Photoshop Lab Color- P9 pptx

If instead of establishing pivot points byCommand–clicking almost under the bridge, where the water is a dull blue, we had clicked at the bottom of the image where it’s bluest of all, th

examining the LABchannels of the afflicted

file We hope, of course, that the problem is

limited to the Aand B, in which case we blur

and head for the gym Here, though, Figure

11.20A indicates that the music is still playing

in the L But because the moiré is blue and

yellow, it’s sharply defined in the B, Figure

11.20B And, with ominous foreboding for the

fate of the unfortunate moiré, please observe

that the interference pattern in the Blines

up perfectly with that in the L

As usual, we work on a duplicate layer,preserving the original channels on the bot-tom We select the jacket by one means oranother; the method doesn’t affect whatcomes next And now we blur the B, and to alesser extent the A, until the moiré is gone

That kills the garish yellow and blue, but thedamage is still visible in Figure 11.20C, as itstill exists in the Lchannel—temporarily

As we will discover in Chapter 15, blendingthe A and/or B into

Figure 11.20 Left, the L channel of Figure 11.19 Center, the B Right, the composite after blurring the A and B channels.

the Lis one of those things that in theory

sounds as crazy as an intoxicated loon and in

practice is maddeningly powerful The key is

to use one of the blending modes in which

values of 50% gray are ignored The usual

suspects are Overlay, Soft Light, and Hard

Light They work in different ways, but they

all lighten the target wherever they are

themselves lighter than 50%, darken where

darker, and do nothing at exactly 50% And

exactly 50% is the midpoint of the Aor B, a

value of zero, neutrality Blend the original B

channel (that’s why we saved the original on

its own layer) into the Lin one of these modes

and nothing much will change except for the

only area that is significantly off 0B, namely

the moiré

I used Hard Light, the most violent of thethree, which at 100% opacity not only obliter-ated the moiré in the Lbut reversed it (Figure11.21A) Note that in the Image: Apply Imagedialog, Invert must be checked Otherwise,yellower (lighter) areas of the Bwould lightenthe L, exacerbating the existing problem Weneed the opposite And inverting the channeldoes nothing to areas that were 0B I foundthat an opacity of 70% wiped out the moirécompletely The final result is Figure 11.21B

Retouching is a difficult topic, and this hasbeen a difficult chapter The idea that a majormoiré can be obliterated with a blend from achannel that looks like a gray blur is, shall wesay, a foreign concept, even to experts It’sonly when you visualize the Aand Bchannelsjust by looking at the color composite that the fix becomes obvious

I have tried to point out some of the areas

in which retouching in LABis superior tomore customary alternatives If you followedevery example, pat yourself on the back,because most experts would not be able to

For sure, you shouldn’t be upset if you didn’tforesee blending an inverted Binto the Ltokill a moiré

To use LABin retouching, you don’t need aticket Admission is free If you want to havethe advantages that we’ve seen in most of theexercises in this chapter, you don’t have to doanything new Do whatever it is that you’vebeen doing so far—just convert to LABfirst

The Bottom Line

LABhas major advantages in many types of

retouching Color blends are more believable Noise

can be reduced more effectively Certain colors can be

targeted for enhancement Damaged areas are more

easily filled Most painting tools are more effective

Using the dodge and burn tools on the AandB

chan-nels is a more sophisticated alternative to

Photo-shop’s sponge tool

LABis the tool of choice for eliminating moiré

Spec-tacular results can often be achieved by blending with

theAand/orBchannels into the Lin Overlay, Soft

Light, or Hard Light mode

eputation is an awesome thing.LABhas one, and it’s fied: extremely powerful, difficult to learn By getting this far

justi-in a challengjusti-ing book, you’ve justi-indicated that you’re willjusti-ing tomake the effort to study and understand, and as a result allkinds of imaging frontiers are now open

There is, however, another, much larger group of folk:

those who are tempted by what LABhas to offer but want to be able totame it without knowing exactly why or how it works From time to time I drop LABstuff into my magazine columns It always brings forthdemands for more simplification and more step-by-step, as in Chapter 1

In early 2005, I went further, offering a recipe for a spectacular use of

LABthat didn’t even require learning what the individual channels do

Naturally, it involved curves and color variation It was too blunt aninstrument to introduce previously, but it’s worth a look now

In the first half of the book we stuck with one basic form for the

ABcurves Now, we will look at some advanced curving that takes fulladvantage of LAB’s power, but it may make more sense if we start with the demonstration that so impresses the uninitiated

