Bash script for bulk optimizing PNG/JPG files, quickly and reliably (includes download function, and GUI filepicker). It uses lossy compression, with barely noticeable differences except for filesize. Similar to tinypng.com.

Download and install to /usr/local/bin/

wget https://raw.githubusercontent.com/644/compressimages/master/compressimages
sudo install -m 0755 compressimages /usr/local/bin/

Install dependencies

• Ubuntu

  apt install -y jpegoptim pngquant parallel zenity imagemagick
  

• Arch Linux

  pacman -Syu jpegoptim pngquant parallel zenity imagemagick
  

Run it with

compressimages [options] [files] [directories]
-0         use NUL as delimiter for stdin rather than newline
-t INT     threads to use (default: 16)
-p PATH    path to save compressed images (default: $HOME/compressed)
-nh        don't ignore directories starting with . or ..
-depth INT limit find's maxdepth to INT
-cc        convert images to JPG and PNG, compress, and keep the smallest file

Read from stdin

find . -type f -name '*.png' -print0 | compressimages -0 -p output/

It can detect whether it's a directory or file, or even a (s)ftp/http(s) URL

compressimages -depth 2 Downloads/ someimage.png Pictures/ https://website.com/image.png

Running

compressimages .

will recursively find and optimize all PNG/JPGs in the current directory, saving them in compressed/ by default

To open the zenity filepicker, just run

compressimages

To bulk convert all images (including webp,svg,etc) to both JPG and PNG, compress them, and keep the smallest file

compressimages -cc

To read filenames/folders/URLs from a .txt file

compressimages < images.txt

For directories, it will scan them for more images.

For files, it will check they are PNG/JPG before continuing to optimize them.

For URLs, it will attempt to match http(s)/sft(p)/file links to images, download them using wget, and add the downloaded file to the queue for testing if it's a JPG/PNG image.

For the benchmark I created 1,000 unique JPG and PNG images of ~2.0MB each. The script successfully compressed all 2,000 images in under a minute, saving 1.7GB of space total.

Script to create the unique images

#!/usr/bin/env bash
for ((i=0; i<1000; i  )); do
	convert -size 1000x1000 xc:gray  noise gaussian "image-$i.png"
	convert -size 4000x1000 xc:gray  noise gaussian "image-$i.jpg"
done

Screenshot taken with mpv of a 4K BluRay, uncompressed (7.0MB)

Compressed version (889KB)

Transparent image of the Rubik's Cube, uncompressed (79KB)

Compressed version (24KB)

There are very small differences to the human eye in these examples, yet significant amounts of space was saved. Of course this example isn't 100% accurate (different test types on different images should be performed, like in: https://testimages.org/), but this should hopefully provide a rough idea as to what this script can do thanks to pngquant and jpegoptim.

More useful links: https://tinypng.com/ https://tinyjpg.com/

• parallel
• jpegoptim
• pngquant
• zenity (optional)
• bash >= 4.0

MIT