Check out the photo below. It looks normal, right? Well, it never really happened. It’s the result of Google’s new Add Me feature for Pixel 9 phones that allows you to add the photographer into a group photo with the help of augmented reality.
Would you have guessed this moment never actually happened?
When I first heard about this feature during the Made by Google Pixel launch event, I had some reservations. I was worried about the idea of capturing a moment that didn’t actually take place and how that could blur the line between real and fake content.
Read more: Google’s Pixel 9 Phones Have a Wild New AR Camera Feature
That being said, trying Add Me quelled my anxiety. The feature is far from perfect, so you can’t convincingly “fake” every possible scenario, which is reassuring. And I do appreciate its intended purpose of inclusivity.
But I still have some concerns about Google’s other photography features, like Best Take and Magic Editor, which also let you create a “fake” moment. Unlike Add Me, which is designed to include everyone in the photo, these features seem aimed towards editing out imperfections to get a flawless photo. This concerns me because posting these photos online without disclosing that they’ve been edited could contribute to unrealistic standards for how perfect our lives should look.
Here’s another photo created by Add Me.
How I discovered Add Me’s limits (and its potential)
The simplest way to explain how Add Me works is that it involves combining two photos. But there’s a lot more going on behind the scenes. You start by taking one photo of the group (minus the photographer, of course) and then a second photo where only the original photographer stands in the frame. To help you line up the second shot, the phone displays virtual representations of the rest of the group with augmented reality. Then, the software merges the two photos to create one picture with everyone.
Read more: How to Use Add Me on Pixel 9 Phones to Get Everyone in the Shot
I tried Add Me with my fellow CNET-ers, technology reporter Abrar Al-Heeti and social producer Faith Chihil. We wanted our final shot to make it look like Abrar and I had our arms around each other, so she started by taking a photo of Faith and I. We made sure to leave a space for Abrar to pose in later.
Then Abrar handed me the phone so that I could take the next photo. Faith and I left the frame, letting our augmented reality stand-ins do the work, while Abrar hopped in. I tried to position her so that it looked like she was in the picture from the get-go. But when Add Me merged the two photos, it didn’t layer them how we had expected, as you can see below. (Check out the video above to hear our giggles when we realized our vision didn’t come to life the way we thought it would.)
Not quite what we were going for.
We tried the arm-around-each-other photo a few times, but Add Me did not seem to want to make it happen. The second time, it put Abrar’s arm over me instead of around me. So we moved on from that concept and kept experimenting.
Not quite what we were going for either.
I’ve found that Add Me photos look the most realistic when there’s a little bit of space in between each subject, like there is in the photo below. This may have been the picture I started this article with, but it took us a few tries to get there in real life. I think it actually looks like we were all together at the same time here.
Alright, now we’re getting somewhere.
Towards the end of our experiment, we continued to test the bounds of Add Me by trying to get the same person in the same photo twice. And yes, it is possible. We got this epic photo of Faith and her alter-ego. Would the image below convince you she has a twin?
Faith and her alter-ego.
Those photos of Abrar and I trying to make it look like we had our arms around each other showed me the limits of this feature. And those limits are exactly what make me feel more comfortable with it. Add Me can’t create a convincing photo of a moment that didn’t actually happen every time… At least, not yet. And when it is successful, I will say that it feels nice to have everyone included in the shot.
Google’s history of pushing the bounds of reality in photography
Add Me isn’t coming out-of-the-blue. Google introduced Best Take with last year’s Pixel 8 series, which allows you to swap your favorite facial expression(s) from a sequence of pictures into another photo from the same batch. Here is a Best Take-generated picture of Abrar and I by our colleague, Patrick Holland. Would you have guessed that this moment never actually happened?
Abrar and I looking sassy (photo created by Google’s Best Take feature).
Magic Editor also debuted with last year’s Pixel 8 series. At launch, it allowed you to do things like remove someone from the background of your picture or move and resize your photo’s subject. With the Pixel 9 lineup, Google has introduced new ways you can use the Magic Editor. Reimagine, for example, lets you transform your photos by entering text prompts. My colleague, Andrew Lanxon, used it to change his pub selfie into a photo in front of a brick wall with a sunset.
Andrew’s original pub selfie vs. his Reimagine-generated selfie.
You can also now use the Magic Editor to expand and fill the edges of a photo beyond its original frame. My colleague, Sareena Dayaram, explored how Best Take and Magic Editor made the line between real and fake content fuzzy when they were first introduced. With Add Me, Google is offering yet another tool to improve your shot, even if the moment you’re capturing is artificial. But I feel more anxious about Best Take and Magic Editor right now. Let me explain.
Google’s AI camera features still make me nervous
In this photo, Magic Editor changed an overcast sky to a blue one.
Like I mentioned above, Add Me (in line with its name) adds a subject in order to promote inclusivity. Best Take and Magic Editor, on the other hand, seem more centered around eliminating or changing things to create the perfect photo, like an awkward face or an overcast sky. This sort of editing worries me more than adding someone in the spirit of inclusivity.
That’s because I fear that posting these images without disclosing that they’ve been edited could make it seem like everyone’s lives should be picture perfect all the time when, in reality, life is messy (and that’s totally normal). Plus, doesn’t part of the beauty of getting a really good shot lie in the fact that it doesn’t happen every time?
Like Add Me, Best Take and Magic Editor don’t always spit out flawless photos. In my colleague Andrew’s pub-turned-brick-wall photo, for example, little details, like the imperfect cutouts around his hair, give away the fact that this image was edited.
Notice how there’s a little bit of the pub left over in the Reimagine-generated photo?
But there have also been many instances when I’ve found it hard to tell that these tools were used to create a photo. For example, I don’t think there’s anything about this picture of Abrar and I that gives away the fact that it was created by Best Take, rather than captured the first time around.
This photo was brought to you by Google’s Best Take.
In a statement to CNET, a Google spokesperson said that AI image generation tools like Reimagine and Pixel Studio (which lets you create images based on prompts) are designed for creativity and that it has policies and Terms of Service related to the type of content that’s allowed. “At times, some prompts can challenge these tools’ guardrails and we remain committed to continually enhancing and refining the safeguards we have in place,” Google said.
And don’t get me wrong, I love playing with Best Take and Magic Editor. It’s fun to get creative and see what you can achieve with them. I’ve even posted one of these images on social media (along with a disclosure that it was edited, since Google only includes this in the metadata).
I had to share this Best Take photo by CNET’s Patrick Holland on the Gram, but I also had to let everyone know it was edited (disclosure highlighted).
There’s always a gap between how we represent ourselves online and our real selves (I’m certainly guilty of only featuring my best moments on my Instagram profile). But in my humble opinion, features like Add Me, Best Take and Magic Editor should be treated with caution because they could make that discrepancy even wider and, ultimately, help normalize unachievable standards. Plus, nefarious actors could use these tools in all sorts of ways to create images of fabricated moments, as The Verge recently pointed out.
Features like these aren’t going anywhere. So the debate around what constitutes a “real” photo in the age of AR and AI is one that will likely continue for years. But for now, it’s not a bad idea to continue to check in with ourselves about how comfortable we feel using these tools as they become available, if at all.
+ There are no comments
Add yours