From: Jeffrey D. Mathias (jeffrey.d.mathias@worldnet.att.net)
Date: 11/09/01-01:41:15 PM Z
Kees Brandenburg wrote:
> Did you see a relation between the resolution of the original file
> and the amount of posterisation?
Posterization is the stepping of tonal values. There is always some
posterization at some level, whether digital bits of data or molecular
chunks or crystals. This is not a concern if enough steps are produced
in certain ranges of tones. This is for the most part independent of
resolution, unless the resolution is low enough that it causes the
actual number of tones per distance to be less than the number of tones
per tonal range required in that distance (unlikely). However, it is the
ability of each piece to obtain a diverse enough tone to achieve fine
enough posterization. Resolution only dictates the maximum number of
pieces per distance.
Any interpolation is to be avoided, and any manipulations in software
are to be careful and minimized, if posterization is to be controlled
and the original image information is to be preserved and transferred to
the negative and then to the print. For example: some texture can be
introduced into a highlight area to replace that lost from excessive
posterization, but this will be similar to painting the texture onto the
final print.
Further and most important, the software, and the printer, and inks must
be capable of delivering enough data information (bit depth) so as to
keep the posterization fine enough so as to not interfere with subtle
tonal values and their discrimination.
Another important consideration for finding the threshold posterization
is that the discrimination of two adjacent tones is more important than
the total number of all tones (although related, but not linear).
Depending on the printing process used (and other aspects), for each
tone throughout the total range there is an adjacent tone that is just
discriminated. The posterization should be fine enough to allow this
discrimination to be determined by the printing process (and associated
aspects). Remember also that this discrimination is not linear with the
range of tones from black to white.
Additionally, the resolution is likely limited by the selection of paper
used for the final print. A smooth paper like Crane's cover-90 will
only provide a maximum resolution of about 600 lpi. A rougher paper
will provide a lower maximum resolution. Blocks of tone in an image are
generally larger than the maximum resolution, and are so required to
independently measure posterization.
"Continuous tone" (or the perception of such) is achieved over a certain
distance when the posterization is fine enough to not limit subtle
differences and the resolution is fine enough to allow enough steps in
that distance. (This perception also assumes the printing process used
to be finer than the eye. Some printing processes may not produce
"continuous tone".)
> Did you use non-kryloned prints too? Were they still sticky as is
> often seen with transparency media?
Only a small corner was not Krylon sprayed. The Krylon did seem to
help, but Joe Smieglitz should be able to provide better information as
to what the Krylon does and how to best apply it.
Mark Nelson wrote:
> Could the original image have been posterized in Photoshop prior
> to making the negative? Did the file have a smooth histogram or
> a "comb-like" histogram?
As I mentioned in the previous post, I feel a 16-bit scan should be
used. Then yes, posterization is likely caused by Photoshop. This is
basically by the reduction to 8-bit data. Also several of the
calculations (functions, filters, and such) can increase posterization
(reduce the number of steps). But even if using GIMP 16-bit image
software, the printer (and/or it's driver) and the inks can reduce the
number of steps.
The histogram has nothing to do with posterization. A step wedge can
give a comb-like histogram and continuous tone a smooth line histogram.
But this is only information as to the total number of pieces of certain
tones comprising the image.
An example:
While one might assume that posterization would not be a problem if only
two tones were present in the image, they should be cautious as the two
tones may not be accurately reproduced. The two original tones may be
5% and 80% and due to posterization (like if using 1-bit data) are
printed as 0% and 100%.
-- Jeffrey D. Mathias http://home.att.net/~jeffrey.d.mathias/
This archive was generated by hypermail 2b30 : 12/10/01-11:12:21 AM Z CST