![]() |
Supplementary Material for Our Paper Multi-Class (Anisotropic) Electrostatic Halftoning |
||||||||||||||||||||||||
Home |
A general requirement of halftoning algorithms is that the resulting
dots represent the underlying image well. When halftoning with dots
of different sizes, however, it is often also necessary that certain
subsets represent the image as good as possible, e.g. all dots up to a
certain size. However, these requirements contradict each other such
that a good tradeoff between the different requirements must be found
in practise. Since the importance of each requirement depends on the
application, we included freely adjustable tradeoff parameters in our
halftoning algorithm which can be used to weight the different
requirements. On this page, we illustrate the effect of these
parameters using only two particle sizes to halftone a grey
ramp. However, as explained in our paper, these concepts immediately
carry over to an arbitrary number of sizes, even if each particle has
a unique size.
As expected, only the overall image is approximated well for C=1 while the small and large particles alone do not represent the image well. Increasing C forces the large particles to approximate the image also without the smaller particles, which also yields an appropriate distribution of only the small particle. If C is chosen too large, however, the large particles nearly ignore the positions of the small particles, which yields an good approximation with only the large particles, but an inappropriate approximation when considering all particles. Instead of forcing the large particles to distribute the image well without the small particles, one might also require the same for the small particles. The resulting halftone is illustrated in the last image triplet. One can notice at second glance that the small particles cluster around the larger ones in dark image regions. Apart from this issue, halftoning works well, though. < Tonemapping for 8-bit channels Main Page Influence of α > |
||||||||||||||||||||||||
MIA Group |