Go to file
Matteo Paonessa d61b709702 Merge pull request #2 from Lymphatus/tiff
Reverted experimental TIFF support
2017-03-18 11:52:18 -07:00
caesium Reverting TIFF support 2017-03-18 19:51:18 +01:00
demo Reverting TIFF support 2017-03-18 19:51:18 +01:00
.gitignore Header does not depend on turbojpeg anymore 2016-12-15 21:30:14 +01:00
.travis.yml Different install paths 2017-03-17 10:49:15 +01:00
CMakeLists.txt Reverting TIFF support 2017-03-18 19:51:18 +01:00
LICENSE Fixing linking mozjpeg on Linux 2016-11-16 09:41:46 +01:00
README.md Moved parameter init to the library 2017-02-21 12:54:52 +01:00
install.sh Initial support 2017-02-25 10:17:33 +01:00

README.md

libcaesium

Build Status

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,
                 const 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. Step by step instructions for each platform will be provided.

Linux and OS X

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

Platform not tested. Compilation not supported yet.

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.