scanning - theory of color

7/29/2019 Scanning - Theory of Color

1/52Scan

ning

T he Creative W orld of Digital D ata


2/52

Introduction

H ello and W elcome!

With t he small book youre now holding, we at

Heidelberg CPS would like to offer and share with

you the knowledge and experience in the field of

scanning which we have acquired in over thirty

years of pre-press development and production.

I t s import ant to r ealize that t his isnt a manual

for your scanner, and so aft er reading this book you

may not be a fully qualified scanning expert.

But youll gain an excell ent i ntroduction into t he

world of scanning and it s vast potential , so that

even as a beginner, youl l easily be able to achieve

scanning results which fulfill high expectations.

To make sure your introduction to scanning is as

excit ing as the topic it self , well only address as

much theory as is absolutely necessary for a basic

but fundamental understanding of the technology.

That way your first priority can be a thorough

famili arization with our useful, practice-oriented

ti ps. (Some of which may be of help to even the

more experienced scanner user.)

Since we have a few surprises waiting for you,

theres no point in dragging out the introduction

any longer.

H ave fun reading, playing, and scanning !! !


3/52

Theory of Color

Bits and Bytes

Opacity and Density Resis

Drum Scanners versus Flatbed Scanners Te

Scanner Handling

System Calibration C

Storage Structure

I mage Assessment

Prior to Scanning

Scan Process From Docum

Scanning Standard Originals

Scanning Heads-Up Originals

Scanning with Image Manipulation

Pixel Retouching

Data M anagement


4/52

But we can see the colors of the visible spectrum

because the electromagnetic radiation in these

areas has a wavelength capable of acting as a

color stimulus for our eyes.

Thus, to perceive color we need

q light

q an object either to reflect or to let the light

through

q the eye and the brain.

Different colors are perceived just as subjec-

t ively as sounds, tastes or smells.

The same color can provoke a number of dif-

ferent reactions in dif ferent people. L ike our

little,blue circle here:

Sometimes these reactions depend not only on a

persons sense of taste but also on his or her color

sensit ivit y.The spectr al sensit ivit y of each persons

eyes varies slightly. That s why two people might

come to different conclusions when judging a

single color. One might call t urquoise green, while

the other might say that its bluish.

Colors are not only perceived subjectively how-

ever, but ar e also infl uenced by external circum-

stances.

One influence might be the kind of light source.

Which one of us hasnt seen a beautif ul colored

sweater in a department store which ended up

looking absolutely terrible in daylight?

At the same time, the size of an object can be

important as well.

I magine that youve had to pick out a fabri c for

your sofa by choosing f rom among small swatches

of fabri c. When you go to pick up the sofa, one

glance and you take a big step back to ask your-

self : is this really the color I picked out? Colors

often prove to be much more intense on large sur-

faces than on smaller ones.

How is it that we see different colors?

To answer t hat question, we have to star t out bytak ing a look at both physics and biology. We pro-

mise, however, to keep it lively and color ful for you.

This graphic shows the whole range of electro-

magneti c waves, of which only very few are visible

to the human eye.

We cant see UV- or i nfr ared rays, for example.

q What a blue! Wow!

q Thats a terribly loud blue!

q A beautiful, glowing blue. I like it!

q and so forth.Theoryof

Color


5/52

Here you see once again our li tt le blue circle, but

this time half of it is on a bright background and

half on a dark background. Clearly, the eff ect of

the color depends on the background against

which it lies.

L ast but not least, you have to t ake the obser-

vers point of view into account.

An objects color can look drastically different

depending on the angle from which youre looking

at i t. The color may look either darker or lighter.

This is especially noticeable among fabrics.

Of equal importance is the objects background.

How does the monitor display colors?

M onitor s and televisions produce color by way

of electromagnetic waves that correspond to red,

green and blue. However, screens arent able to dis-

play the full range of colors which are visible to

the human eye. Their range of color is more limi -

ted.

Any desired color can be produced by adding

the colors of the three color channels (RGB for

Red, Green, and B lue).This is called additive color

mixing. I f t he colors of t wo of the color channels

are mixed in equal propor ti ons, new base color s

are created.

Bl ue and green add up to a bright , light blue cal-

led cyan. M agenta, a bright pink, is made by mix-

ing red and blue. Red and green together make yel-

low.

I f r ed, green, and blue light are mixed equally

together at f ull power, you get white light.

Theoryof

Color


6/52

I magine that a surf ace printed magenta and cyan

is ill uminated by normal daylight. As explained

above, this white daylight can be thought of as

composed of equally bright r ed, green and blue

rays of light. I f t his white light str ikes a colored

sur face, the surf ace absorbs the complementar y

colors from t he light and reflects the rest. Our

surface was printed with equal parts of cyan and

magenta. Cyans complementary color is r ed, while

magenta s complementar y color i s green. Wi th the

subtraction of the absorbed, complementary

colors, all thats left is blue. Blue is reflected, andagain we see our little blue circle.

H ow are colors printed?

Pri nted colors dif fer fr om monit or colors in thattheyre produced by overlaying ink pigments on

paper instead of by combining different wave-

lengths. I f however you were to tr y to print red,

green, and blue on top of one another, you

wouldnt be able to produce very many colors.

You wouldn t be able to make yell ow, for instance.

This is why printing inks work using subtractive

color mixing, which wed like to intr oduce to you

at t his point.

Theoretically, if we were to print a surface wit h

equal par ts of cyan, magenta and yell ow, wed see

black, because all colors would be absorbed and

none reflected.

I n practi ce, however, thi s black looks more li ke amuddy green or brown. That s why four color s are

general ly used in print ing ( you may have heard t he

expression four-color print ). Black is used as

the fourth color in order to achieve a real black.

Thus, a normal four-color print is produced by a

CMYK basis (CMYK stands for Cyan, M agenta,

Y ellow, and Bl ack) .Theoryof

Color

Print ing with Cyan, M agenta Yel low and Black Printing with Red, Green and Blue


7/52

For practi cal purposes, this also means that con-

ventional color printing only tolerates colors that

can be mixed fr om these four base colors. (A ll

other color s have to be speciall y printed.)

The print spectrum is thus smaller than the visi-

ble spectrum and also smaller than the monitor

spectrum.

I f color perception is so very subjective, how can

you make someone understand which color youd

like your new car? How do the scanner and com-

puter make sure that the scanned image is colored

just li ke t he or iginal?

Clearly we need a system that precisely defines

diff erent colors.

For that reason every color has been described

by three base values:

q tone (hue)

q saturation (chroma)

q brightness (luminance)

These three graphics clearly show what the

terms describe.

If you follow the circle of the first illustration,

you can recognize different color tones like green,

yell ow, and red. Looking at t he second il lustrati on,

moving fr om the center outwards, the color satu-

rat ion clearl y increases.The color s become ri cher.

You can see a progression of varying brightness

from the darker bottom to the brighter top in the

third illustration.

All three values of any color can be measured

wit h a special tool call ed a spectr ophotometer. You

can easily alter an image by changing even just oneof t hese values. That s why they oft en appear as

variable parameters in software for image ma-

nipulation.

A number of systems have been developed using

these measurable values. Each of these systems allows

every color to be assigned a clearly defined value.

1 Tone (H ue) 2 Saturation (Chroma) 3 Brightness (Luminance)

Theoryof

Color


8/52

Theoryof

Color

I n thi s coordinate system every color has been

assigned a specifi c point. I nside the space of t his

abstract tr iangle (or shoe sole, or colorf ul what-

ever-it-is) lie all the colors visible to the human

eye. Al ong the x-axis of t he coordinate system you

can see the red colors, while the green colors are

situated along the y-axis. This two-dimensional

illustration shows all colors irrespective of their

brightness. I f you can imagine a thir d axis which

extends perpendicularly through the plane of the

page (and against which youd bump your nose ifyou got too cl ose), the various values of brightness

would li e here. One problem wit h this systems il lu-

stration, though, is that the distances between the

colors dont correspond to our perceptions of color

dif ferences, and that including the thir d dimension,

and thus the brightness, is very dif fi cult .

To avoid confusion, though, wel l only intr oduce

two of them to you:

I n 1931, the Yxy Color model, also called the

Norm Color model was developed by the

Commission I nternationale dEclair age (CI E) . The

CIE defines reliable standards for purposes of

color production and lighting. An ill ustration of

their Norm Color system looks like this:

-bblue

-agreen

+ byellow

+ are d

To address thi s short coming, in 197 6 the com-

mission developed the Lab model in which the

color space was represented three dimensionally.

All colors with equal brightness still lie on a

single plane of the model.

Here you can see how each individual plane

extends along the a-axis from left to right and

along the b-axis from the bottom to the top.

M oving outwar ds fr om the center of the cross,

positive a-axis values increase to the right and

negati ve a-axis values decrease to t he left . Posit ive

b-axis values increase towards the top and negati-

ve b-axis values decrease towards the bottom.


9/52

Posit ive a-values are red, so the redder a color

is, the higher i ts a-value is. Negative a-values are

greenish. I n other words, complementary colors

are aligned opposite to one another. The same istr ue of yell ow and blue. Posit ive b-values are yel-

lowish. Negative b-values are bluish.

Br ightness, meanwhile, increases from t he bot-

tom of the model to the top.

And here we have an illustration of the whole

three dimensional color model for you:

-agreen

+ byellow

+ are d

L

white

black

Theoryof

Color

-bblue

You can name every color exactly by giving its

brightness, L ( for Luminance), and its specific

a- and b-values. I n the real world, though, it might

be a bit troublesome to order a necktie in

L = + 43 .51,

a = + 1 5.4 5,

and b = - 22.85.

So, why are we bothering to explain all this ???Because some scanners and scanner softwares

work within this Lab Color space to ensure an

accurate exchange of data.

Next wed like to explain how these colors are

displayed by the computer on the monitor.


10/52

I f, however, a pixel i s the product of several bi ts,

tones between black and white can also be repre-

sented. That would be li ke a good, old-fashioned

lamp wit h two bulbs and two swit ches, of which

both or neit her or one or the other can be switched

On at any given time. I n this case, there are four(22) dif ferent, possible posit ions.

M ost scanners operate with a bi t depth of 8 bit s.

That means that 8 bits represent each pixel.

Therefore the pixel can take on 256 (2 8) different

posit ions, and the scanner can produce that many

different shades of gray.BitsandB

ytes

On these two pages youll find a number of tech-

nical concepts and defini ti ons. Some of them might

prove to be new to you.

But dont be discouraged, we wont go into too

much detail ! ! !

Wed simply like to pass on to you the funda-

mentals of the technology wit h which your scanner

and computer work because theyre of utmost

importance for the actual scanning.

So lets get down to business.

W hat is a Bit?

When you work wi th a computer, regardless of

whether youre writ ing a text , preparing a picture,

or programming soft ware, calculat ions are being

carri ed out by the computer. I n the process of

these calculations, millions of tiny, electronic

switches are activated. A bi t is one of t hese swit ch-

es. I t only knows two posit ions: On and Off . The

results of computer calculations are therefore

combinations of Ons and Offs.The computer r epre-

sents them in combinations of 1s and 0s. 1 corre-

sponds to On and 0 to Off .

1 Bit

2 Bit

11

1 0

01

00

01

Relationship between bitdepth

and the number of possible shades of gray

All types of digitalized pictures are called bit-

maps. L ike in the second pictur e on the next page,

a bitmap is composed of a lot of tiny squares.

These tiny square picture points are called pixels.

Pixel is formed from the words Picture and

Elements.

Whether a pi xel can take on black, white, gray-

scale or even colored tones depends on the pixels

bit depth. This also sounds much more complicated

than it really is.

When exact ly one bit represents a pixel, depen-

ding on whether the bit is switched to On or Off ,

the pixel wil l be white or black. I magine a lamp

wi th a single bulb and a single power swit ch, a

spotlight for example. I t can be eit her On or Off ,

and not slightly On or slightly Off .


11/52

For this reason its important to note both the

original and the reproduced sizes of the document

when determining the scan resolution.

At the same time, youl l have to consider that a

higher scan resolution always means that youre

working with a larger amount of data.This in turn

slows down your computer. Thus, you shouldnt sel-

ect a higher resolution than is absolutely neces-

sary.

High-quality scanner software automatically

calculates the scan resolution required for the

selected sizing factor.

I f youd like to start scanning right now, and

need information for the calculation of scan reso-lution, turn to the chapter entitl ed Prior to

Scanning . Thats where youll find the exact f or-

mulas. But be careful: you are bound to miss

something in between.

I f color images are being represented, the scan-ner functions with a bit depth of 8 bits for each of

the color channels. That means that RGB-repro-

ductions are produced with 256 x 256 x 256

(2563) different combinations of red, green, and

blue, and that over 16 milli on dif ferent colors can

be represented.

Hi gh end scanners work wi th a bit depth of up to

16 bits.They can thus represent 65536 x 655 36 x

65536 dif ferent colors. Such a huge number i s

really hard to imagine, isnt i t?

Aside fr om the bit depth, bit maps are described

by the scan resolut ion, the dimensions of t he docu-

ment and the selected color model.

The term scan resolution describes the number

of individual samples that the scanner takes over a

specifi c area. I t s measured in dpi ( dots per inch).

Thus the scan resolution is actually pretty

import ant, because the precision of the scan

depends on this resolution.

BitsandB

ytes

I f you scan a document wi th l ow-level resolu-

ti on, the image wil l be made of only a few dots per

inch. I ndividual pi xels will be visible as small

squares in the pict ure. This is most cl early visible

along curves or diagonal lines. They of ten appear

to be fr ayed or stepped, as though someone had

been nibbling on them.

The qualitys dependence on scan resolution

becomes especially obvious when the image is

enlarged for print ing. I f you dont increase the

resolut ion when enlarging an image, the individual

pixels themselves will be enlarged and become

visible. A resolut ion that s applied to a small docu-ment will suffice as long as the picture is to be

reproduced in the same size. I f youd like to

enlar ge the pictur e, however, you should select a

higher scan resolution.

Bitdepth: 1-bit , 2-bit , 8-bit

from black and white to color

The enlargement causes

the individual pixels to become visible


12/52

OpacityandDensity

Opacity

Opacit y (O) r efers to t he color layer s resistance

to letting light through. I n real lif e, this resistance

is never consistent across the whole surface of the

document. M ore light is let t hrough and reflected

by bright areas of the document than by dark

areas. The less light an area of a document l ets

through, the greater it s opacity. Opacit y can bedescribed by a single value the total amount of

light which str ikes the surf ace divided by the amo-

unt of light let through.

For example:

An amount of light equalling 1000 light units

stri kes a slide. The color layer of a bright ar ea on

the slide allows 100 light unit s to pass through.

This area of the slide has an opacity of 10.

The same is true for reflective documents.

I f a secti on of a newspaper clipping reflects 200

of the original 1000 light units , then this sec-

tions opacity is 5.

Theoreti cally, opacit y can have an infinite value,

but in practi ce the value usually li es between 1 and

1000.

I n practice, when light strikes eit her a t rans-

parent original (like a slide) or a reflective original

(l ike a photo or a newspaper cli pping), not all the

light t hat stri kes the original will be let t hrough or

reflected. Some light i s always absorbed. The

strength of this absorpt ion depends on the opacity

and densit y of the original s color layer.

10001000 1000

1000

1000 1000 1000 1000

900

100

101

800 20020 2

Regardle ss of whether its a transparent

or a reflective original,

the darker the surface, the higher

the opacity and density


13/52

OpacityandDensity

Density

I n principle, densit y (D) designates the samething as opacit y: the document s resistance to let-

ting light through.

Density also tends to vary across the document.

Dark surface areas have a higher density and

bright surface areas a lower one.

M athemati cally, densit y refers to t he power t o

which we rai se the opacity value. Therefore it is

general ly a smaller, more user-f ri endly value than

opacity.

That sounds pretty confusing, but in trut h it spretty simple:

Lets take another look at the graphic for our

tr ansparent document. On a relat ively bright ar ea

of the document where 100 light units are let

through, we calculate an opacity of 10. On a dark-

er area of the document where only 10 light units

pass through, the opacit y is 100 or 102. The densi-

ty of this darker area is thus 2.

