libcaesium

Libcaesium is a simple library performing JPEG and PNG compression/optimization using mozjpeg and zopfli.

Download

Binaries not available yet. Please refer to the compilation section below.

Basic usage

Libcaesium exposes one single function to compress, auto-detecting the input file type:

bool cs_compress(const char *input,
                 char *output,
                 cs_image_pars *options);

Parameters

input - input file path
output - output file path
options - pointer to the options struct, containing compression parameters (see below)

Return value

true if the compression has successfully ended, or false if any error occurs.

Compression options

Libcaesium supports a few compression parameters for each JPEG and PNG. You need to initialize the default values before compressing by calling initialize_parameters().

They are defined into a top level struct containing each supported file parameters, as follows:

typedef struct cs_image_pars
{
	cs_jpeg_pars jpeg;
	cs_png_pars png;
} cs_image_pars;

JPEG

typedef struct cs_jpeg_pars
{
	int quality;
	bool exif_copy;
	int dct_method;
} cs_jpeg_pars;

The first 3 parameters matters, in term of compression, while the others will be set by the compressor/decompressor during the compression progress and thus they will be overwritten.

• quality: in a range from 0 to 100, the quality of the resulting image. Note that 0 means optimization (see below). Default: 0.
• exif_copy: set it to true to copy EXIF tag info after compression. Default: false.
• dct_method: one of the turbojpeg DCT flags. Default: TJFLAG_FASTDCT.

PNG

typedef struct cs_png_pars
{
	int iterations;
	int iterations_large;
	int block_split_strategy;
	bool lossy_8;
	bool transparent;
	int auto_filter_strategy;
} cs_png_pars;

Those are the zopflipng compression parameters.

• iterations: number of iterations (more means more compression). Default: 10.
• iteration_large: number of iterations for large files. Default: 5.
• block_split_strategy: filter strategy. Default: 4;
• lossy_8: convert 16-bit per channel image to 8-bit per channel. Default: true.
• transparent: remove colors behind alpha channel 0. Default: true.
• auto_filter_strategy: legacy.

Compilation and Installation

Libcaesium uses cmake to build and install the library. Before compiling, be sure to have all the requisites. Libcaesium requires mozjpeg and zopfli installed as shared/static libraries. Please refer to their own documentation for detailed instructions.

OS X

Requirements

You can install mozjpeg and zopfli with homebrew $ brew install mozjpeg zopfli

Compile

Provided you have all the requirements, building and installing from git is as simple as typing

$ git clone https://github.com/Lymphatus/libcaesium.git
$ cd libcaesium
$ mkdir build
$ cd build
$ cmake ..
$ make
$ sudo make install

This will compile both the shared and static caesium library, the required header and a small demo application named caesiumd.

Linux

Requirements

You need basic developer utilities. On Ubuntu you can install them with

sudo apt-get install libtool autoconf git nasm pkg-config cmake

Then run install.sh to get the latest mozjpeg and zopfli.

Compile

Provided you have all the requirements, building and installing from git is as simple as typing

$ git clone https://github.com/Lymphatus/libcaesium.git
$ cd libcaesium
$ mkdir build
$ cd build
$ cmake ..
$ make
$ sudo make install

This will compile both the shared and static caesium library, the required header and a small demo application named caesiumd.

Windows

Compiling on Windows is somehow tricky. You can achieve it with MinGW (tested) or Cygwin (not tested), but it's better to stick with the binaries provided.

Compression vs Optimization

JPEG is a lossy format: that means you will always lose some information after each compression. So, compressing a file with 100 quality for 10 times will result in a always different image, even though you can't really see the difference. Libcaesium also supports optimization, by setting the quality to 0. This performs a lossless process, resulting in the same image, but with a smaller size (10-15% usually).
PNG is lossless, so libcaesium will always perform optimization rather than compression.