You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be nice if we could specify the direction of scanning:
top-down / bottom-up
left-to-right / right-to-left
"serpentine scanning" ‒ horizontal direction of scan alternates between lines
Another, more interesting feature, is a non-constant kernel. For example, we could specify a set of kernels ‒ e.g. A = 7 3 5 1; B = 3 2 3 8; C = 4 4 4 4 ‒ which would be applied to successive pixels in sequence (A B C A B C…) or randomly, etc.
An even more interesting feature, although this is no longer Floyd-Steinberg dithering, would be a multi-pass error distribution, where the error can be pushed to previously processed pixels in the first pass(es), and the last pass would either be a standard forward-only distribution (a la Floyd-Steinberg), or the back-propagated error would simply be discarded.
The text was updated successfully, but these errors were encountered:
It would be nice if we could specify the direction of scanning:
Another, more interesting feature, is a non-constant kernel. For example, we could specify a set of kernels ‒ e.g. A = 7 3 5 1; B = 3 2 3 8; C = 4 4 4 4 ‒ which would be applied to successive pixels in sequence (A B C A B C…) or randomly, etc.
An even more interesting feature, although this is no longer Floyd-Steinberg dithering, would be a multi-pass error distribution, where the error can be pushed to previously processed pixels in the first pass(es), and the last pass would either be a standard forward-only distribution (a la Floyd-Steinberg), or the back-propagated error would simply be discarded.
The text was updated successfully, but these errors were encountered: