Google has announced it is developing a new image enhancement tool that can tease more detail out of low-resolution images, similar to the computer trickery used on shows like CSI that we all laughed at for being impossible.
Google Brain revealed the new two-step process that sees Google’s algorithm first create a map of what a full-resolution version of the image might look like by mining similar images with higher resolutions. Then it simply adds details to the missing areas.
The results aren’t fantastic, as you can see in the example above, but it’s quite an extraordinary achievement. And, of course, this functionality will only continue to improve.
Google’s AI system can turn an unrecognisable image into something distinguishable, which could serve many uses, from aiding police investigations to photographers wishing to clean up fine details when blowing up their images.
Google also recently rolled out its RAISR technology for producing high-quality versions of low-res images.