Introducing the Man from Mars

This strategy for the LAB-ignorant involves creating a document with theoriginal image on the bottom layer and an impossibly colorful adaptation

in an adjustment layer on the top, and then cutting the opacity tosomething more palatable We’ve seen this method in action twice, mostrecently in a shot of a bison, Figure 7.9A Our first example, the wild-looking Figure 1.15B, was a man’s head and shoulders I used that same

Command, Click, Control

Advanced LAB curving can be astoundingly effective at driving objects apart without actually selecting them Starting the process requires only

a click of the mouse—and an understanding of why we sometimes need to produce men from Mars.

12

image to begin the magazine piece,

and named the procedure after it,

cal ling it the “Man f rom Mars

Method.” Here’s the recipe,

remember-ing that it’s designed for people who

don’t understand LAB’s structure

• Find an image with flat-looking

color, like the Venetian shot of Figure

12.1A Convert it to LAB

• Layer: New Adjustment Layer>

Curves The L curve appears by

de-fault

• Locate an area that is typical of

the flat color you’re trying to affect—

not too colorful, not too dull

Com-mand–click there, and a

correspond-ing point appears on the curve

• Drag the lower left point of theL

curve to the right, half the horizontal

distance to the point that you clicked

into the curve in the last step

• Turn your attention to the Acurve

and Command–click another point—it

will be called a pivot point from now

on—into the curve

• Drag the bottom left point of the A

curve to the right until the top right

corner stops being a curve and snaps

into a straight line Compare the top of the L

curve in Figure 12.1 to that of the Aand Bto

see what I’m referring to The Aand Bcurves

hug the top of the grid; the Lis still an arc

• Go to the Bchannel and repeat what was

done with the A Click OK, and behold the

Man from Mars, Figure 12.1B, wildly colorful

• Reduce the top layer’s opacity to taste I

chose 22% for Figure 12.1C If you think that

the yellow-sky look of Figure 12.1B is

roman-tic, you could choose a bigger number

Why the Recipe Works

Those who’ve never heard of LABbefore don’t

understand the purpose of these steps, but

they certainly appreciate that there is no

simple way (if one exists at all) of going from

Figure 12.1A to Figure 12.1C in another space The reason is the same striking varia-tion in colors that made the canyons of Chap-ter 1 such easy pickings for LABcorrection

color-The key to success is choosing the rightpoints to click into When we look at an un-tamed Man from Mars version, we need tosee that colors have been driven into all fourcorners of the LABspectrum

In Figure 12.1B, that goal is clearly beingattained, especially in the B At the bottom ofthe image the water has gotten much bluer,but the sky has turned yellow The impact inthe Achannel is less obvious, but it’s there:

the water farthest from us is now green Onlythe laundry in the left foreground can truly

be described as magenta, but the buildings

must have a strong magenta component or

they would not have become so orange

If instead of establishing pivot points byCommand–clicking almost under the bridge,

where the water is a dull blue, we had clicked

at the bottom of the image where it’s bluest of

all, the entire picture would have turned

yellow Similarly, if we had clicked low on thehorizon into the sky, which isn’t as blue as thewater, everything would have gotten moreblue We wouldn’t have neededLABto help

us do either of those stupid things But only

LABlets us blast these very similar bluesapart, making some yellow, some bright blue

Figure 12.1 The Man from

Mars Method The drastic

curves at right are applied to

the original image (opposite

page) on an adjustment layer,

producing, above left, a

wildly colored picture Above

right, the layer opacity is

sharply reduced to produce

a final version.