As you might well imagine, it wouldnt do for a

scanner t o scan only dark ar eas properly, or f or

that matt er, only bright areas. The scanner has towork with a documents entire density range if

its to redisplay the natural contrasts between

the dark and bright areas. We can figure out a

documents density range by looking at the dif-

ference between the area with the highest density

(the image shadow) and the one with the lowest

density (the image highlight).

I f we measure, for example, a highest densit y of

2.4D and a lowest densit y of 0.3D, our document s

density range equals 2.1D.

Reflective originals generally have a density of

2.0D. Transparent originals might have a densit y

of 3 .0D or even more. Your scanner has to be

that much more sensitive if youre hoping to use

it to scan the greater image range of transparent

originals wit h sati sfying result s. The exact sensi-

tivity of your scanner is listed among its other

technical specifi cations.

Image shadow

Image highlight


14/52

DrumScannersversusFlatbedScanners

From Light to Electricity to Digital D ata

The active scanner reads light reflections from

reflective originals (li ke a newspaper clipping) and

picks up the amount of let-through-light when

scanning transparent originals (li ke a slide). The

light source projects a laser-like beam which spi-

rals across the original while the cylinder is rota-

ting.

The reflected (newspaper clipping) or transmit-

ted (slide) light is then picked up and passed along

by the scanner s eye-like sensor unit . The incominglight signal is divided into three color channels

red, green, and blue. A separate photomultiplier

tube (PM T) senses the light f or each color and

transforms it into electric signals.

Once inside a photocathode, these light rays

produce electrical charges. Unfortunately, the

resultant stream of charges is still too weak to be

of much use. But once the stream is passed along

a charge intensifi er , the available charges are

sufficient to produce a strong electric signal.

Because, however, computers arent designed to

work wit h analog, electri c signals, the signals have

to undergo one final tr ansformation: they have to

be digitali zed. This signal tr ansformation takes

place in an analog/digital converter (alias A/D

conver ter), which turns the analog signals into

useful, digit al values.

Drum scanners and flatbed scanners dont just

look diff erent, but are also diff erent in terms of

handling, technology, ...

Drum Scanners

M ounting of Originals

All the documents which you intend to drum

scan must be mounted on the scanners acrylic

glass cylinder. The documents should be thin and

flexible so that they can be laid flat on this drum.

To simpli fy the process, you can r emove the cylin-der from the scanner and attach it to a specially

designed mounting unit . Once the cylinder i s

mounted, it s easier t o posit ion the documents

exactly as youd like them to appear on the screen.

Careful posit ioning is important: the document

image appears on the screen in the same way as

you scanned it , and rotat ing the image angle after

scanning costs not only time but diminishes the

qualit y of your r esults. Ef ficiency can in fact be

fur ther improved by working wit h addit ional cylin-

ders: whil e the fir st set of or iginals is scanning,youll be able to mount other originals on another

drum.


15/52

Drum

ScannersversusFlatbedScanners

P ractical A dvantages and D isadvantages

The full-fledged application of drum scanner tech-

nology offers several advantages:

q high-speed efficiency

q high sensitivity (density resolution of up to

4.2D) , which lets you scan documents that ei-

ther reflect poorly or dont let much light

through

q minimal noise

q optical r esolution of up to 9,600 dpi, which al-

lows 3,000% enlargement ( enough for you to

enlarge a small slide to medium poster size)

The point-by-point scanning guarantees some fur-

ther advantages, li ke:

q diminished interference from stray light

q consistently high resolution across the entire

scan surface (which means absolutely no change

in scan quality)

High-quality drum scanners are universally ap-

plicable, but the mounting of originals requir es

some dexterity.

Drum Scanner Signal Processing

Original

Mirror

M irror

Color F ilter

Light Source

Photomultiplier Tube


16/52


F latbed Scanners

M ounting Originals

I n contr ast to drum scanners, documents for

flatbed scanning are laid on a horizontal tray.

Thus, they don t necessaril y have to be thin and

flexible. Some scanners will even allow you to scan

documents of pract ical ly any t hickness (books and

fabrics for example).

From Light to Electricity to Digital D ata:

A New Conversion

Flatbed scanning also requires careful posi-

ti oning of t he originals, but t his is much simpler on

the horizontal tray than it is on the cylinder of a

drum scanner. On most fl atbed scanners, the tr ay is

attached to the scanner it self. What that means

however, is t hat you can t simul taneously scan and

be busy preparing another tr ay with originals. (A t

the time being there are only a few exceptions

which of fer a removable document tr ay.)

As wit h a drum scanner, in a fl atbed scanner

light signals are first transformed into electricalsignals and then into digit al signals. I nternally

however, the fl atbed scanner s technology is quit e

different to that of a drum scanner. A halogen lamp acts as the sole light source.The

lamp shines on a single stri p of the original and the

original is scanned line-by-line.

The light reflected or let through by the original

is picked up by thousands of extremely small

photodiodes. These photodiodes are aligned on a

sili con chip. The selected charges set f ree by the

light are collected for a while and then transmitted

into a CCD (Charge Coupled Device)-line.

From here on, the signal proceeds simil arly to

the way it would in a drum scanner. The charges

are read out from the line and finally turned into

computer-ready digital data by an A /D-conver ter.

I f the CCD in question has a single fil ter for r ed,

green and blue color channels, then all thr ee color

channels can be scanned in a single pass (also

known as the Single Pass Technique).

If however the CCD is working with the help of

ext ernal f il ters, the scanning has to be divided

among a total of three passes (called the Triple

Pass Technique) t o all ow for a fi lt er change before

the star t of each pass. Almost all modern fl atbed

scanners operate with the Single Pass Technique

because it allows the added advantage of higher

efficiency.


17/52

Drum

ScannersversusFlatbedScanners

A Practical Comparison to the Drum Scanner

q most fl atbed scanners aren t as eff icient as drum

scanners

q the optical resolution is determined by the num-

ber of CCD-elements and can only be increased

or diminished by software interpolation

q because it s more sensit ive to str ay light, fl atbed

line-by-line scanning is more likely to produce

images with faults than dot-by-dot drum scan-

ning

q lower density resolution than with a drum scan-

ner

Despi te these disadvantages, a high-end flat bedscanner performs comparably with a drum scanner

when scanning normal originals.The drum scanner

is truly superior only in cases demanding very high

qualit y work. Fur thermore, new CCD technology is

continually being developed and even now the

more moderate price and simpler handling often

make flatbed scanners the preferred choice.

CCD Unit

Original

Light Source

M irror

M irror

Flatbed Scanner Signal Processing

Digital Cameras

I n the fi eld of photography, CCD technology is

also being used to produce computer images of

lar ge and three-dimensional objects. The case of

digital cameras has been outfitted with individual

matrix-shaped CCDs.

The result is that a subject can be captured

directly in digital data without resorting to film.

This technology is primarily used in photography

for catalogues, by journali sts (who send press pho-

tos over modems to their editing offices) and in

advertising for studio photography.


18/52


This table gives you an overview of the various

types of scanners, their main applications and

much more.

I n many offi ces nowadays, the scanner is just as

common and as much at home as the computer.

I t s oft en used to r ecord text and for document

archives.

Scanners are becoming increasingly popular at

home as well . When you combine a scanner wit h a

reasonably pri ced computer, you can slowly f ami-

liarize yourself with digital image preparation.You

shouldnt expect professional qualit y result s, but

youll definit ely have a lot of f un with i t.

Flatbed Scanner Drum Scanner

Office DTR Lower M id Range Upper M id Range High-End

Office Applicat ions Office Applicat ions, Pre Press, Pre Press, Pre Press,

( for example, OCR), DTP and Adver t ising DTP and Adver t ising High-End High-EndDiscr iminat ing Agencies Agencies I mage Product ion I mage Product ion

Home Users

Reflect ive Reflect ive or Reflect ive or Reflect ive or Reflect ive orOriginals, Transparent Transparent Transparent Transparent

Originals, Originals, Originals, Originals,flexible or st i f f, flexible or st i f f, flexible or st i f f, flexible or st i f f, flexible,

var ious thicknesses, a lmost any a lmost any 0.4 inch thick thinthi ckness for thi ckness for

Reflect ive Reflect ive

Originals Originals

(3D-Objects) , (3D-Objects) ,b& w and color b& w and color b& w and color b& w and color b& w and color

low average average to high high very high

300 x 300 dpi to 600 x 1200 dpi > 1200 dpi > 3800 dpi > 9600 dpi400 x 400 dpi

< 3,0 D < 3,0 D > 3,0 D > 3,7 D > 4,2 D

The vast majorit y of scans, however, are stilltak ing place in other fi elds: in the high end region,

by which we mean not only professional image pre-

paration but also small companies with their own

DTP departments as well as in advertising agen-

cies for graphic design jobs. These are exact ly the

sorts of places where a scanner can be a helpful

tool for the support of creative assignments.

