This is a C++ project that generates a blue noise pattern through simulated annealing. Further details can be found in the SIGGRAPH 2016 paper "Blue-noise Dithered Sampling" by Iliyan Georgiev and Marcos Fajardo from Solid Angle. You can access this paper and others on the Solid Angle research page here: https://www.solidangle.com/arnold/research
Compiling and running the program will create four .pgm files, the original white noise pattern, and the new blue noise pattern with their respective Fourier transformations. It will also produce a text file with the blue noise data in to be used somewhere else. The images just display the first dimension, but the text file should contain as many dimensions as specified by the depth parameter. Sample values in this file are in depth first order. Parameters are currently defined at the beginning of the main function, as members of the sData variable. The only real restriction is that the m parameter (exponent of two defining the resolution) has to be greater than one.
It might also be helpful to note that it was relatively trivial to convert this to use AVX2 instead of SSE3 instruction sets. However this would require at least Haswell architecture, and so I've left the implementation here as is.
This is the input noise and it's Fourier transformation using the default settings
This is the resulting blue noise and it's respective frequency representation
- CMake 2.8
- TBB 4.2
- SSE 3