No one area of this Venetian image was

vastly more important than any other Often,

though, we encounter images with two or

more distinct points of interest that need to

be moved apart from one another If so,

LABcan do it The most efficient way is to

Command–click some pivot points into

the curve The Man from Mars Method

requires only one such point per curve In

the rest of the chapter, the images need

more We’ll start with a simple example

The Invisible Background

Figure 12.2 is a silhouetted but

other-wise unedited digital capture, shot for a

leading outdoorwear catalog This

merchant’s design style of showing

product only, without a background

or a model to cause trouble, makes

life much easier

I see the jacket, which is described

in the catalog as being blue, as too

pur-ple and too dark Moreover, there’s no break

between the body of the jacket and the

shoul-der and waist areas, where there should be a

significant color change, not just because the

two areas are of different colors, but because

our own visual systems insert simultaneous

contrast when we view such things

The action will be in the Bchannel, since

it controls the yellow-to-blue axis We must

increase the distance between the two key

areas Possibly we may do it in the other two

channels as well, but in the Bfor sure

The problem is that such curves threaten

to affect two innocent areas of the image

The first threat is an empty one, just like the

background it covers When the silhouetting

was done in RGB, the background was

deleted to a pure white: 255R255G255B On

conversion to LAB, it became 100L0A0B, and

when it was without further intervention

converted to CMYKit became the similarly

blank 0C0M0Y

Mess with the Aand Bcurves in this image,

and the background may go far enough off

0A0Bto become an imaginary thing of a distinct hue, yet defined as being

color—some-as light color—some-as blank paper If so, when it converts

to CMYK there will be dots where no dotsshould be

But if that happens, it’s no problem —provided we’ve had the foresight to apply thecurves to a duplicate layer and not the origi-nal file The background is easily restoredusing the Lightness slider in layer BlendingOptions, by excluding everything that has avalue of 100Lon the underlying layer

The bigger problem is the jacket’s lining,which is neutral in Figure 12.2 and shouldprobably remain so

With the file converted to LAB, it’s sensible

to begin with the critical Bcurve before ciding what to do with the Land A Start withthree Command–clicks, one taken in each

de-of the three areas we’ve identified The lining

is essentially neutral, a value of 0B; the

Figure 12.2 The body of the jacket needs to be bluer

and more distinct from the grayer color found in the shoulders and waist of the jacket.

shoulder/waist section is higher on the curve at

around (25)B; and the body of the jacket is highest of

all at around (45)B

By raising the top point while lowering the middleone and holding the position of the bottom one, we

achieve the goal The body gets bluer, the

shoul-der/waist section less blue, and the lining stays

the same

After finishing the B, it’s easier to figure outwhat to do with the A In this image, it turns out

that a similar three-point approach works well

The jacket body is more magenta, less neutral,

than the other two points So, just as in the B,

the jacket color can be intensified, the shoulder

color made more neutral, and the lining held

The garment shows little luminosity tion, so no three-point approach is available

varia-in the Lcurve Instead, we simply drop the

quartertone value, making the darker half of

the curve steeper and increasing contrast

throughout the jacket

After a final check of the Info palette to

be sure that the

back-ground will still convert

to pure white when the

file enters RGBor CMYK

(otherwise, use Blending

Options to exclude areas

that are white on the

underlying layer), we

have reached Figure

12.3, and the jacket is

ready to wear

And That’s No Fish Story

LABis an essential part of the toolbox of the

undersea photographer Marine life is often

so brilliantly colored as to baffle any output

device Unfortunately, the background is

often strongly colored enough itself to

dis-tract from colors that have already been

sub-dued by being brought into the CMYKgamut

The clown triggerfish of Figure 12.4 isnoted for its brilliant yellows The impact of

its mouth coloring is substantially reduced by

Figure 12.3 The curves below force the

two outside colors of the jacket apart, while retaining the neutrality of the lining.

competition from yellow areas in the ground below the fish’s belly and, to a lesserextent, on the fish’s own back

back-This picture is unmanageable in RGBor

CMYK without selecting the fish Thosecolorspaces have no way of reducing theintensity of one yellow while increasing that of another In LAB, it just takes a fewCommand–clicks in the Bchannel

The Bcurve in Figure 12.4 has only threepoints other than the two endpoints, butthere should really be four, perhaps five,

Command–clicks, because there are that

many areas of importance in the image

Going from the most neutral to the most

yellow, they and their typical measurements

in Figures 12.4A and 12.4B are

• The white spots on the fish’s abdomen,

80L(3)A2Boriginally, 86L0A(1)Bafter the curves

• The yellow areas on the fish’s back While

these are certainly significant, I did not sure them They appear to fall about midwaybetween the yellowness of the backgroundand of the mouth As those two areas willneed to be driven apart, the yellow on theback will fall into place of its own accord

mea-• The extremely yellow areas around thefish’s lips were 88L1A65Boriginally, 92L4A84B

after the curves

As with our first two images, the idea is todrive a wedge between relatively similarcolors With this fish, we also nod in thedirection of gray balance by trying to keep the spots on the abdomen close to 0A0B.And from that, it’s an easy step to the nextimage, where, in addition to there beingspecific colors that need to be handledseparately, huge areas appear that should

be just about as neutral as the fish’s whitespots are

The Search for the Scapegoat

When a printed job does not meet theclient’s expectations, it can be for a variety

of reasons, such as poor ink densities, adequate control of dot gain, bad pressquality-control procedures, and the like

in-Printers tend to lump these issues underone inclusive technical label, namely, “badphotography.” Photographers, for theirpart, often blame bad printing for theirown foolish color-correction practices It’s

a great spectacle Color reproduction is soboring a topic that we can always appreci-ate any amusement, regardless of source