ScannerOverview

Area ofApplication

SuitableOriginals

Work Ef ficiency

OpticalResolution

Density Range


19/52

ScannerH

andling

I n principle, your scanner i s just as fr iendly or as

fini cky as any other electronic device. That means

that t he scanner doesnt like to be too hot, too

cold, or t oo damp. But when preparing a scanner

work stati on, you should keep the foll owing consid-

erat ions in mind:

Place your scanner on a sturdy surface.

I t s really annoying to have to r emain absolut e-

ly still just because the slightest motion could

cause the table under the scanner to r ock, which in

turn might lead to faulty scan results.

Set up a work station in an area which can be

fair ly well protected against sun light. Reflections

on the monitor can be bothersome even if youre

just wr it ing a text . I n work that invol ves pictures,

reflecti ons can be a real pain. I mage detai ls can

become unrecognizable, and that doesnt make

your work any easier.

Shopping T ips

I f youre establishing a new DTP depart ment in

which several people are to work , you should con-

sider setting up a special station intended exclu-

sively for scanning purposes. On the one handyoull avoid crowding at t he station, and on the

other hand the stat ion can be supplied wit h it s own

specialized equipment.

Youll need a high performance computer with

plenty of memory to process and store the large

amounts of data. And, if you want t o reproduce the

scans yourself , youl l also need an appropriate

print er. Above all however, you should have a real-

ly good, large monit or wit h a suitable 24-bit

graphic card.

Only high-quali ty, well -equipped systems all ow

for truly good work.

A good scanner by itself doesnt amount to a

good scanner workstat ion ! ! !


20/52

H ow does calibration work ?

High-qualit y scanners leave the fact ory properl ycalibrated. I f aft er a while you start to notice

changes in the port rayal of i ndividual originals, you

should recalibrat e your scanner. The simplest way

to test the calibration is to rescan an original that

youve al ready successful ly scanned and saved.You

can then compare the earlier and more recent scan

results side by side on the monitor.

For calibrat ion, youre going to need a calibrati on

set which consists of:

q a calibrati on document (f or instance, a standard

IT8 document)q a disk with the determined Lab values of the

individual color fields of the test document

q calibration software

I n case youve just unpacked and instal led your

new scanner, you should check to see that your

system (which consists of your monitor, scanner,

and printer) i s full y calibrated. By calibrat ion we

mean that the pieces of your system conform to

one another or are in tune with each other.

W hy does the system have to be calibra ted?

A scanner senses red, green and blue (RGB)

light dif ferently than the human eye does. That

means that a scanner, or even a monitor, sees an

original diff erently than our eyes would, and as

a result displays the colors in a dif ferent way. Even

wit h the intr oduction of special color fi lt ers, the

image wont conform 1 00% to our expectations.

A scanner might also need recalibration if you

are install ing a new system component, even if thi s

new component is replacing an old one of the same

make and model. You should always recalibrate

your system when you change scanners, monitor s

or pri nters, or when you replace a part of a com-

ponent.

A certain aging process in the scanners halogenbulb can also bring about changes in scan results.

Once in a while you should give your scanner a

check-up. I f t he need arises, adjust or r epair

trouble spots that have developed.

Such test runs and adjustments are not only

simple but quickly taken care of.SystemC

alibration

T his is what an I T 8 document looks like

(lar ger in real life, though)


21/52

System

Calibration

The actual calibration begins with a scan of the

test document (known as the Calibration Scan).

The values for individual color fields obtained

through this scan are then compared with the

determined Lab values on the disk. From t his com-

parison, the computer can calculate a specific

color profi le for your scanner.

For al l f ur ther scan processes, the scanners

color profile will be assigned to the RGB-data of

the scanned image.

Keep in mind however, that scanner calibration

alone doesnt suff ice.The monitor and printer have

to be calibrated as well . Ei ther your soft ware will

allow you to select the proper monitor and printer

calibration settings so that you can undertake a

similar calibration, or you can request the color

profile of your component including their calibra-

tion settings from the manufacturers.

Only completely calibrated systems will guaran-

tee that the original image on your monitor looks

the same as both the unscanned original and the

original produced by your printer. By the way, an

IT8 document has a gray scale spectrum which is

very well suit ed for the check-up.

A P ractical Tip

For t he calibration check-up, rescan a gray-

scale original in color mode because its changes

are easier to notice than those of a color original.

I f you see a color cast, it s time for a new

calibration.

I T 8 Document

RGB-TIFF Fi leScanner

Calibration File ComputerStored

Lab Values


22/52

Avoid Chaos !!!

Dont be surprised that the question of data

storage arises befor e youve even star ted scanning.

Before you scan the fir st original, you should

definitely review the following:

A I s obvious and easy to expl ain

Youre working with large amounts of data.

M ake sure that you have enough memory at

your disposal before star ti ng work. Especiallywhen several images are being scanned in one

batch. You might not notice until the last

second that theres not enough disk space left

to save the final scans. You shouldnt let

annoying things li ke that happen to you.

B Takes a little longer to explain

Try to establish a definite structure for data

storage before you star t working. Decide

which folders to set up and by which system

the files should be named. I f you dont , youmay end up spending a lot of time searching

for misplaced data. This can be especiall y cri -

tical when several people are preparing the

data and each develops his own system for the

purpose of data storage. Even if you know

which folder to search through, you can stil l

have problems finding the file.StorageS

tructure


23/52

StorageS

tructure

This text (the one youre reading right now)

could for example, be stored under t he fol lowing

file names:

q stostruc, according to the text content. This

system would be fairly conclusive but might be

awkward when working with pictur es.

For pictures you might need a data name like

womanwi thch i l dunderpa lmt ree inga rden in -

fr anceonabeauti fulday, which would probably con-

flict with the softwares restrictons for the number

of permitted characters.

q 072495 , according to the date on which i t was

produced.

That could prove to be tr oublesome, though,

because after a while you probably wouldnt

remember when each file was produced.

q paul5 , because it s the fi ft h fil e on which Paul

worked on a certain day.

not so nice, however, when you pass on your

data to a third party who knows that M r. Smith

gave him the data, but who doesnt know that

M r. Smith s name is Paul, and thus doesnt know

in which stage of t he job M r. Smith worked on the

data.

That sounds trivial, but i t can be a real problem.

Thus, you should establ ish a single system to which

everyone always keeps.

Restructuring a full hard drive after the fact is

much more time consuming than keeping to one

storage str ucture from the star t.


24/52

Because different sorts of originals have to be

scanned in dif ferent ways, you should always begin

the scan process by assessing and classifying your

originals.

I n the fol lowing pages weve prepared a number

of groups and examples which should be of help in

assessment and classification.

First of all we should consider those originalswhich are most easily scanned. I n practi ce, these

Standard Originals make up the vast majority

of al l or iginals. I ncluded in the group of Standard

Originals are:

Color I mages

By color images we mean color slides or photos.

Bilevel Im ages (L ine Ar t I mages)

By bilevel images we dont necessarily mean

images which are split into two sections.

Bilevel images are any originals that consist of

only black and white no gray, no color. A hand-

written text (provided its written on white paper,

using black ink) e.g. is a bilevel image.

ImageAss

essment

Grayscale Images

Grayscale images are what you might generallycall black and whit e. These images are actually

black, white, and gray. I ncluded are slides or pho-

tographs.

Of course these originals are only considered tobe standard if they dont contain particular sub-ject-based featur es, li ke str ong color cast or lowcontrast.

I f t hey did contain such part icular f eatures, or i f

the image had to be scanned in an unusual way,

they would be counted among the Heads-Up

Originals.


25/52

ImageAss

essment

Color Cast Images

Color cast images contain a color imbalance.

