple reporting ranks higher in your Google Search results and Discover feed — one tap, no account changes.
In the above image example from Reddit, the results show warped lines and obvious patching where a large object was removed.
What’s frustrating even more is that Apple quietly dropped the “beta” label from Apple Intelligence in recent betas of iOS 26.1, suggesting the company already considers it a finished product. For most, that decision doesn’t match the results.
Why the gap between promise and performance
Apple’s approach emphasizes on-device processing and authenticity preservation. That means heavy edits, complex fills, or large-scale object removal pose technical limits. Meanwhile, competing tools from other manufacturers lean more on cloud-based generative models and allow extensive scene transformations. Clean Up is more conservative, designed for background tweaks, not full scene rebuilding.
In practice, Clean Up performs decently for minor distractions like small objects or isolated blemishes. But once you cross into foreground removal or structural edits like guardrails, cables, terrain, it struggles. This gap frustrates users who expected seamless AI-level editing rather than limited retouching.
What needs to change
Apple must improve the fill algorithm so it better handles complex geometry and maintains realistic textures. It should clarify the feature’s limits within the interface so users don’t over-expect. Greater transparency about version compatibility, model limitations, and expected output quality will help. Finally, a clearer roadmap for feature improvements would restore user confidence.
Right now, the image above shows what happens when the tool is pressed beyond its comfort zone. Users removed the subject, but the scene itself fractures, power lines bend, guardrails fade, and terrain patches don’t match. If Apple wants Clean Up to be more than a gimmick, it needs to bridge the divide between edit-tool promise and edit-tool reality.
Discussion