Culpability for Figure 12.5A is probablyshared Many pressmen would simplyblame the photographer, because, sincepressmen work in the room being pictured,they know full well that it has no yellowcast Photographers might well start blam-ing the pressmen in advance of the nextjob, since it’s hard to evaluate what aprinted product looks like under nonsensi-cal lighting conditions Fortunately, there

Four Tips for the Command-Clicker

Breaking colors apart can be a delicate operation, in which

slight slips cause major problems Here are three ways to

minimize the chances of something going wrong

First, apply the curves to a duplicate layer or adjustment

layer How much to move colors apart from one another is

a subjective, not a by-the-numbers, decision There’s no

disgrace in creating too strong an effect and then changing

the layer to a lower opacity to move back in the direction of

the original Also, this type of curving sometimes creates

problems in other areas of the image By using a separate

layer, you permit the possibility of excluding these areas by

means of layer Blending Options

Second, particularly when the points you have clicked into

the curve start out fairly close together, be careful that you

only adjust them in the north-south direction, straight up

and down Moving the control points diagonally may

defeat what you’re trying to achieve You may find it useful

to move a point by selecting it and then using the up or

down arrow key, which forces the point to move

north-south only

Third, start with the most critical color channel In the

jacket of Figure 12.2 and the fish of Figure 12.4, the key

colors are blue and yellow, so the Bis the critical curve Get

that one right before progressing to the Aand finally the L

Fourth, take advantage of the option to show the curves

grid at a larger size (To toggle between large and small

dialogs, click the icon at lower right.) Ordinarily, the larger

size is a waste of valuable screen space, because we don’t

usually need extreme precision in placing points But it

makes it easier when we do—as in these examples

Figure 12.4 Undersea life is

often brilliantly colored, but

the background can compete

with it, as it does in the

orig-inal image, top The fish’s lips

are supposed to be a vivid

yellow, but the yellow areas

of the background spoil some

of the effect The curves at

right drive the two yellows

apart and result in the

corrected version, bottom.

A

B

are controlled-lighting booths in the nearby

pressroom for just such purposes We hope

Forgetting the politics, we face a gross

yellow cast In Chapter 4, we learned how to

stifle such things by steepening the ABcurves

while moving them to whichever side of

their original center point would serve to

neutralize the cast

That’s not going to work this time, for four

separate reasons

• This image is like some of those shown in

Chapter 7, where the digicam has “helped” us

out by extending the range so far that the

endpoints can’t possibly have a cast because

they would otherwise fall outside of the RGB

gamut The ceiling lights all measure 100L—

which, since Figure 12.5A wasn’t touched

after it left RGB—means that you don’t

need to hear about the other two channels,

because 0A0Bis automatic The shadow areas

within the equipment aren’t quite 0L, so they

have positive values in the Aand B, but not

nearly so positive as the concrete floor and

the metallic areas of the machinery

• I think that when finished, both the

metallic areas and the floor should be nearly

gray However, certain areas of the floor are

around 15B more positive than the

equip-ment is Steepening the Bcurve as a whole

would drive these colors even farther apart,

whereas the correct approach must be to

bring them closer together

• Critical parts of this picture are blue The

yellow cast deadens them to a certain extent

If we just steepen the Bcurve and move it to

the left of the center point, we’ll get a glowing

blue that won’t be appropriate

• The bright yellow areas aren’t nearly as

large as the blue ones, but, being hazard

warnings, they’re important Particularly

that one on the delivery chute in the

fore-ground That ledge is right about shin height

for a six-footer A close encounter with such a

thing at high speed, please believe me, causes

sensations beyond all names of pain Wiping

out the overall cast by moving the Bcurvetoward blue is likely to weaken these yellowsunacceptably

The A channel is almost meaninglessbecause nothing important is either partic-ularly magenta or particularly green The L

may be able to accept a minimal S-shapedcurve, but almost the entire tonal range isalready in use Any drastic move wouldprobably plug the shadows