When looking at one you have the feeling you are

looking through a color filter (or toned sunglas-

ses). Old yellowed photos are a good example.

Low Key Images

Low key images are low contrast just like high

key images. The dif ference is that l ow key images

are exceptionally dark.

Catchlight Images

Catchlight images contain small very bright

reflections, or catchlights , just like you might

see on windows or chrome metals.

Wi thin the group of Heads-Up Originals you can

find:

H igh Key Im ages

High key images are images which depict a

very bright subject wi th very lit tl e contr ast, like the

picture below. A f urt her example of a high key

image is the famous polar bear in snow .


26/52

U nfocused Original s

Unfocused originals are those originals in which

lines arent clearly defined.The outlines appear hazy.

Not only the subject but also the subject pr esen-

tation can contribute to difficulty in scanning anoriginal. For example we have:

Ar tistic and Computer Ge nerate d Originals

This palett e of various techniques includes, for

example, watercolors or oil paints. But also com-

puter generated images show some particularities

which you have to regard.

I mproperly Li t Originals

These originals are either very bright and over-exposed or very dark and underexposed. I n these

cases brightness or darkness is the product of a

technical mistake and not related to the depicted

subject of the image.

ImageAss

essment


27/52

Negative Originals

These are grayscale or color originals presentedin the form of negatives.

A screen dot which is supposed to be 25% black

(you might al so call i t light gray) i s smaller t han a

50% black screen dot ( which is a bit darker) ,

which is in turn smaller than a 100% black dot. A

single screen dot is always made up of several

image dots which can be either black or whit e.The

dark -er the screen dot is supposed to be, the more

black image dots you have. Just as is the case wi thbit depth, the number of representable color l evels

depends on the number of image dots that form

each screen dot.The minimum number of gray sha-

des required to represent an optically continual

color spectr um is 64.Thus, a screen dot consists of

a matrix of 8 x 8 picture dots.

The width of the print screen is described by the

number of screen lines per unit area. The unit of

measure is lpi, or li nes per i nch.

Screened Originals

Any original, be it grayscale or color, is screened

as soon as it is reproduced in a pr inted format (f or

example, newspaper clippings). The screen repre-

sents different hues with dots of various sizes.

ImageAss

essment


28/52

PriortoS

canning

Dont rush off to start scanning now that you

have assessed the originals ! ! !

Nows the time to ask yourself a fundamental

question:

Do you want t o reproduce your image exact ly as

it is now (same size, poorly li t, etc.) , or should you

alter t he image in one way or another? I f, for

example, you want to change the size of t he image,

you should take this into account in the scan reso-

lution.

Quality scanner software takes over the resolu-

ti on calculat ion for you, as the formulas presented

below are already integrated into t he program. Al l

you have to do is input the desired print-out size

and the screen ruling manually.The soft ware t akes

care of everything else.

For t hose of you however, who would like to

learn the exact formulas, here is all youll need to

calculate the scan resolution for the most used

printout forms.

For printers that can reproduce gray shades

(like a thermosublimation printer)

I n case youll later want t o print t he image wit h

a printer t hat r eproduces gray shades, you can cal-

culate scan resolut ion using the following method:

The sizing fact or is found by dividing the desir ed

size by the original size.

For example: You want to scan an original that s1x1 inches. The original i s too small , so you decide

to enlarge the reproduction to 3x3 inches and

print it with your 300 dpi printer.

The following calculation results:

I f you want to reduce your image, the sizing factor

is accordingly small er than 1. For a r eduction of

the 1x1 i nch image to 0.5 x0.5 inches, the sizing

factor would be 0.5.

The scan resolution depends above all on the

medium wit h which the image is to be repro-

duced later on.

Sizing factor = 3/1 = 3

Scan Resolution = 30 0 dpi x 3 = 90 0 ppi

Scan Resolution =

Resolution of the Printer x Sizing Factor


29/52

PriortoS

canning

Scanning for Screen P rinting

The stor ys the same, whether you want t o scan

a grayscale or a color original. I f you want t o print

the image at some later date (for use as a maga-

zine illustration, for i nstance), use the following

formula to calculate the scan resolution:

I n real li fe it is ext remely rar e that a single pixel

correlates exactly t o a single screen dot. I f the

scan resolution were to correspond to the screenresolut ion, the screen dot could take on an incorrect

color value. This would be especiall y visible at

sharp edges in the image. I n order t o avoid that ,

you have to include the screening factor in the

equation. This fact or increases the scan resolut ion

and lets the output device (for example a printer)

calculate the color value of the screen dot more

precisely.The screening factor should be roughly 2.

Then the color value of each screen dot will be cal-

culated from four pixels (a 2x2 matrix) and thus

be much more precise.So, if we stil l want to enlarge our 1 x 1 inch

image to 3 x 3 inches, but t his time for t he purpose

of printi ng it in a 150 lpi raster for a magazine, we

should calculat e the scan resolut ion li ke this:

Printing with a Screening Printer

Quite a lot of computer printers also work with

print screens.You alr eady have the formula for thiskind of printout. I n principle, it s the same as for

screen prints.

All thats left to calculate is the one value with

which well replace the printout screen ruling.This

value is dependent on the maximum print r esolut ion

of the printer and the number of desired color levels.

Screen Ruling = 15 0 lpi

Screening Factor = 2

Sizing Factor = 3

Scan Resolution = 150 lpi x 2 x 3 = 900 ppi

Scan Resolution = Printouts Screen Ruling x

Screening Factor x Sizing Factor


30/52

Back to a litt le math:

We can calculate the desired value by dividingthe maximum print resolution of the printer by the

square root of the number of desired color shades.

Dont put the book away yet !!!

This is as easy to understand as it is to explain.

Youre once again scanning our now famous

1 x 1 inch image and stil l want to enlarge it to 3 x 3

inches. L ater, youre going to print it on your

600 dpi printer and want to achieve a minimum of

64 diff erent color shades.

Youll want calculate it li ke this:

Therefore, you need the fol lowing scan resolution:

Theres a relationship between the screen ruling

and the number of r epresentable color shades. I f

youre trying to produce higher quality work by

work ing wit h a finer screen, the number of r epre-

sentable shades is r educed accordingly. Color vi-

gnettes in such a case often cant be displayedwi thout obvious gaps.

Unfor tunately, scan resolut ion has technical

limit ations. I t cant be raised to arbitr arily chosen

levels. For l arge-scale image enlargements, you

have to rely on another source of assistance to

ensure a quality printout.

This software-based assistance is called interpo-

lat ion. I nterpolat ion adds new pixels to the original

ones so that these originals dont have to be enlar-

ged as much as they did before.The color values of

the new pixels are determined on the basis of the

surrounding pixels.

6 4 = 8

600 dpi / 8 = 75 lpi

Screening Factor = 2

Sizing Factor = 3

Scan Resolution = 75 lpi x 2 x 3 = 450 ppi

Too much interpolation can blur the individual

pict ure lines, causing the enti re image to appear

fuzzy. Therefore you should never attempt to

increase the scan resolut ion of bilevel originals by

way of int erpolation. The exact scan resolut ion of

bilevel originals is determined using the formula:

Scan Resolution = Printer Resolution x Sizing

Factor. You never want to overstep the maximum

resolution of the output device.

BE CAREFUL

PriortoS

canning


31/52

PriortoS

canning

So, now to the question of whether the images

subject should be depicted just as it appeared in

the original, or i f changes in the tonal range and

gradation should be set up in order to manipulate

the image during the scan. Whats that? We stil l

havent explained what the tonal range of an

image is? No problem. Well clear that up right

now.

Take another look at our standard original for a

grayscale image:

Now take a look at the diagram which accom-

panies this image:

On the diagram you see an axis numbered 0 to

255 which ext ends fr om black to whit e. There are256 ver ti cal bars arranged along this axis. (The

number 256 should be familiar to you from the

Bi ts and Bytes chapter. I t cor responds to t he num-

ber of shades of gray that an 8 bit scanner can

distinguish.)

The height of a bar indicates exactly how many

pixels display that i ndividual shade. Taken coll ec-

ti vely, the bars form the original s histogram.

You can see that the heights of the bars are quite

consistent, as is the distri bution of various shades

of gray on the original. This consistency extendsfrom total ly black to completely whit e.


