Google's new algorithm can compress JPEGs 35% more than now available methods

Google's new algorithm can compress JPEGs 35% more than now available methods


The researchers said Guetzli targets one aspect of JPEG compression, quantisation - dropping the number of colours in an image to cut its size. The name Guetzli means cookie in Swiss German. Left: The uncompressed original.

Secondly, Guetzli would seem set to have a significant resource footprint compared to the long-established image compression libraries now in use (mostly over Apache or other Linux-based frameworks). Google said the breakthrough is the result of an algorithm that balances quality reductions and file size by searching for differences between JPEG and Guetzli representations of image data.

Google has developed a new compression algorithm for JPEG images that shrinks them 35% smaller than today's methods without image loss.

'[While] Guetzli produces smaller image file sizes without sacrificing quality, we additionally found that in experiments where compressed image file sizes are kept constant that human raters consistently preferred the images Guetzli produced over libjpeg images, even when the libjpeg files were the same size or even slightly larger. Its psychovisual model "approximates color perception and visual masking in a more thorough and detailed way than what is achievable by simpler color transforms and the discrete cosine transform", and the search algorithm that Guetzli uses to achieve this is slower than alternatives like libjpeg.

Obsessed fan has been 'stalking' Taylor Swift for months
Jaffar has now been charged with burglary and stalking, and because of this is also banned from getting in contact with Swift. In court papers, Swift also states that Jaffar returned two other times, ringing her buzzer for up to an hour straight.

The underlying compression algorithms behind popular implementations of JPEG have always been lossy; once a web version has been generated from a high-quality original master, the lost pixel information can not be recovered (except, arguably by AI - another field of interest for Google's image researchers). According to Google, this makes the slower compression a "worthy tradeoff".

Google is offering the Guetzli encoder as an open source software project for anyone to check out and possibly improve. For example, a multi-megabyte image stored in BMP or PNG format can look nearly exactly the same converted to a JPEG that's only a few hundred kilobytes in size.

Note how the Guetzli compression on the right is smoother but lacks some richer colors of the libjpeg compression in the center. It also aims to enhance usability for graphic designer, in addition to aiding future research on video and image compression techniques.

  • Kyle Warner