In such cases, I prefer to start where theaction is, with the Bcurve It may be possible

to steepen the Aafterward, but we won’tknow how it fits into the overall picture untilthe Bsituation is clarified

Therefore, I commenced hostilities byCommand–clicking four points into the B

curve: one each for the yellow hazard signs,the blue equipment, the metallic parts of theequipment, and the yellowish areas of thefloor Because the latter two points wererather close together and therefore difficult toseparate, before starting to move the points

I erased the one for the floor and substitutedanother slightly lower on the curve

I raised both the middle points, but thelower of the two a little more, creating a curvethat looks somewhat like an inverted S I thenturned to the Acurve and Command–clickedpoints for the yellow and blue areas, anddrove them apart slightly

After I added the Lcurve, the new numberswere as follows, from the bluest to the yel-lowest areas:

• The blue areas start at 45L(8)A(11)B andbecome a significantly bluer 46L(15)A(24)B

• The metallic parts of the equipment start

at 63L3A14B and become a nearly perfect gray, 75L1A0B

• The yellow areas of the floor start at

62L7A29B and become 73L8A8B—still warm

as opposed to gray, but much closer to theneutral metals than they were in the original

• The yellow hazard warnings start at

84L(4)A75Band get more yellow, 93L(8)A98B

Figure 12.5 Four points were

Command–clicked into this B curve From top to bottom, they represent the blue parts

of the equipment, its metallic areas, the left side of the floor, and the yellow hazard warnings The curve forces semi-neutral areas closer to one another while allowing blues and yellows to remain pronounced.

A

B

Warm Things Are Doubly Positive

Drastic inverted Scurves don’t work in other

colorspaces A more subdued version is

occa-sionally appropriate when an image has

crit-ical details in both highlights and shadows

and we’re willing to sacrifice midtone

con-trast to bring them out But we can’t make the

center of the curve as flat as the one we just

saw in the B curve of Figure 12.5 without

blasting all detail out of the entire midrange

There’s the LAB advantage again In the

Bchannel, there’s no detail to be harmed,and blue and yellow things land in the samelocation on the curve regardless of how darkthey are The curve shown retained (or evenintensified) existing strongly blue or yellowareas, while neutralizing everything else Wecouldn’t have achieved the same result withthe simpler straight-line curves that we used

to kill casts in Chapter 4

Deciding when to use the complex curveshapes described in this chapter as opposed

to simple straight lines can be ing, so there’s a box on this page reviewingwhy this chapter’s images needed specialhandling In the pressroom image of Figure12.5A, straight lines would clearly not haveworked We could have made the overallpicture a lot more neutral, but certainparts of the floor would have become evenyellower than they were originally

confound-Sometimes, though, it’s hard to choose

The twilight conditions in which Figure12.6 were shot created a coolness that thephotographer could not easily control If

we start Command–clicking into everyarea of significance in this picture, how-ever, carpal tunnel syndrome becomes alive possibility Therefore, on first impres-sion we might wish to treat it in the style

of Chapter 4 The image is somewhat toodark, which can be fixed in the L All colorsare subdued, so we steepen the Aand B

curves, while moving them to the right

of their original center point to makeeverything warmer

The question is, how far to go? There aretoo many unknowns Certainly the facesare not red enough, but we don’t knowexactly how red they should be The man’sshirt is probably gray, but it might also be

a blue The road and car in the backgroundmight both be blue, both neutral, or one

