Google works to make JPEG graphics smaller but not sucky
- Author: Tracy Ferguson Mar 18, 2017,
Mar 18, 2017, 0:56
'[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.
Coming back to Guetzli, meaning "cookie" in Swiss German, this compression technique is quite similar to the Zopfli algorithm which is used to shrink PNG and gzip files without needing to introduce a new format. Google Research's Zurich office led the new project, and that's why the name is Guetzli. Last, we hope that the new explicitly psychovisual approach in Guetzli will inspire further image and video compression research. Uncompressed original on the left. Ultimately, that means these smaller images will look just as good to the average person. If that's not something you care about, you can head over to the Github destination instead and check out all the files, instructions on installing the tool, and the code necessary for using it.
JPEG compression has several steps, including color space transformation, discrete cosine transformation, and quantization. The result is more detail with smaller file sizes, at a cost of substantially higher resource utilization at compression time. Mozilla, maker of the Firefox web browser, began a project in 2014 called Mozjpeg created to improve on standard compression engines. Like Google's other web-centric compression algorithms, Guetzli doesn't change the base JPEG format so all current devices and browsers can view the generated images. Presumably Google are considering optimisations and rational solutions to speed up Guetzli's compression speed performance. Guetzli strikes a balance between minimal loss and file size by employing a search algorithm that tries to overcome the difference between the psychovisual modeling of JPEG's format, and Guetzli's psychovisual model. The second image is encoded with libjpeg and the third image uses Guetzli.