Ever since its debut, Google’s Pixel series has always handled smartphone photography differently from other brands, particularly Android ones. While the likes of Samsung, Xiaomi and Huawei chased better and better hardware with more cameras, more megapixels, larger sensors, Google put out relatively modest hardware, and instead let its software handle the bulk of the work. Google’s view was that software was more important than hardware when creating a digital image.
I didn’t fully agree with Google’s take, as there’s only so much software tricks can fix, and ultimately, it’s not an either/or decision. Brands like Xiaomi, Huawei and Samsung gave us new hardware with software smarts to boot. Their software may not be quite as advanced as Google’s, but it’s not like they’re small companies that can’t afford to innovate.
So starting about a couple of years ago, Google put more effort into hardware, and the Pixel 6 and 7 series made a leap. With the new Pixel 8 series, which launched last month, Google is again bringing better hardware and software. And this time, there’s one component that is responsible for both.
A.I.-driven silicon
The Pixel 8 Pro is powered by the Tensor G3, a chip designed by Google specifically to handle the company’s on-device machine learning algorithms, including speech and image recognition models. Google says the Tensor G3 is also the first silicon to run the exact same text-to-speech models used by Google’s own data centers. This means the most powerful and intelligent software company on Earth is making some of their internal machine learning models available on device in a smartphone.
The Pixel 8 Pro does indeed feel very smart. If I have a news article open on Chrome, I can ask the phone to read the article aloud to me, in English or another language; or I can have the phone summarize the article. It takes about 30 seconds, and sometimes the summarized cliff notes miss some important details, but it works well enough most of the time.
I can record a conversation with another person, and then the Pixel 8 phones’ voice recorder app can transcribe the text and even identify the two speakers.
But the most impressive parts of the Pixel 8 phones’ new A.I. power is generative AI, which is the ability to create text and even image files from nothing. This is most notable in the new Magic Editor, which allows me to remove objects from a photo, and then the phone will use generative A.I. to create entirely new pixels in place of the erased item.
For example, in the collage below are two images: the original shot on the left, and the edited version on the right. I used Magic Editor to remove the distracting elements of the shot, like the traffic light, road block barrier and brown apartment building. The Pixel 8 Pro didn’t just remove them but also created new visuals in place, like additional trees.
Here’s a more impressive example of generative A.I. at work. I removed the action figure of Hong Kong actor Tony Leung from the foreground of the image, and then the Pixel 8’s generative A.I. had to use machine learning to guess what was behind the action figure’s head and create pixels to fill out the missing frame. You can see the image is not perfect—most notably you can see a hand floating on its own, but A.I. did accurately fill in a lot of missing visuals, including part of the horned individual’s face, and the outfits of both action figures. If you remove the floating hand, the image on the right looks convincingly real.
The Google Pixel 8 phones isn’t doing this all entirely on device, as Google is sending some images to its data center for processing, but the Tensor G3 is handling a lot of the initial processing. This editing process is also surprisingly fast, taking 10-20 seconds per image.
Some have argued that this is opening up a can of worms, that it is becoming too easy to manipulate images and create a fake reality. Google’s own Pixel 8 marketing materials show people doctoring photos to make their athletic feats look more impressive. I am sure professional photographers may not like this direction. As a writer whose work could be replaced by A.I. one day, I am also a bit concerned, but it is the future and I choose to embrace it.
Elsewhere, the Pixel 8 Pro is a fine phone, it’s got a 6.7-inch OLED screen that gets brighter than before. Its new main camera sensor is upgraded from the Pixel 7 Pro; it now takes in more light naturally. The zoom lens also got an improvement, so overall the Pixel 8 Pro cameras are very good, among the best in the industry.
And the phone overall looks great, keeping the design language established by Google. Battery life is above average, too, thanks to a large cell. Every aspect of what a smartphone should be, Google gets right this year. But ultimately, what makes the Pixel 8 phone appealing is that new A.I. powered chip.
The Pixel 8 starts at $700, and the larger Pixel 8 Pro with an extra zoom lens sells for $999. The phones are available in North America and select Asian markets like Japan, Singapore and Taiwan. Google’s vision and zigging where others zag seem to be paying off, as the Pixel is one of the faster growing Android phone series in the world.
Read the full article here