Google's new algorithm shrinks JPEG files by 35 percent
- Author: Desiree Holland Mar 18, 2017,
Mar 18, 2017, 1:23
This sets the algorithm apart from other methods of reducing JPEG file sizes, like RNN-based image compression, RAISR, and WebP.
Google has named the new encoder Guetzli, the Swiss German term for cookie. Other compression algorithms such as "libjpeg" can also produce images similar or larger in size than Guetzli, but the blog post confirms that its human raters consistently preferred the images produced by Guetzli.
There is a slight downside (of sorts) for the new algorithm: it takes a little more time to compress, but the benefits outweigh the processing costs, as users are on the receiving end of the image - and needn't foot the CPU bill.More news: 'Will challenge Calcutta High Court order in Supreme Court'
While Guetzli creates smaller image file sizes, the trade-off is that these search algorithms take longer to create compressed images. The company's Guetzli JPEG encoding algorithm improvement effort looks to have a greater immediate impact because of its ability to potentially reduce image file sizes by up to 35% at the same image quality level when compared to traditional encoders, while maintaining decoding compatibility with existing browsers. Guetzli strikes a balance between minimal loss and file size by employing a search algorithm that tries to overcome the difference between the psychovisual modelling of JPEG's format, and Guetzli's psychovisual model, which approximates colour perception and visual masking in a more thorough and detailed way than what is achievable by simpler colour transforms and the discrete cosine transform. Smaller file sizes means less data is being transferred around the Internet and websites will load faster. According to Google, this makes the slower compression a "worthy tradeoff". It would be interesting to see if Guetzli scores a wider acceptance. You can view Guetzli's repository on GitHub here. Even though there are other third-party ways to compress an image, but Guetzli has its way to get the data in order first and then follow up the compression office.
Note how the Guetzli compression on the right is smoother but lacks some richer colors of the libjpeg compression in the center. "Guetzli is rather slow to encode", the researchers said, suggesting it's most likely useful on image-heavy websites. It also aims to enhance usability for graphic designer, in addition to aiding future research on video and image compression techniques.