Google Guetzli can compress JPEGs by up to 35% LOSSLESSLY

Google Guetzli can compress JPEG images by up to 35% losslessly

Google have announced Guetzli – the JPEG encoder that can lower image file sizes by up to 35%! Even better, this incredible new algorithm is completely open-source – meaning any website owner will be able to improve page load times significantly!

An in-depth explanation can be found as to how this incredible algorithm works can found on both the official blog post/press release or Google’s GitHub repo.

Guetzli (which is Swiss German for “Cookie”) essentially takes a look at every single data-aspect of the input JPEG images and structures it in a way that is significantly easier to compress – resulting in smaller file sizes! Even better, the algorithm is able to compress these images WITHOUT the image quality degrading! The only downside is that this form of compression does take significantly longer when compared to the likes of libjpeg.

Example images

Google Guetzli example image 1
Courtesy of the official Google blog – 16×16 zoomed in image: Guetzli on the right, libjpeg in the middle, original on the left
Google Guetzli example image 2
Courtesy of the official Google blog – 20×24 zoomed in image: Guetzli on the right, libjpeg in the middle, original on the left

The compression speed tradeoff is one that webmasters will have to consider. However when you take into account that:

Guetzli-generated images are typically 20-30% smaller than images of equivalent quality generated by libjpeg. Guetzli generates only sequential (nonprogressive) JPEGs due to faster decompression speeds they offer.

It’s a no-brainer!

If you’re considering using Guetzli, make sure you check the prereq’s/notes that Google have listed on their GitHub page, as the algorithm is heavy on system resources:

Note: Guetzli uses a large amount of memory. You should provide 300MB of memory per 1MPix of the input image.

Note: Guetzli uses a significant amount of CPU time. You should count on using about 1 minute of CPU per 1 MPix of input image.

Note: Guetzli assumes that input is in sRGB profile with a gamma of 2.2. Guetzli will ignore any color-profile metadata in the image.

Will Stocks

Will Stocks

As a career "IT person" and technology enthusiast, I've been around technology for over 8 years now. From enterprise-grade hardware to consumer equipment, IT Support to Systems Administrator - I'm passionate about all forms of tech, learning how it works, integrates and the scenarios in which different people would use them. I started in 2017 and have also contributed to other websites around the Internet.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.