32/52

PriortoS

canning

For color images, a histogram can eit her indi-

cate the brightness of the entire image or be set

to provide the histograms of each, distinct color

channel.

For the purpose of comparison with the stan-

dard original, take a look at these high key and low

key images along with their histograms:

Clearl y, the dist ri bution of bars and thus the dis-

tribution of the pixels are by no means consistent

along the axis. Rather, the pixels of t he high key

image are bunched in the brighter region while

those of the low key image are gathered in the

darker region. I f youre interested in manipulatingthese images so that you can print them dif ferent-

ly than they originally appeared, you should now

direct your att ention to gradation.


33/52

Take a look at this diagram:

Along the horizontal axis you see the inputvalues for the various shades of gray. The ver t ical

axis displays the corresponding output values. For

the represented gradation curve, this means that

all the values will be expressed in the same shades

in which they were originally entered.

With help of these gradation curves you can

manipulate your original.

Just take a look at our slightly manipulated

standard grayscale original, it s histogram and this

altered gradation curve:

For comparison, heres the original prior t o

manipulation:

M anipulat ing an image by introducing infl uences

to the gradation curves requires some experience.

You should practice by testing the effects of diffe-

rent changes on the images of various originals.

Now t o the actual scanning !! !

You see that we have darkened the dark tones

and brightened the bright t ones. Thus, the tonal

range has been str etched out. The histogram now

depict s holes at a few places. That means that a

few intermediate shades are no longer present.

Detail s are disappearing fr om the image. This

effect is called posterisation. PriortoS

canning


34/52

Next, an automati c tonal range correction is

perf ormed to guarantee an optimi zed data t ransfer

of the RGB image.

Finally, the computer will automatically analyse

the original. I t will basically be repeating the same

analysis which you alr eady went through:

q Is this a standard original with a clearly defined

image shadow and image highlight?

q What does the tonal range look like?

q Does the image have a color cast?

The automatic tonal range correction however,

could cause a high key image to become a lot

darker because the scanner mistook it for an over-

exposed image.

Clearly, that s not our goal. We want the compu-

ter image to look j ust like the original t hat we liked

in t he fir st place.

To that end, before doing the scan you have to

choose the option of either a Standard Original

scan, a Heads-Up Original scan, or a M anipulati on

scan.

Following the scan, these options str eamline

once again into a single path.I f necessary youll retouch the image. Then

youll save the personally quality-controlled image

that youve personally scanned in a personally sel-

ected file of a personally selected data carrier.

But now to the differences of scanning in full

detail ! ! !

The prett y ill ustr ation on the right page shows in

full detail the exact sequence of the scan process.

Never theless, we want t o share a few explanatory

points with you.

Youll begin the scan process by mounting the

originals. You should tr y to ensure that t he

originals and the tray are as dust-free as possible.

M ake sure you lay the originals fl at on t he glass.

Correct folds or waves in the originals as they

arise.

An entire scan process can consist of three indi-

vidual scans.

The first of these scans is the overview scan.

On the overview screen you see the tray with all

the mounted originals. From this overview, youll

select t he area that youd li ke to scan, and enter, as

far as your software allows, the type of original

that is to be scanned (negati ve, reflecti ve, tr anspa-

rent, etc.).

Af ter t he overview scan, you activate the pre

scan. Before your scanner star ts the pre scan, your

software will probably do the following three

things:

Fir st, the computer will alt er the original

gradation so that the monitors RGB image corre-

sponds to the perception of the human eye.

ScanProc

ess


35/52

ScanProc

ess

Tip:

If you try to wipe the dust from the originals

wit h a normal dust rag, you might end up making

the situation worse despite your good intentions.

Rubbing an original brings about a static charge

which att racts even more dust. Cleaning wit h an

anti-static rag isnt a good idea eit her, since theyre

often soaked in an emulsion that can leave streaks

on the original.

The most suitable tool for cleaning the originals

is an anti-stati c brush, like the kind you used to

clean vinyl records wi th. These brushes should be

available in all art and design stores.

Pr e Scan

Pixel Retouching

Overview scan

Scan

I mage M anipulation

Scan

Heads Up Originals

Scan

Standard Originals

M ounting the Original

Data Storage


36/52

Scanning

StandardOrigin

als

This is a quick and easy chapter. Scanning stan-dard originals like this one is a very simple and

almost automatical procedure.

As long as youve thought out the appropriate

resolution in advance by taking all the relevant

points into consideration, and as long as youre not

considering any manipulations of the image, all

that you have to do now is enter the desired reso-lut ion, activate the scan, lean back and wait f or t he

final result.

But before you do that, you should review all the

relevant points one final time.

Are you sure that youre really dealing with a

standard original?


37/52

Scanning

StandardOrigin

als

Have you calculat ed the resolut ion properly?(I f youre scanning for printi ng purposes, pay

attention to the screen ruling.)

Should the original really be scanned without

manipulation?

Have you answered all these questions wit h Yes?

Well t hen, go ahead ! ! ! Scanning can be that simple !!!


38/52

Scanning

Heads-UpOriginals

Weve named these originals Heads-Up Originals

because theyre all unusual in ways that have to be

accounted for when youre scanning. That doesn t

necessarily mean that theyre more dif ficult , only

that they cant be scanned as quickl y or, most

import antly, as automatically as Standard

Originals.

In this chapter youll find the most important

types of Heads-Up Originals, their i ndividual fea-

tures, and ti ps for work ing with them so that j us-

tice is done to their unusual characteristics.

High Key and Low Key Images

The trouble with scanning high key images has

already been briefly mentioned. I f t he automatic

tonal r ange correction was acti vated, the result ant

image might look like this:

because the scanner mistook it for an overexposed

original. The image shadow would be reset to a

lighter area of the image (the value of which is too

high) and the entire image would end up darker

than the original image.

The same is tr ue for l ow key images, except t hat

here the image highlight is reset to a darker area

(wit h a changed highlight value thats too low) and

the enti re image is brightened.This is how i t l ooks:

I n order to avoid such a result , you should foll ow

the directions on the next page:

Original

Original


39/52

Scanning

Heads-UpOriginals

Go through the pre scan using all the available

automatic functions (the correction function should

be set t o normal ). Now take a look at the preview

image on the monitor. Do you like it? Does it look just

li ke the original? Good. Act ivate the scan.

M ore likely, though, you wil l see the eff ect weve

just descr ibed and t he image wi l l have changed, so

that now the high key image looks darker or the

low key image appears bright er. I f this is the case,

you should see if you prefer the result once you

have manually adjusted the image shadow or high-

light.

You can achieve a number of dif ferent eff ects bychanging the image shadow and image highlight.

Try a few different values and see what the preview

images look like. I f you like one of t he sett ings, go

ahead and activate the scan. M anual adjustment

of image shadows and image highlights calls for

some practi ce, but in t he end youl l be able to scan

and reprint your high and low key images so that

they look just like the originals.

Catchlight Images

The earli er pr inciple applies here as well : prescan with all the available automatic functions,

take a look at what comes out of it and then de-

cide what you want to do.

The problem with catchlight images is that the

program can set the image highlight in a catch-

light, resulting in an image thats too dark. I f t he

catchlights are small li ke reflecti ons on water, the

program sometimes recognizes what they are, and

the problem doesnt ari se. Larger catchlights can

become tr oublesome, though, for example, a bright

window.

This effect is very similar to what happens when

youre taking pictures wit h an automatic aper ture

selection. I f you take a picture against the light

and the sun is par t of the picture, the lighting will

be set accordingly. People or t hings in the pict ure

may end up looking like black outlines.

I f you notice that t he preview image is clearly

darker than the original, you should once again

manually set the image highlight to the proper

posit ion. This is the brightest area of the image,

aside from catchlights. But as before, manually

setting the image highlight will definitely take

some practice.

Tip:

I f youre dealing with one large catchlight, you

can select an image section for the original analy-

sis which doesnt include the catchli ght. That way,

the catchlight wont be taken into account in the

analysis, and the above-mentioned problem in

determining the image highlight wont come up.

