Unsharp Mask for newsprint
Megan
you're using far too high a radius for newsprint .
(0r anything else for that matter)
try about 100 threshold and 3 pixels and remember it's
better to use unsharp mask twice at a lower value than once
at a higher.
also, check your printing inks setup in preferences. the default is
for coated paper. choose newsprint and the programme will automatically
compensate for dot gain.
with these values whta you see should be pretty much what you get.
Sean Gallavan
format and have recently purchased a slide scanner to help in this
respect. Recently, we ran our first color photographs which we scanned
from our Poloroid SprintScan 35. We used an unsharp mask in PhotoShop
with what I was told to be the "standard" settings:
Amount: 200
Radius: image dpi/200 (for us 170/200=.85)
Threshold: 4
When we ran the picture with these settings, it appeared drastically
oversharpened. Our ad design manager insists that it must appear
oversharpened on the screen in order to print properly to the newsprint
where the dots will inevitably bleed. Our photo editor insists that the
newspaper he worked for this summer says not to sharpen the photos at
all. I'm the systems manager and would like to find out what the pros
think. We're printing to newsprint with an 85lpi screen out of a 1200 dpi
laser printer. I can give you additional specs if they are needed.
Thank you,
--Sean Gallavan
Doug Foss
>Subject: Unsharp Mask for newsprint
>Date: 10 Apr 1995 02:44:30 GMT
My job requires that I work with many newspapers and many, many digital
images. My experiences have shown that if an image looks SLIGHTLY too
sharp on the monitor, it will usually print very well. I know that 'slightly'
is not easy to demonstrate in text mode, so here are some procedures
and guidelines that I have seen work.
1. Zoom in when applying unsharp masking. this will allow the operator
to check for objectionable haloes, and sharpening of random pixels.
If the current parameters cause the above problems, just undo and
change the values and try again. The final numbers will work for
most of your photos.
2. Too low of a Theshold # is usually the culprit, I rarely use less than 6.
Doug Foss
Technical Service Rep.
Konica Imaging U.S.A. Inc.
Deerfield Beach, FL
JLJPixel8
grain structure of the film used, the degree of enlargement, and the
sharpness of focus of the original.
In practice, I ordinarily use 350% at .7 pixel radius, 3 threshold.
Headshots can usually stand 500% at .5 pixel radius, 3 threshold.
"Grainier" films will take a smaller percentage, say 200% at 1 pixel
radius, 10 threshold.
Always evaluate sharp at 1:1 magnification ratio. Then you can "see" what
your imagesetter will "see" when it goes to rip the file. I always try to
make files look like a sharp photograph on the screen.
USM errors I've seen in print:
1) too high a threshold; no skin detail, but sharp edged faces
2) too high a pixel radius; visible halos
For a range, stay within 150 - 500%, .3 - 1.5 pixel radius, 0 - 10
threshold.
Higher quality originals will use a higher percentage, a smaller radius,
and a lower threshold. Lower quality originals will use a lower
percentage, larger radius, higher threshold. Try a different USM for each
file until you get a better "feel" for what you're about. The real trick
is to hide the halo just within your screen frequency...
We output an 80 line screen with a 240% total ink limit (100% black limit)
for letterpress (yes, letterpress.)
Hope this helps...
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Jim Johnson (JLJP...@aol.com) (616) 722-0320 x354
Digital Imaging Technician (616) 722-2552 fax
The Muskegon (MI) Chronicle, USA
I speak for myself and nobody else; they like it better that way.
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
"The finest carver cuts little..." --Tao Te Ching
sco...@freenet.edmonton.ab.ca
: We are in the process of moving our daily newspaper to a fully digital
: format and have recently purchased a slide scanner to help in this
: respect. Recently, we ran our first color photographs which we scanned
: from our Poloroid SprintScan 35. We used an unsharp mask in PhotoShop
: with what I was told to be the "standard" settings:
: Amount: 200
: Radius: image dpi/200 (for us 170/200=.85)
: Threshold: 4
: When we ran the picture with these settings, it appeared drastically
: oversharpened. Our ad design manager insists that it must appear
: oversharpened on the screen in order to print properly to the newsprint
: where the dots will inevitably bleed. Our photo editor insists that the
: newspaper he worked for this summer says not to sharpen the photos at
: all. I'm the systems manager and would like to find out what the pros
: think. We're printing to newsprint with an 85lpi screen out of a 1200 dpi
: laser printer. I can give you additional specs if they are needed.
