Google’s Pixel 9 Pro is close to AI overload, but it’s still a great phone

Beyond these features, quieter examples of AI continue to make the Pixel a great smartphone, from improving call quality and even answering suspicious calls for you, to being able to pull up web results for anything you see if you circle it.

Take photos

Pixels have long had excellent point-and-shoot cameras that often outperform other high-end smartphones, and this phone continues that trend. On the Pro models, the new 42MP front-facing camera takes excellent selfies in almost all lighting conditions, and can pull back to wide angle for group photos. On the rear camera, the combination of wide, ultra-wide, and telephoto delivers sharp results within 2 cm of the subject (in good light and with a steady hand) and at 10x zoom.

I think that, all things being equal, Google takes the most realistic photos compared to current Apple and Samsung phones. Colors aren’t oversaturated or overly warm, and artificial lighting tends to be properly accounted for. Portrait blur and panorama modes are certainly not foolproof, but they generally work well.

However, Google’s focus on building photography systems going forward is clearly on AI tools, and there are more of them in the Pixel 9 Pro than ever before. While they all work, the results always have an AI whiff, and applying AI edits to individual photos still doesn’t feel as comfortable as the usual cropping and filters.

The new feature, Zoom Boost, is more or less the magic trick TV detectives use on CCTV stills, except in this case it uses generative AI to guess how an object will look in more detail than the camera actually captured. This feature is automatically applied when you take a photo if you zoom in more than 15 times, but you can also edit any photo you take with the phone, zoom in and activate Zoom Boost to clean it up. It’s better than the blur or pixelation you’d expect on most phones, but photos will definitely end up with obvious AI artifacts, like strange skin textures, wrong colors, or thick ink lines.

Another new feature, Add Me, is designed to solve the common problem of not being in a group photo because you’re the one taking the photo. You activate the mode, take the photo, then hand the phone to a friend and head to the group photo location. Your friend will see your photo superimposed on the photo you took, and when they press the shutter, the system stitches the photos together. The feature has issues if it needs to place you on a plush couch or in a complex shadow, but it’s about as competent as a decent Photoshop.

And more AI

The AI-powered Magic Editor also has some new features, and you can use it to touch up even your oldest photos taken on other phones. Auto Frame will analyze the photo and give you some framing options based on different photography techniques. It might straighten, crop, zoom, or even move away from the subject. Obviously, many of these options require the AI ​​to create new context entirely, so results will vary.

loading

For example, I reframed a photo I took of my son from behind while he was riding a scooter. The system flipped the photo from portrait to landscape and moved the subject from the middle of the frame to the bottom third, which made the photo look much better at first glance. But it also meant that about half of the photo was completely fake, and if you look closely you can see melted car bodywork, wires that aren’t connected to anything, and what looks like a bunch of cement blocks with some high-visibility stuff stuck to them.

The “Reimagine” option in Magic Editor is more targeted, as you can highlight any photo element (just as if you were going to erase it with Magic Eraser) and enter a text prompt to transform it. This could mean asking for fireworks to be set off in the sky behind you, or turning an unwanted “No Smoking” sign into a band poster. As always with these things, some results are nearly indistinguishable from reality, while others are downright terrible.

For video, most of the AI ​​features are enabled when you turn on Video Boost, which sends your backup videos to Google servers for processing. Videos you shoot are available immediately, but the enhanced versions typically take a few hours to arrive. Boost removes blur and adds detail to videos zoomed in more than 5x, fixes colors, makes low-light shots clearer, and optionally increases the resolution to 8K so you can take decent still photos from your video. In my limited testing, it did a great job improving nighttime videos, just as Night Sight does for still photos. I’d be curious to try it out at a concert, though, to see how the AI ​​handles capturing details of performers at a distance and with a lot of motion.

Get news and reviews on tech, gadgets and games every Friday in our Tech Newsletter. Register here.