Original


40/52

Color Cast I mages

Once again, the question i s whether t he compu-ter image is supposed to correspond to the original

or i f it should be alt ered. Should the color cast be

kept or f il tered out? M aybe you want t he yell owed

photo from the Fifties to look the same once its

scanned. I f your soft ware is equipped to correct

color casts, you should deacti vate the function by

sett ing it to uncorrected . You should also deac-

tivate the color cast balancing function if your pro-

gram is unable to recognize that a high-percent-

age color cast is probably intentional. Otherwise,

the red might be filtered out of that excellent sun-

set photo. I f you actually do want t o filt er t he redout of t he sunset, the automat ic color cast correc-

tor (i f you have one) won t suffice. Youll eit her

have to choose the option stronger for t he color

cast corrector manually or alter the saturation of

the red area by hand. That would be a shame for

the sunset, though, wouldnt i t?

Over- or U nderexposed I mages

These recommendations also depend on whether

the later computer image is supposed to actually

look like the original or if you want to correct it.

I f you dont want t o change anything, scan your

underexposed original l ike a Standard Original and

set the correction function to uncorrected so

that it is not acti vated. The scanned image will

then corr espond to your original, and be just as

poorly lit.

Corrected

that has a higher value.The result is that all areas

which are darker t han this one wil l be set to black.

The image will turn darker overall . I n underexpo-

sed images, the image highlight wont be set in t hebrightest ar ea. Al l ar eas brighter t han the image

highlight wi ll be set to white, and the entir e image

will look brighter.

If your software doesnt allow you this option,

following the pre scan youll have to reset the

image highlight or shadow by hand.

However, if you do want to correct t he image,

and your software lets you address the problem as

a poorly exposed image, it s rather simple.

By entering underexposed , you automatically

acti vate a special program which perf orms exact ly

the same job t hat weve just described for high and

low key images (except that there we didnt want

it ). The image will be analyzed and automatically

corr ected. The image shadow of an overexposed

original will be set in a somewhat brighter areaScanning

Heads-UpOriginals


41/52

Scanning

Heads-UpOriginals

U nfocused Original s

M ost scanners offer an aut omati c sharpeningoption. Just like all the other automatic f unctions,

this will also have to be shut off if the computer

image is to remain as unfocused as the original.

Then the original will be possible to be scanned

just li ke a standard original . However, this might

only rarely be the case.

I f you tr y to sharpen the image by activating the

fil ter option, the technical r esult is this:

This graphic shows you the tonal value jump be-

tween darks and brights on one section of the

image. I f t hese were to be a sharp edge in the

image, the transit ion would be very dir ect and the

curve very steep. Here, however, you see the line

climb relat ively slowly. The tr ansit ion is soft .

An even less focused copy will be produced from

the unfocused image. (Sounds illogical? Dont

worry, it ll be clear i n a minute! ) The transit ions

between the individual gray shades in the less fo-

cused copy are smoother and less conspicuous

than in the original.

When the curve of the less focused image is sub-

tracted from the curve of the unfocused original,

thi s curve result s.

I f we now add this mask to the line of the ori-

ginal image, small peaks appear at the points of

tr ansit ion. This means that at t hese posit ions the

dark region turns a litt le darker, and the bright

region a li tt le brighter. The overall contr ast

between dark and bright intensifies at this point.

The resultant image looks more focused.


42/52

Scanning

Heads-UpOriginals

Be careful !!! Too much sharpening can bring

about unpleasant side effects.

Fir st of all , the image may suddenly look very

sharp but also terribly art ifi cial. Something will

look wrong wit h the image, but you wont be able

to pinpoint the exact nature of the problem.

Fur thermore, the image contr ast might be too

drasti c, so that halos form around the bright areas

in t he image.These bright surf aces spread out and

as a resul t f ewer detail s can be seen.This effect is

similar to posterisation.

A related side effect arises when sharpening not

only affects the image subject but also causes the

individual pixels or even fi lm grains to become visi-

ble. Here, eff ects like mott ling or speckli ng can pop

up. M ott ling i s taken to mean the lar ge scale deve-

lopment of dots on the image. Film grains or un-

evenness in the fi lm text ure can be seen. Speckling

refers to the pixel inaccuracy, oft en caused by

noise, in which bright individual pixels appear i n

dark regions.

Speckling

resulting from too much sharpening

Ar tistic and Computer Ge nerate d Originals

A number of different problems can arise when

youre scanning artificial and computer generated

images. Fir st, the colors worked with oft en lie out-

side the range of printable colors and are there-

fore diffi cult t o reproduce. The same is tr ue of

color effects such as neon colors.

The str uctur e of t he image can be another prob-

lem. Oil paintings for instance, often come alive

not just because of the colors but also because of

the visible color layering.This cant be reproduced

in the scan or subsequent print.All you can do here is to try different settings

and see which of them comes closest t o por tr aying

the original.

Negative Originals

Negative originals are truly a topic in themsel-

ves.

Lets take a case in which were dealing with a

negative original with nothing unusual in the sub-

ject, a t rue Standar d Original wi thout anythingpar ticular ( not high key, not l ow key, not poorly

exposed and definitely no color cast) : thi s negative

can be scanned like a regular Standard Original.

Generally your software will let you enter the ori-

ginal as a negati ve. The reversal process proceeds

automatically so that a positive image appears on

the monitor directly after the pre scan.


43/52

Scanning

Heads-UpOriginals

Even speciali sts, however, have to resor t to good

old tr ial and error when dealing with more dif ficult

subjects.

Heres another practical ti p:

Dont wait until youre pressed for time to work

wit h negatives for t he fi rst time, but go ahead and

experiment beforehand to see what effects you can

create.They can be pretty excit ing: for instance, not

just a col or cast but a red cast f or the bright areas

and at the same time a green cast for the dark

areas is possible among many other effects.

As we just said, if youre not pressed for time

and have a full half hour to produce a top quality,

full y normal image, it can be a lot of f un!

Screened Originals

I f you scan screened (printed) originals as you

would Standard Originals, youll oft en end up wit h

the unwanted moir effect (the little stars or

bright lines).

For scanners which can`t descreen, the following

tip:

You can achieve good results if you can make

your scanner think that its had an unfocused ori-

ginal f rom the start . You engage in a lit tl e hard-

ware defocusing, so to speak.

This is achieved by simply laying a glass pane or

a fil m layer over the document t ray. The originals

will be assembled on the pane of glass and thus bebeyond the range of f ocus. A pane of glass is

actual ly better suit ed because film can bring about

stray light eff ects. I t should be about one tenth of

an inch thick and above all have polished edges.

However, be careful: make sure that you dont

scratch the document tray in the scanner!

M oir

How can you avoid this effect so that you canscan and later even re-use a newspaper clipping?

Your soft ware has a descreening function. This

function helps you to artificially defocus the origi-

nal. For t hat, however, you have to know the scre-

en ruli ng with which the image was print ed. Once

again back to the screen counter!

All this assumes that defocusing and later refo-

cusing will effectively deal with the moir effect.

But unfortunately we cant offer you a 100%

guarantee of success.


44/52

Youve already read a lot about image manipu-

lation in the section on scanning Heads-Up

Originals. Every t ime you darken an overexposed

image or filter out the offending shade in a color

cast or iginal, youre perf orming an image manipu-

lat ion. Above and beyond that, though, theres stil l

a lot of room for manipulation ! ! !

1 pict ure says 1000 words, and on these two

pages alone we have 6 pict ures. Here you see

our standard color original as a grayscale ori-

ginal and our usual grayscale original as a bilevel

original.

On the following page weve manipulated ourStandard Original in a number of very different

ways. Aft er the pre scan, all you do is alter the

gradation cur ve of t he image and star t the scan on

the basis of t hese al tered sett ings. Weve depicted

the gradation curve that goes with each of t he ima-

ges in order to show you how we set up the mani-

pulation.

Weve achieved this effect by intentionally mis-leading the scanner. When you enter t he sor t of ori -

ginal youre supposed to be scanning, no ones forc-

ing you to tell the tr uth. Any time that youre

mounting a color original on the document tray,

you can claim t hat i t s a grayscale original. The

scanner automatically sets the conversion into

action.

