Google has revealed more details of its new RAISR method of image processing, which promises to recreate the details of high-resolution images in low-res versions to dramatically cut down on data usage.
Announced in November 2016, RAISR – or Rapid and Accurate Image Super Resolution – is the company’s new smart technology that aims to predict how a low-resolution image can be increased to high-resolution by filling in gaps between pixels.
Google says RAISR employs machine learning that scans pairs of high- and low-resolution images to develop filters that can recreate details in the low-res version that are similar to those in the original image.
Google has deployed the technology on its Google+ platform, and the company claims it’s now using RAISR to process more than a billion images per week.
Google claims RAISR is saving up to 75% bandwidth per image, and we expect to see it roll out more widely across Google’s many platforms later this year.
Take a look at the demo video above to see just how it works.
Via itechpost