of each The woman’s sweater appears to

be magenta or purple, but as to the preciseshade we have no clue

When to Use This Method

The work in this chapter is more difficult than that of the

first four chapters, where the adjustment curves were

mostly straight lines Both methods break colors apart,

so it can be hard to know which to use

With many pictures, like the canyons of Chapter 1 or the

Venice image of Figure 12.1, straight lines are appropriate,

because no one part of the picture is so important that we

would be willing to suppress some other part Plus, we

want to enhance all, not just some, colors

Command–clicking and adjusting the points up and down

makes sense when we wish to drive two or more specific

objects apart Often, this entails deliberately neutralizing

one area so that it doesn’t compete with another

In the fish image of Figure 12.4, consider three areas: the

white areas on the abdomen, the yellowish patches in the

background, and the extreme yellow around the mouth

Straight lines would create more variation than in the

original, but would do so uniformly: the space between

abdomen and background would increase, as well as that

between background and mouth Instead, we’d prefer to

put extra space between the two yellows—in other words,

to keep the white and the paler yellow at a constant

rela-tionship, or even bring them closer together

The industrial scene of Figure 12.5 does the same thing for

a different reason Straight lines would increase the

varia-tion between the relatively neutral areas and those where

the yellow cast was more violent We’d prefer to keep them

together, and to drive both toward neutrality

To decide whether to use straight lines or a more complex

shape, look on the important objects of the image as

being children If they are all playing happily together,

straight lines will work fine If they are being hostile, and

competing with one another, Command–click away

Nevertheless, certain possibilities can beruled out Warm colors—reds, browns, or

oranges—are positive in both AandB Three

parts of this image must meet that test:

the skin, the faux-wooden bridge, and the

woman’s hair The woman’s sweater is very

positive in the A because it’s far more

magenta than it is green, but we have no way

of knowing whether it should be more yellow

than it is blue The man’s hair could be

brown, but it’s dark enough that it might

be black, in which case it would be 0A0B

The woman’s hair could not It is lighter and

therefore must be some shade of brown

The initial measurements show how faroff the original is The man’s hair is 18L0A(10)B

His face is typically 69L9A(6)B The woman’s

face is a little more yellow, at 65L8A(1)B

In all of these numbers, theAand Bshould

be a lot closer to one another As human skin

and hair, barring visits to the tattooist and/orthe beauty salon, can’t possibly be more bluethan it is yellow, we know that the Bis hugelybiased toward blue By inspection, we seethat the faces aren’t pink enough, so wesurmise that theAneeds to move away fromgreen and toward magenta

To get a hint of how much, we look forother clues The man’s shirt is typically

75L(5)A(20)B, but its B value is quite able—on our left, I’m seeing something like (8)Band on the right more like (25)B Inthe background, the center of the car’s trunkmeasures 44L(6)A(32)B Pavement of aboutthe same darkness is 40L(4)A(18)B

vari-If the car is that much more B–negativethan the pavement is, then it’s a blue car

It can’t be gray, or the pavement would beyellow So perhaps the pavement is blue andthe car even bluer, or maybe the pavement

Figure 12.6 This original image is so heavily biased toward blue that it’s a candidate for the straight-line curve treatment

discussed in Chapter 4 However, Command–clicking control points into the curves and adjusting them up or down may

help bring out variation in key areas.

is gray and the car blue With the man’s

shirt being so variable, there’s no correct

answer But the Bcurve must move at least

far enough to the right to turn (10)Binto 0B—

and that’s only on the dubious assumption

that the man has black hair If it’s brown, we

need to go farther

The Achannel seems like an easier call

I don’t see why any of these three objects

should be more green than they are magenta

If anything, they should be biased in the other

direction So, the minimum move makes (6)A

become 0A.There isn’t any one right way to handle this image with straight-line A Bcurves

The exact angles are therefore up for debate

Figure 12.7 is a reasonable shot—except forone little thing

The colors in the original are subdued—

except for the woman’s sweater, which checks

in at a loud 45L50A(10)B It’s a printable color

in Figure 12.6, but not in Figure 12.7, where

Figure 12.7 Using straight

lines for the A and B curves while moving both to the right of the original center point wipes out the original cold cast while intensifying all colors Unfortunately, the woman’s sweater is driven out

of the CMYK gamut, so detail

is lost when the file is converted for printing.

the curves have basically doubled all Avalues

in the interest of getting pinker faces So, the

sweater is out of CMYKgamut when the file

leaves LAB When that happens, most of the

detail vanishes

Figure 12.8 is the Command–click native There are two interior points in the

alter-A curve One establishes the fleshtones,

the other the assumed neutral areas The

neutrals go slightly toward magenta and

away from green The sk in goes more

strongly toward magenta Meanwhile, thelower left point is raised, suppressing themagenta component of any object that was very magenta to begin with—namely,the sweater

In the B, the three interior points, left toright, are the light parts of the woman’s hair;

the skin plus the magenta sweater, whichshare the same range; and the backgroundpavement The point in the upper right is not a Command–click It’s an arbitrary point

Command, Click, Control 255

Figure 12.8 This version uses

the same general approach as

Figure 12.7, but brings up

the left side of the A curve to

prevent the sweater from

getting more colorful Also,

two points in the top of the B

curve bring the background

car away from the

back-ground pavement.