Sean,
I'm sorry that Ihave to be brief at this time. I have some _serious_
deadlines to meet and it's 3:30 a.m.!
Lets start by saying that unsharp masking is a process whereby the detail
contrast is enhanced. This is accomplished by using a mask created from
the original image, but made 'unsharp' or in other words - blurred. The
mask, like all other masks, is then applied to the original to locate
edges of detail.
Technically, a percentage of the difference between the two images is
applied to the original image, therefore expanding the local contrast of
image detail. In other words a sharpened image.
What does this mean - well, in Photoshop you have three controls. Ammount
(ammount of the difference that is applied), Radius (how blurry the mask
is made) and Threshold (how much differece before sharpening is applied
to a piece of detail).
In laymens terms:
Ammount = Volume control
Radius = edge detection
Threshold = regulator
As detail gets further apart - soft focus, big enlargement, the radius
must increase.
As radius increases - amount must be reduced to control oversharpening.
As noise increases, threshold must also increase to compensate for 'grain'
The perfect ammount of USM is quite subjective and relates to the
original image as well as the size of reproduction and the halftone ruling.
How to apply...
first of all, your scanner has a built-in USM type algorythm. When I last
spoke to folks from Polaroid (a friend is project leader) the controls
were fixed based upon enlargement.
If you are applying USM in the scanner, you must dramatically reduce the
amount of USM in Photoshop. I prefer to scan raw and sharpen later
because I often retouch dust and apply tone controls _before_ sharpening
to get better effects.
Start by setting your radius based upon image and reproduction, then
apply more or less ammount to control enhancement. The two operate like a
teeter-totter, as one goes up, the other needs to come down (after a
reasonable setting is obtained)
Threshold minimises the effect of the ammount control, and should be used
with caution and in moderation. With apropriate control, the use of
threshold should be only about 10% of images IMO.
Try a few images with various settings to get a feel for the controls.
Vary only one setting at a time and watch the effect. Remember, it is
supposed to enhance the detail of edges.
I hope this helps 'cause I gotta run, I've spent 10 minutes writing this
which is 15 more than I had <G>
Scott
--
SCOTT SANDEMAN-ALLEN
Roderick Scott Corporation
email:
sco...@freenet.edmonton.ab.ca (eMail only)
rsco...@tibalt.supernet.ab.ca (SLIP)
70274...@compuserve.com
rsco...@aol.com (as little as possible)
----------------------
Confidence is the feeling you get
Just before you fully understand the problem...
Unknown.
Toby Thain
> Sean Gallavan (galla...@nd.edu) wrote:
> ...Our ad design manager insists that it must appear
> : oversharpened on the screen in order to print properly to the newsprint
> : where the dots will inevitably bleed...
This is quite true, mainly due I think to the low-pass filtering effect of
halftone screening (high frequencies - fine details - tend to drop out).
The coarser the screen ruling, the lower the frequencies (coarser details)
that will reproduce.
> : Our photo editor insists that the
> : newspaper he worked for this summer says not to sharpen the photos at
> : all.
In my experience virtually all flatbed scans (even PhotoCD) need sharpening.
One rule of thumb I've used is, set the USM radius to the ratio of the
image resolution (at final size) to twice the screen ruling (i.e. 300dpi
scan reproduced in 65lpi newspaper = radius of 2.3). Even drum scans will
benefit from extra sharpening when the screen ruling is coarse, especially
if scanned assuming typical quality offset reproduction.
Mad Dog
to...@netspace.net.au (Toby Thain) wrote:
That is about right. You have to remember that at 65lpi the distance from
dot to dot is getting wider than at 150lpi, sometimes even sharpening cant
get good detail at that low of lpi.
--
There are no strangers on the net, only passerbys