Google’s New Algorithm can Compress JPEG without Compromising on Quality

March 20, 2017, By Sanjeev Ramachandran

JPEG compression is a thing of high priority in the modern day web world. But despite the availability of multiple tools, JPEG compression have always created headache due to its ubiquity. One wouldn’t definitely want to have pixelated images loaded, and neither would they want to wait too much for an image to get loaded while browsing.

Hopefully, Google says that it has found the right solution for this conundrum. Called Guetzli, the new JPEG encoder from Google is capable of compressing the JPEG sizes by 35% without creating any distortion to the original picture. Or in other words, Guetzli will be able to create better quality images at original sizes of the JPEG.

Guetzli focuses on quantization which transforms disordered data those are hard to compress into ordered data that are comparatively easier to compress. The difficult part was retaining the sizes, and that was managed by Google by introducing a new psycho-visual model, which they call the Butteraugli.

The psychovisual model is similar to human visual processing system, and the exact working is a bit complex and deals with detailed color perception and visual masking. In short, this helps Guetzli to distinguish the colors and details to be kept and discarded. By doing so, the algorithm will be able to keep small details that are else wise obliterated as single blocks of color.

Google’s new algorithm will be mostly beneficial in reducing the sizes of images in web pages, though it can also be used to tweak those captured using your devices. The algorithm is open-sourced and is available for download to all. Simultaneously, the technology can also be used to improve the quality of JPEGs while keeping the file size unaltered.

© 2008-2012 - All rights reserved | Privacy Policy