Google Is Stretching the Boundaries of AI and Photography

[This newsletter is 100% written by me, not AI.]

Hello! Thanks for reading The Smarter Image. I appreciate your attention. If you want to help support this newsletter, please recommend it to a friend. Thank you!

First Thing

If you own my book Adobe Lightroom: A Complete Course and Compendium of Features, I’ve created an addendum that covers features added to Lightroom and Lightroom Classic since the book was printed. The addendum is available as a free downloadable PDF from my publisher Rocky Nook. It covers Generative Remove, Point Color, HDR editing, and more. It’s also a helpful reference even if you don’t own my book. (But if you do want to buy this extensively researched and beautifully printed book, use code CARLSON40 at checkout when ordering from Rocky Nook to get 40% off!)

Add Me (to the List of People Interested in the Pixel 9 Camera)

One of the most interesting modern technological arms races has been the one between Apple and Google over phone cameras. Technical advances such as what processor is powering a smartphone aren’t nearly as interesting to most people as what the camera can produce, because we’re all carrying cameras and all making photos with them. Each company takes an advantage one year, in low-light quality or portrait mode or something else, and the other catches up the next year.

(I’m generalizing by mentioning Google and its Pixel phones as stand-ins for Android. Samsung and Xiaomi are also pushing the state of the art of mobile photography, but they also run Android.)

To stand out, Google has been more aggressive about including AI-based capture and editing features on their phones. For example, one feature coming natively to iPhones is the Apple Intelligence-powered ability to remove objects or people from a photo, with the area filled in using generative AI. Google has had that ability for a couple of years now.

However, in my experience the quality of the results on Android has been subpar. (The iPhone feature is not yet available, even in the iOS betas.) So it looks very cool to circle a person in the background and erase them, but the pixels left behind are usually a mess.

But it’s important that Google is pushing these kinds of boundaries. They’re introducing new ideas and new ways to apply machine learning in useful ways.

At the Made by Google event on August 13, one camera feature that raised eyebrows was Add Me, which solves a particular problem. When you want to take a selfie with friends, your options are to hope someone has long arms and you all scrunch together to get in frame, or you find someone nearby who looks trustworthy and hand the phone to them to take a picture of your group.

This happens so often that bystanders are often quick to offer to take the picture when it’s clear someone is trying to capture their group. But you’re also handing over your expensive communication slab that holds most of your most sensitive data.

Add Me lets you take a picture of the other people in your group from a normal distance away. You make sure to compose the shot with enough room to include you. Then get one of your companions to swap places with you, and take your place in the shot. The phone keeps the subjects of the first shot on screen—by this point, phones can quickly and fairly accurately determine the subject. Your friend lines up the guide and you in the next shot, and then the phone creates a composite that includes all of you.

How to take Add Me photos (source: Google).

It involves quite a lot of effort and footwork, honestly, and guarantees that Add Me photos are going to look staged; there’s not much room for impromptu reactions when you’re doing a handoff dance behind the camera. But that’s OK, because you were going to create a staged shot anyway.

I cynically suspect that Add Me will be a neat technology that doesn’t get used much, because it will be easier to just ask someone nearby to take the photo (and hope they don’t take the phone instead).

But it’s still a novel idea that tries to solve a problem in a way that only software can do.

Let’s Connect

Thanks again for reading and recommending The Smarter Image to others who would be interested. Send any questions, tips, or suggestions for what you’d like to see covered at jeff@jeffcarlson.com. Are these emails too long? Too short? Let me know.

[Apologies for the long gap between newsletters. A lot of other work has been keeping me busy. I’ve also been reluctant to keep supporting Substack, but it’s been easier so far to stay there than to transition to another service.]