Scanning

withImageManipulation


45/52

Scanning

withImageManipulation

I f your soft ware has a Sector Correction

menu option, one simple manipulat ion possibi li ty

stands out. Wi th this option you can change the

saturation or even the tone of an individual color

(hue), which quickly and easily allows you to

achieve effects like these:

These pages depict only a small proportion of

the possible manipulat ion eff ects. Experimentat ion

in this area will not only be worth t he effor t, it s

also lots of fun !!!

Variation of all color channels

Variation of cyan channel

Variation of magenta channel


46/52

Technical ly speaking, when retouching youre r e-

placing individual pixels whose color values have

been altered by dust or scratches. I n actual prac-

tice what youre doing is copying a neighboring

pixel. Even though it sounds complicated, this is

very simple if you use the right tool.

Since however the typical image usually doesnt

contain two neighboring pixels with exactly the

same color value, it wont suff ice to simply recolor

all the fault y pixels in the same tone. The re-touching would be just as str ik ing as the imper fec-

tion.

To successful ly remove disruptive fact ors, you

need a tool with which to reconstruct the diverse

arrangement of the individual pixels.

To that end, youl l copy an enti re group of pixels

from a neighboring area with similar colors and

use it to replace the faulty pixels.

This retouched picture will be much less con-spicuous.

If however there are large imperfections across

the entir e image, this practice can be rather time

consuming.Thus its really important that you pro-

tect your originals fr om unwanted fluff, water, and

so on.

Dust, fluf f, scratches, water marks are what you

dont want to see on your originals. But unfor tun-

ately, in t he real world you cant always protect

them from the environmental influences which

bring about these side effects.

Af ter t he pre scan, you can check to see if a bi t

of fluff or dust appears on the digital image on

your screen. These imperf ections wil l be especial ly

clear t o the eye aft er drasti c enlar gements.

You should wait until after the scan to remove

them, though, because then the problem can be

seen in finest detail.

Even though much is entir ely automati c, pixel

retouching still has to be carried out by hand.

PixelReto

uching


47/52

PixelReto

uching

I mperfect pixelsreplaced by one single color

I mperfect pixels

replaced by neighboring pixels


48/52

T his might make the situation clearer:

Your scanner software operates with a specific

color model (CI EL AB, for example). I f you want

to use the data for a number of different applica-

tions you should definitely save the data in a for-

mat that corresponds to this color model (a CIE-

L AB format, in this example). From here, the data

can be reformatted to correspond to further appli-

cations.

If however you save your data in a special for-

mat (like CM YK for offset printing), and then

afterwards have to use it for another application

with a second special format (like RGB data for amult i-media show for instance), youre going to

end up reformatt ing twice. The data will fi rst have

to be converted to match the color model of the

scanner software before it can be changed to the

new format .This double conversion not only wastes

time but also threatens the quality of the end

product.

Here you see that the space requirements from

one piece of data to the next vary depending not

only on the kind of original involved (whether its

a graytone or color image, for example), but also

decisively on the color space ( RGB, CM YK, CI E-

L AB ) . Never thless, you shouldnt simply select

the format for your data which requires the least

amount of memory. When selecti ng a format you

have to consider the datas intended use. I s the

data going to be used only once or over and over

for different purposes?

After the scan and pixel retouching (if neces-

sary) all you have left to think about is saving the

data and handing it over. At thi s stage, there are a

couple of guidelines you should follow to ensure

that everything fl ows as smoothly and eff icientl y as

possible.

Its important that you take time to consider

data and image formatt ing before you star t t o save

your data.This table shows the space requirements

for various originals in different image formats:

DataManagement

I mage ( 6.3 x 7.9 in ) Disk Space RequiredStorage as: TI FF File

without Compression

Bilevel Original, 600 ppi 2,195 KB

Bilevel Original, 1,200 ppi 8,729 KB

Grayscale I mage, 200 ppi 1,984 KB

Grayscale I mage, 300 ppi 4,416 KB

Color I mage saved as RGB or Lab File, 200 ppi 5,856 KB

Color I mage saved as RGB or Lab File, 300 ppi 13,120 KB

Color I mage saved as CM YK F ile, 200 ppi 17,472 KB


49/52

General Tips for D ata Transfer

You should be sure to ask five times over whatkind of data carrier youre supposed to use when

transferring the data.

Diskettes are often inappropriate for the trans-

fer of large amounts of data.

DAT tapes are also problematic because they

can be formatted in various ways.

Magnetic optical disks (MOD) are generally

tr ouble-f ree, as are removable cart ridges for which

the data carriers designation has been standard-

ized. But even here there are a number of di ff erentformats (for example cartridges for 44MB,

105M B, 200M B, etc.). You should therefore ask

about t he hardwar e prerequisites early on.

I n case the data are to be used for pr int ing, be

sure to ask ten times over what data format you

should use. Veri fy w hich printi ng process, paper

and screen ruling have been lined up.This verifica-

tion is without a doubt faster than reworking the

data or scanning the entire original set once again.

(A n especiall y import ant point, since right before a

deadline youre not likely to have a lot of time atyour disposal.) Time lost due to unreadable data

can be a terri ble aggravation.

I t s even more aggravating when fal se or mis-

sing information leads to mistakes in data storage

and transfer, or t o printed work that doesnt at all

look like its supposed to.

M ake sure that you have all the necessary infor -

mation!

DataManagement

The same is true for the image format:

The most common image format is TIFF

(Tagged Image File Format) in which the imagesare saved pixel by pixel. Al l of t he most common

systems support the TIFF format.

So even if youre constantly using different pro-

grams for di ff erent purposes, you can general ly

import all your data as TIFF data into the pro-

gram youre using at the moment and continue

your work worr y-free.

I n practi ce, thi s means that youre always on the

safe side as long as you save your data in t he color

model of your scanner software and in the TIFFformat.

Now, if you know exact ly how you plan t o use the

data and if youre really only going to use it for a

single purpose, you can go ahead and deviat e from

this sure fire practice.

So, you know for example that the originals are

to be used for layout work and wi th a specifi c DTP

program. I t s probably the right choice to save the

data directly in CMYK -EPS format. EPS stands

for Encapsulated Postscript. I f you save a CMY K

image as EPS data, a total of five data pieces willbe produced.They first of all contain a main file in

either TIFF or PICT format that offers a rough

preview of the entir e image. M eanwhile, the fine

data for the image are saved in four different files

arranged by the four, distinct CM YK color chan-

nels.

The rough preview can be of use when posi-

ti oning your l ayout. The small er amount of data in

this image simplifies and speeds up the work in

comparison to the fine data.

I f youre absolut ely sure that the scans are to be

printed and you know the exact printing process,

the exact screen ruling and the sort of paper to be

used in printing and you also know that the scans

wont be used for another printing process some-

time next week then, and only then, you should

save the data in t he corresponding CMYK format.


50/52

Well , that s the fir st step as far as were concerned.

We hope we were able to make this booksdemanding content understandable and even en-

tertaining, despite all the theory.

Now t hat your scanning career s getti ng star ted,

be sure to take time to t r y things out. Experiment!

Play with the possibilities that your scanner and

soft ware off er and see what kinds of eff ects theyl l

let you create.

Youll soon discover that its not only fun but

quickly helps you gain valuable experience.

One la st pra ctical tip

Jot down all the effects that you see while ex-

perimenting and especially note the settings that

you used to create them. I f l ater on you uninten-

tionally produce something similar, it l l be a lot

easier for you to figure out how and why its sud-

denly come up again.

Well , there are only two things left to say in this

book:

Thanks for your interest !!!

And most of a ll, we wish you success

in scanning ! !!

OneFinal

Word


51/52

Publisher

Heidelberg CPSColour Publishing Solutions GmbHDu Pont-Strae 1D-61352 Bad Homburg

Heidelberg CPS Company425 Oser AvenueHauppauge, NY 11788

Heidelberg CPS Ltd.St. James Place, Knapp RoadChelt enham, Gloucestershir e GL5 0 3QR

Al l images are scanned, using TOPAZ orChromagraph 3900 from HeidelbergerDruckmaschinen.

PostScript is a registered trademark of AdobeSystems I nc.T I FF is a registered trademark ofM icrosoft Corporation and A ldus Corporation.Kodak is a registered trademark of Kodak AG.The ill ustrations for the Lab- and theNormcolormodell are published with kind permis-sion of M inolta GmbH, Ahrensburg.


52/52

ning

scanning - theory of color

Documents