Compilation:
make
Converting a file using a single algorithm:
# method can be one of [xbr2x|xbr3x|xbr4x|hq2x|hq3x|hq4x]
# 2x means 2x bigger, 3x means 3x bigger, 4x means 4x bigger
./a.out input.png output.png [method]
Converting using all algorithms:
python3 convert_all.py [input.png]
This will generate the files input-xbr2x.png, input-xbr3x.png, etc. for each of the methods.
This library implements the xBR pixel art scaling filter developed by Hyllian, and now also the hqx filter developed by Maxim Stepin.
I needed a standalone implementation, but could not find any ready-to-use ones. I found versions of the algorithm used in various open-source projects and I eventually settled on a modified version of the one used in FFmpeg's libavfilter.
I also wanted to use hqx and needed to make some changes to the code, so I wound up merging this into the same codebase/library. This project is the result.
Original source for the xBR implementation: http://git.videolan.org/gitweb.cgi/ffmpeg.git/?p=ffmpeg.git;a=blob;f=libavfilter/vf_xbr.c;h=5c14565b3a03f66f1e0296623dc91373aeac1ed0;hb=HEAD
Original source for the hqx implementation: https://code.google.com/p/hqx/
This version is a pure C library that operates on 32-bit ARGB data. It scales images up to 2x, 3x and 4x sizes, and preserves and interpolates the alpha channel as necessary.
Project files included to compile as a DLL with Visual Studio 2012. Code should also compile successfully under GCC on other platforms, and probably with other Visual Studio versions.
A crude example application is included that uses libpng to apply the available filters to 32-bit images.
Licensed under the GNU Lesser General Public License, version 2.1. (See the
header in xbr.h and xbr.c.)