I Tried Out the Updated Cleanup Feature in the Photos App on Apple Photos, and It Actually Works Pretty Well.
Along with the all-new Siri AI , Apple is introducing a number of new Apple Intelligence features, including three new AI-powered tools in the Photos app. We’ve seen this before: Apple previously released AI-powered image editing features like Clean Up, which didn’t always live up to expectations compared to similar tools from competitors like Google or Samsung. But this year, the situation is somewhat different: Apple’s latest models, including those that work off-device, improve existing features and introduce new tools. Overall, this seems like a step in the right direction.
Apple’s cleaning tool is much better.
Perhaps the most important update here is the new Cleanup tool. In iOS 26, the Cleanup tool used Apple’s built-in AI models to remove objects, but its performance was inconsistent. Cleanup did a decent job of basic tasks, but I found it unable to remove surrounding shadows and replace an object with something that looked like it belonged in the image.
Clean Up now uses a hybrid approach. For simple edits, such as removing a small object, it uses the on-device model, as in iOS 26. But for larger, more complex tasks (such as removing an obstacle around a face), the task is delegated to Apple’s powerful Foundation models, hosted on Apple’s own private cloud servers. These servers, according to Apple, are completely private and encrypted. Apple claims it doesn’t access your photos or use your data for training.
To find these new tools, tap the “Edit” photo, then select “Tools” at the bottom of the toolbar. Here, tap “Clean Up.” By default, the feature is in Automatic mode, which is a hybrid approach as described above. From here, you can also switch to “High Quality” mode to force Apple to use cloud-based models.
Then proceed as usual. Use your fingers to select the object or part of the image you want to remove. If you’re using the “Quick” feature (available on your device), the cleaning process will begin instantly. If you’re using the “High Quality” feature , you’ll need to tap “Clean” and wait for Apple models to do the job. In my experience, the waiting time can even stretch out to several minutes if you’re removing sharp, large objects.
After using this feature to process multiple images, here’s the best advice I can give: always use high quality. Fast mode is similar to last year’s feature, and while it removes an image, its replacement leaves much to be desired, as you can see in the example of the mismatched tabletop in the image above. Even if you’re removing an individual object from the table, high quality better reproduces the tabletop, as well as the shadows cast by other objects.
Improvements continue in facial recognition. A new feature in iOS 27 can use generative AI and your own photos to reconstruct hidden parts of your face. In my test (which you can see above), iOS 27’s Cleanup feature removed 99% of the image of my coffee mug (though the frame somehow remained). However, in iOS 26, the result is simply laughably bad: a mishmash of surrounding colors.
Expanding storage for viewing photos in iOS 27 works flawlessly.
The “Expand” feature, as the name suggests, enlarges your photos. Let’s say you have a photo that’s off-center or just looks unbalanced. Tap “Expand” in the “Tools” menu, then pinch and drag the image. The surrounding areas will blur, indicating the areas that iOS will fill in using Apple’s generative AI models. Since Apple uses cloud-based models, this may also take some time. Tap “Expand” and wait.
Overall, Apple’s generative AI for expanding images and filling in details is quite good—with some limitations. This isn’t surprising, as it’s trained on its own Gemini models , which are excellent at image processing. I tested this feature by expanding a frame outdoors and indoors. It accurately guessed my surroundings and even added an arm and leg that weren’t in the original shot. However, the image was completely overexposed, so while more objects fit into the frame, the sky is completely lost. In the photo of the coffee shop, the Photos app took the coffee bags that were sitting on the shelf and simply replicated them in the expanded image. It’s a pretty clever way to make the expanded image look more realistic.
Apple’s Reframe feature needs some work.
The Reframe tool lets you change the angle or perspective of a photo. Perhaps you regretted not moving your phone just a little to the right before taking a photo of your significant other, and while the moment is gone, the angle may still be there. With Reframe, you can move your phone around the image to change the perspective, just as you would adjust the angle when you originally took the photo. The app displays a preview of how the image will look (since this is just a preview, it will show some unnatural distortions, but this won’t be in the final result). Then tap Reframe and let Apple’s cloud-based models do their work. After a short while, the reframed image will be ready.
During my testing, I found this to be the most inconvenient tool. It gets the job done, but it struggles with faces. It’s best used for small angle adjustments, nothing more. For a stress test, I pushed the tilt angle as far as the Photos app would allow. The result was an image of a tilted face that looked more two-dimensional than three-dimensional (I’ll spare you the horror of my face with distorted eyes). However, as Apple works to improve its cloud models, this may improve.
As a reminder, all of these features are currently in beta testing. Apple may continue to improve the user experience in subsequent beta versions and with the official release of iOS 27 this fall.