Geutzli Is Slow. So?

The main complaint I often read about Geutzli, Google’s new JPEG encoder is that it takes a long time to work.

If you are compressing images based on how fast it’s for you, then I honestly don’t think it’s a fair criticism. The way Geutzli works is a very different kind of compression than other methods we employ so, yes, it does take a while to work.

Recently, I had the need to compress a sizable number of JPEGs. Keeping image integrity high, the JPEGs were saved without much compression. So, it was not a shock that the result of the image file sizes from a Photoshop exports combined with an ImageOptim pass didn’t result in large file sizings.

Setting up an automated process for Geutzli, I got up and walked away for a lunch break. When I came back the images were all processed.

Selecting the JPEGs and then pressing Control + Command + I on Mac OS, I'm able to get combined file size of the images and get a comparison of the file savings.

It turns out, by leveraging my time, I was able to get them to 60% of their original file sizes. Then I ran the images again through ImageOptim and was able to shave off another additional 1–2% file savings.

While we might not be able to point Geutzli put into a Grunt task and expect it to wrap up in a few seconds, that doesn’t mean you can’t toss a large number of images to be converted during an extended break or overnight or even over a weekend. When you come back, your users could be enjoying a substantial increase in web performance without a perceptual loss in image quality.

That can never be a problem.


Originally published at Christopher Schmitt.