Depixelize PixelArt

Depixelize lets you to create beautiful looking anti aliased images using the 1990s pixel art. An implementation of the algorithm stated in 'Depixeizing Pixel Art' by Johannes Kopf and Dani Lischinski published in SIGGRAPH 2011.

Summary

The algorithm is divided into following parts -

• Input the image
• Initialize the similarity graph on the basis of pixel colors
• Resolve crossing edges using different heuristics
• Create voronoi regions using the similarity graph
• Extract B-spline curves and optimize

Examples

Input:

Output:

Building

Make sure you have CMake installed. Install any recent version (>= 3.15.0) of CMake.

With OpenGL

If you are compiling with -DCOMPILE_OPENGL=ON, you will need the following

• OpenGL
• GLUT - The OpenGL Utility Toolkit For glut:

On Linux

sudo apt install freeglut3-dev

On Windows

Download glut libraries from here (MSVC or MinGW according to the compiler present on your system) and extract it in a folder. Then, use GLUT_ROOT_PATH variable to use the downloaded package.

Steps to build

cmake -H. -Bbuild
# Add -DCOMPILE_OPENGL=ON -DGLUT_ROOT_PATH=<path/to/glut/installation> if required.

cmake --build build

# Binary should be present at ./build/depixelize-*
# NOTE: On windows, the GLUT DLLs are copied alongside the binary, thus if you moe the binary, you want to move them together
cmake

Running

./build/depixelize-gl ./test/dolphin.bmp
./build/depixelize-svg ./test/dolphin.bmp ./test/dolphin.svg

Acknowledgments

• Depixelizing Pixel Art by Johannes Kopf and Dani Lischinski]
• YUV/RGB Conversion formulas
• BMP loading library from sol-prog/cpp-bmp-images
• SVG Creation library from adishavit/simple-svg