Binary continuous image decomposition for multi-view display




  • Paper topic: Virtual Reality
  • Software type: Code
  • Able to run a replicability test: True
  • Replicability score: 5
  • Software language: C/C++
  • License: unspecified
  • Build mechanism: None
  • Dependencies: fftw / opencv
  • Documentation score {0,1,2}: 1
  • Reviewer: Nicolas Bonneel <>
  • Time spent for the test (build->first run, timeout at 100min): 10min

Source code information


Although no makefiles nor build instructions are given, it is quite straightforward to compile -- in my case creating a Visual Studio 2017 solution, putting all files, and setting include an library paths for FFTW and OpenCV. Requires OpenCV libraries are opencv_imgproc412.lib;opencv_highgui412.lib;opencv_imgcodecs412.lib;opencv_core412.lib.
To prevent a (known) opencv issue relating to 'ACCESS_MASK' being an ambiguous symbol in winnt.h, I had to #include <Windows.h> at the very first line of main.cpp.
To activate OpenMP, in addition to addin openmp support in the visual studio configurations, you need to specify the preprocessor definition _OPENMP.
In addition to the code, a compiled version (for Windows) is also provided. Both the code and the provided executable produce the same result on the given examples (those of Fig 8 and 6), and they correspond to the results shown in the paper. The archive containing the compiled files also contain batch files which are useful to see which parameters were used to produce the results (number of bits, iterations, etc.).
Additional instructions on the hardware side are provided in a pdf document in the same folder as the code in supplementary materials.

If you want to contribute with another review, please follow these instructions.

Please consider to cut/paste/edit the raw JSON data attached to this paper.