A new (slow) open source JPEG algorithm makes images 35% smaller and looks better than older compression systems

Guetzli is Google's new free/open JPEG compression algorithm, which produces images that are more than a third smaller in terms of byte-size, and the resulting images are consistently rated as more attractive than traditionally compressed JPEGs. It's something of a web holy grail: much smaller, better-looking files without having to convince people to install a plugin or browser makers to support a new file-format.


In a blog post that accompanies a scholarly paper, two of the algorithm's designers describe Guetzli as "very similar to our Zopfli algorithm, which produces smaller PNG and gzip files."


Guetzli achieves new efficiencies in the quantization stage of JPEG compression, a major culprit for the loss of visual quality. Guetzli has its own psychovisualization model that improves on traditional JPEG models, discarding information more intelligently.

One downside: Guetzli is "extremely slow" relative to traditional compression, which might hinder the production of on-the-fly images. The authors don't quantify "extremely" slow, but note that it "may be too slow for many practical uses."


Guetzli is a new JPEG encoder that aims to produce visually indistinguishable images at a lower bit-rate than other common JPEG encoders. It optimizes both the JPEG global quantization tables and the DCT coefficient values in each JPEG block using a closed-loop optimizer. Guetzli uses Butteraugli, our perceptual distance metric, as the source of feedback in its optimization process. We reach a 29-45% reduction in data size for a given perceptual distance, according to Butteraugli, in comparison to other compressors we tried. Guetzli's computation is currently extremely slow, which limits its applicability to compressing static content and serving as a proof- of-concept that we can achieve significant reductions in size by combining advanced psychovisual models with lossy compression techniques.


Guetzli: Perceptually Guided JPEG Encoder [Jyrki Alakuijala, Robert Obryk, Ostap Stoliarchuk, Zoltan Szabadka, Lode Vandevenne and Jan Wassenberg/Computer Vision and Pattern Recognition]

Announcing Guetzli: A New Open Source JPEG Encoder
[Robert Obryk and Jyrki Alakuijala/Google Research]


(via Kottke)