Quanta is a microservice for JPEG image optimizations built using Swift programming language. It is similar to other online tools like jpegmini.com, tinypng.com or compressor.io but it is an in-house solution and based on a proven solution like MozJPEG.

Compared with other mozjpeg wrappers, this software has low memory overhead and fast performance, which can help with horizontal scaling with a large number of instances and simultaneous requests.

We used Swift and Vapor 3 to handle HTTP requests. Check also excellent libraries without which quanta can't work:

• mozjpeg

Quanta requires MozJPEG to be installed on host.

Run brew install mozjpeg.

You can use the provided script install-linux-dependencies.sh to compile and install mozjpeg. Also, follow official documentation in the repository: https://github.com/mozilla/mozjpeg

Run make to compile the software.
Go to .build/release/ directory and run ./quanta.

There are multiple ways to use Quanta to recompress the JPEG file:

• REST(-ish) API
• proxy method
• via GUI

Quanta provides REST API. This simple endpoint optimizes image by normal HTTP request and returns image (JPEG).

!!! note Quanta always preserves original format. If you send JPEG - you will get (optimized) JPEG.

$ curl -x POST https://quanta.com/optimize/jpg/ -H "Content-Type" \
    --form fileBytes=@/tmp/file_to_optimize.jpg
    --form quality=75
    --output /tmp/optimized_file.jpg

According to the table below, as a result, you can get binary data or JSON struct with an error message.

The easiest way to integrate with an external system because quanta get an external resource and processing it and send you optimized version.

$ export IMAGE_TO_OPTIMIZE=https://quanta.com/images/quanta.jpg
$ curl -x GET https://quanta.com/from/?url=$IMAGE_TO_OPTIMIZE&quality=20

Simple UI allows you to upload pictures and compare various compressions settings.

In this section, you will find information about the performance of quanta. All samples are from the production of Allegro.

Name	Description	Size	Preview
Sample #1	a few colors, headings, irrelevant details	88 kB (960x252)
Sample #2	3 main colors (red, white, black), irrelevant details	88 kB (960x252)
Sample #3	a text (black on white), details with various colors	508 kB (1600×572)

Size reduction between 16.889% and 61.0427%.

Size reduction between 13.4709% and 64.1687%.

Size reduction between 55.7364% and 78.1365%.