When I first learned about Reimagine, the new feature on the Google Pixel 9 Pro XL and Google Pixel 9, I initially thought it was just another AI photo editor. However, after trying it out, I realized it’s the most powerful editor I’ve ever seen.
The functionality is incredibly impressive, to the point that it can produce images that are virtually identical to real ones. It’s so realistic that, at times, it made me feel uneasy.
How Reimagine Works
Reimagine is integrated into Google Pixel’s Magic Editor, so you need to have the photos you want to edit backed up in Google Photos. To activate it, you simply select any part of the photo. And when I say any, I mean any.
Not only can you replace skies or objects, but you can also create whatever you want by selecting any part of the photo by using prompts.
Generated using Reimagine.
This is a picture I took on a route heading to the coast of Granada, Spain. I took it with a Google Pixel 6, and it still maintains excellent quality. However, not everything you see in the picture is real. First, it wasn’t really cloudy. It was a gorgeous day.
Moreover, there were no hooligans drawing graffiti, and there were certainly no tires littering the ground. With this, I want to tell you two things: Reimagine works on any photo you have in your gallery, and its potential is practically unlimited.
Generated using Reimagine.
It’s simply fascinating how, even without the depth of field data from other phones (this photo was taken with an iPhone 14 Pro Max), it’s able to perfectly isolate the subject, human or otherwise.
This feature understands all kinds of instructions, such as adding a slight depth of field. This is especially important so that the photos don’t look like you’ve just copied and pasted them. Want to see the original photograph? It’s a little less striking…
In fact, if it weren’t for the text in the background, which AI is still having trouble generating in photos overall, I would’ve bet all I have that the AI-generated photograph in the forest was the original, not this one.
Generated using Reimagine.
The examples are as amusing as they are endless. My dog, Nexus, looks very cute in the first photo, except that his name is neither Nexus nor is he a dog. It doesn’t exist. I asked the AI to generate a white dog lying on the ground.
In the second picture, something similar occurs. Although it looks like a delicious plate of spaghetti, the spaghetti wasn’t actually there. The photo is just a comparison test of the white balance of an identical photo. The proportions of the dish have been maintained, the bread has been removed, and a new dish has been created. The prompt I used was simple: “spaghetti.”
In this case, the image on the right was generated using Reimagine.
If you want to test the limits of the AI, no one is stopping you. For instance, in the example above, I replaced the panther with a lion and changed the dull gray museum background to a sunset on the African savannah. With a little editing, the second photo could easily blend in as a real one.
The main drawback of Reimagine is that it’s part of the paid Gemini Advanced suite of tools. Luckily, the Pixel 9 includes a free year of the subscription. After that, you’ll have to pay for the service.
Overall, Reimagine is evidence of what generative AI can achieve with any photograph in your collection.
Image | Xataka
View 0 comments