Google Pixel Video Boost Review: Is Google's AI Fixing Pixel Video?

For almost as long as Google has been making genuinely good smartphones, there has been one thing the Pixel has been particularly good at: taking photos.

Google more or less made computational photography one of the defining features of the Pixel. HDR+ became synonymous with the Pixel look, followed by Night Sight and, more recently, Ultra HDR.

Even when the hardware wasn't necessarily the best in the business, Google's image processing could often make up the difference. But video has always felt like a different story.

That doesn't mean Pixel phones take bad videos. They don't. But compared with what Apple and Samsung have been doing with smartphone video, Google has often felt like it was paying a little less attention to the moving picture.

So when Google introduced Video Boost with the Pixel 8 Pro, I was curious.

The idea is that instead of asking the phone to do all the work itself, Video Boost sends a version of the footage to Google's servers, where its computational photography models can process it using cloud computing.

So, does Video Boost make Pixel video noticeably better, or does it just make an already good video a little cleaner after you've waited for Google's computers to have a go at it?

Blueroomcare Review: Is Its Therapy App Worth It?

The care was great. Getting to it was the issue.

TechloyLouis Eriakha

I put Video Boost to the test

There was really only one way to find out.

Rather than relying on Google's claims, I tested Video Boost in two very different lighting conditions. One was deliberately difficult, with very little light. The other had plenty of it.

The first test was the one I expected to tell me the most.

Test 1: Video Boost in the dark

For the first test, I went somewhere that was genuinely dark and shot three pieces of footage: a normal photo, a regular video, and then the same scene again with Video Boost enabled.

The photo, as seen below, gave me a useful point of comparison. If Google's processing could pull an impressive amount of information out of a still image in the same environment, I wanted to see how much of that advantage carried over to video.