What you need to know
- Here’s our first look at Google Photos’ new Colorize feature in action.
- Colorize will use AI to convert your black and white images to color.
- The feature is not yet available and there is no word on when the beta could be released publicly.
Colorize is a feature for Google Photos that will use AI to convert black and white images to color. The feature was first teased way back at Google I/O 2018, and since then, Google has been mostly silent on the subject.
That was until last May, when the project lead David Lieb tweeted out a teaser photo along with the news that the team hoped to have a beta coming “soon.” It might have been a few months since then, but it appears the beta might be nearing a release. That’s after the folks over at 9to5Google were able to enable Colorize in version 4.26 of Google Photos.
The screenshots shared by 9to5Google show that when Colorize goes live, it will be able to determine automatically when an image is black and white and show up as a filter when you enter the photo editing menu. The feature clearly shows it is in beta using the “BETA” tag on the filter, as well as popping up a message when you first use it.
While testing out the beta feature, 9to5Google notes that images must be first uploaded to Google Photos before it can colorize them and the process only takes a few seconds.
The fact that you must first upload the image suggests that Google’s servers are doing the heavy lifting in colorizing your image, instead of it happening on your phone locally. This should ensure that the feature works quickly and seamlessly across all phones no matter how powerful they are.
The photos used here were first converted to B&W using Snapseed and then passed through Colorize with Google Photos. That makes it easy to compare just how well the Colorize filter works.
As you can see, the resulting images are far from perfect. The colors are muted, less vibrant, and some areas lack any color at all. Perhaps we can see why this feature has been delayed for over two years now, as training AI to accurately interpret color in a black and white image is no easy task.