Supercell Dither Masks with Constrained Blue Noise Interpolation

When halftoning images using dither halftoning algorithms, traditional
clustered dot dither algorithms trade off cluster cell size versus the
number of gray levels.  Supercell clustered dot dither algorithms are
used to obtain a small cluster cell size while allowing a large number
of gray levels to be rendered.  However, commonly used supercell
clustered dot dither algorithms have unpleasant patterns at certain
gray levels.  We present an algorithm to use constrained blue noise to
correct such unpleasant patterns.  We also illustrate how this
algorithm can generate nonlinear dither masks which can compensate for
dot gain without the use of a separate tone reproduction curve.