Google's new open-source image compression algorithm reduces JPEG file sizes by 35 percent

[Shawn Knight]

Google has developed a new open-source image compression algorithm that it says can reduce the size of JPEG images by 35 percent.

The algorithm, dubbed Guetzil (that’s Swiss German for cookie), is said to create smaller file sizes without sacrificing much in the way of quality.

Robert Obryk and Jyrki Alakuijala, software engineers at Google Research Europe, highlight in a recent blog post that the algorithm tries to overcome the difference between the psychovisual modeling of JPEG's format and Guetzli’s psychovisual model which 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.

(Figure 1. 16x16 pixel synthetic example of a phone line hanging against a blue sky — traditionally a case where JPEG compression algorithms suffer from artifacts. Uncompressed original is on the left. Guetzli (on the right) shows less ringing artefacts than libjpeg (middle) and has a smaller file size.)

That technical jargon won’t mean much to the average reader which is why they’ve provided a couple of visual examples to show just how effective the algorithm is.

The benefit of using Guetzil, aside from the reduction in image size and barely-noticeable changes in image quality, is that the files it turns out remain compatible with existing browsers, image processing programs and the JPEG standard itself.

(Figure 2. 20x24 pixel zoomed areas from a picture of a cat’s eye. Uncompressed original on the left. Guetzli (on the right) shows less ringing artefacts than libjpeg (middle) without requiring a larger file size.)

The one major shortcoming with Guetzil, however, is that the algorithm takes significantly longer to create compressed images compared to currently available methods. That may seem like a deal-breaker but because of how well it performs in the quality department, Google feels the slower compression time is a worthy trade-off.

Guetzil is available for anyone to download and try right now from GitHub.

Lead image courtesy Marcio Jose Sanchez, AP

Permalink to story.

https://www.techspot.com/news/68565-google-new-open-source-image-compression-algorithm-reduces.html

 

[MoeJoe]

Progress

 

[Legionnaire]

Thought WebP might be worth a mention here...

 

[DelJo63]

Always stick with W3C standards for greatest compatibility. *IF* and *WHEN* Google convinces the W3C to adopt this, then-and-only-then is it worth consideration :grin:

 

[Yynxs]

The best part of the new standard is your picture gets sent to Google automatically every time it's viewed.

 

[Kotters]

Always stick with W3C standards for greatest compatibility. *IF* and *WHEN* Google convinces the W3C to adopt this, then-and-only-then is it worth consideration :grin:

Apparently it's just a different algorithm for generating jpegs, as the article has been written. But don't let that stop you.

>the files it turns out remain compatible with existing browsers, image processing programs and the JPEG standard itself.

 

[Igrecman]

The JPEG 2000 format ( .jp2 .jpx ) has been around for 17 years. It has a much superior compression ratio and quality than jpg, but it never got mainstream usage on the Web unfortunately. So why would we need Google's jpg format? There already is a better jpeg format. Same with .gif, there are better options but we keep sticking to what's been around for ever.
https://en.wikipedia.org/wiki/JPEG_2000

 

[Kotters]

The JPEG 2000 format ( .jp2 .jpx ) has been around for 17 years. It has a much superior compression ratio and quality than jpg, but it never got mainstream usage on the Web unfortunately. So why would we need Google's jpg format? There already is a better jpeg format. Same with .gif, there are better options but we keep sticking to what's been around for ever.
https://en.wikipedia.org/wiki/JPEG_2000

It's not a new format. It's a new library for encoding jpegs.

 

[Jack007]

No use creating a jpeg image engine that is unefficient and slow. if it cant do batch processing faster than current jpeg engine it has to do back to the drawing board for more modifications so it can get faster

 

[Kotters]

No use creating a jpeg image engine that is unefficient and slow. if it cant do batch processing faster than current jpeg engine it has to do back to the drawing board for more modifications so it can get faster

yeah, **** h.265. MPEG 2 for life.

 

[OutlawCecil]

The best part of the new standard is your picture gets sent to Google automatically every time it's viewed.

It's amazing you still connect the internet... You daredevil you. Always taking risks. *rolls eyes*

 

[Yynxs]

It's amazing you still connect the internet... You daredevil you. Always taking risks. *rolls eyes*

New drugs. Soon as they get the dosage right I can go outside if I can get my helmet back.

 

[Yynxs]

On a more serious note:

https://techxplore.com/news/2016-12-compression-format-timbre-music-loss.html
https://phys.org/news/2012-11-music-ears-science-timbre.html (2012)
http://www.aes.org/e-lib/browse.cfm?elib=18523

are about 'emotional' loss in sound compression. More importantly, this may have affected what music is created and sold in the industry.

I see from the article that "...less ringing artefacts than libjpeg..." are a 'feature' of this compression. I'm wondering if there is any exploration of perceptual factors and emotional involvement similar to 'timbre loss' in the visual industry?

edited to clarify the question parameter