Re: Proofs with imagesetter dot simulation
Re: Proofs with imagesetter dot simulation
- Subject: Re: Proofs with imagesetter dot simulation
- From: Roberto Michelena <email@hidden>
- Date: Tue, 27 Mar 2001 09:08:18 -0500
>
Sounds very interesting although there are two things that have been nagging
>
me with this type of solution which I hope you can clarify for me. The
>
first is how to achieve a reliable dot reproduction of a 2400 or more
>
imagesetter with printers that only go up to around 1440 or so.
I believe (some educated speculation here) that they combine subsampling
with error diffusion, and maybe some sort of cellular automata algorithm
(as in the Game of Life).
In short, they downsample the original screen from, ie 2400dpi to 600dpi.
But then you end up with a lot less graylevels. So that error is
compensated for in a diffused pattern among the newly created dots. So
you end up with some mottling, as if you had done a very aggressive
supercell (like 150lpi at 1200dpi), but correct gray levels in the area.
And maybe some cellular automata is used to finetune the dot shapes,
giving higher scores to more regular shapes or to better fits to the
original 2400dpi shape.
The fact is, it works. But except for newspaper screening (90-120lpi),
where it really works incredibly great, in commercial printing it's
usefulness is debatable. Why? it's true you get screening, and a more
offset-like look. But then you usually get some mottling and patterning
in certain shades (specially some flat tints) that you won't get in the
printed offset piece, and that you won't get in a "conventional digital
proof" like an error-diffused BestColor proof. Open to opinions here, as
I said it's debatable.
>
>
The second is a question I have had which I also saw posted on this list
>
once or on another, I don't recall which. The problem is that if you have
>
the same dot structure but different ink color and paper color, you will get
>
different colors on your proof. Unless I use the same exact CMYK inks in
>
the printer, then necessarily I will have different colors from the press if
>
I have the exact same screen, dots, etc. Am I outdated here? I hope so.
(More speculation, but again with grounds) This is similar. They just
contaminate them when needed, placing the extra needed inks on top in the
form of error-diffused printer dots; also, if needed they take out
droplets (printer dots) in the middle of the halftone dot, which will
decrease density but won't be perceived as a hole because the void will
be filled by gain of the surrounding ones.
However, the color is matched on an area (a supercell) not a single
halftone dot. It is unlikely that it could be matched on each single
halftone dot since it's made up of too few printer dots to have accuracy
on color matching.
Barco calls this "dot-on-dot". I believe all do a similar thing
regardless of how they call it. Surely it gets more complex when using 6
inks to simulate, but on the other hand it gives you more accuracy in the
simulation.
best regards,
-- Roberto Michelena
EOS S.A.
Lima, Peru