Google has developed a new JPEG algorithm that shrinks file sizes by up to 35 percent.
The latest JPEG encoder from Google is called Guetzli. For the record, it is a Swiss German word for "cookie". The project was led by the company's Zurich Research office.
Google reduced JPEG sizes: Reasons
According to CNET, website texts are commonly smeared with tray pixels because of JPEG files. Now, Google says it found a better way to compress the said file types. Through Guetzli, people will be able to store more photos on their phones and PCs.
Additionally, websites are expected to load photos faster with Guetzli encoded files. Do note, though, that Google does not intend to replace JPEG. Instead, it just aims to minimize the likelihood of errors when images are being squeezed.
Well, faster websites translate to happier customers. To further explain, better loading speed means more sales online, more page views, and more engagement from users. The only downside of the JPEG compressor is slower encoding. Nevertheless, since it produces much smaller images, Google assures it is worth the tradeoff.
Not the first Google JPEG compressor
According to Mashable, Google also released WebP in 2010. It is a new file type that supports pictures which are 34 percent smaller than JPEG formats. However, it did not gain enough attention since it is only available for Chrome and Opera.
Obviously, if Safari and Firefox refused to support WebP, it eventually died. Fortunately, the new Google function, the Guetzli JPEG compressor, do not have such incompatibility issues. All browsers could cater to it because, technically, it is still a regular JPEG file.
In a statement released by Google Research, it said that "the visual quality of JPEG images is correlated to its multi-stage compression process." It mentioned about color space transform, discrete cosine transform, and quantization. Apparently, Guetzli focuses on the third process.
Google shows that with the more visual quality loss, the smaller the resulting file will be. It "strikes a balance between minimal loss and file size" through an algorithm that overcomes the difference between the psycho-visual models of JPEG format and Guetzli. Simply put, the Quantization process decides which colors are to be preserved or to be dropped.
