Photoshop Lab Color- P11

Chia sẻ: Cong Thanh | Ngày: | Loại File: PDF | Số trang:30

Thêm vào BST

Báo xấu

93
lượt xem 6
download

Download Vui lòng tải xuống để xem tài liệu đầy đủ

Photoshop Lab Color- P11: LAB has a reputation for enormous power, yet virtually all reference materials that advocate its use illustrate its capabilities with a single class of image. This chapter introduces the basic LAB correction method and explains why it is so extraordinarily effective. if you happen to have a picture of a canyon.

Chủ đề:

Bình luận(0) Đăng nhập để gửi bình luận!

Lưu

Nội dung Text: Photoshop Lab Color- P11

Once for Color, Once for Contrast 301 have them, a blend won’t change things. If Curving the L channel would help, but we not, the blend by deﬁnition will bring them should try to engineer extra weight into the closer. Thus, the blend isn’t troubled by neu- picture ﬁrst. We’ve done it earlier with skies; trals that are wrong at some darknesses and now we’ll do it with the entire image. right elsewhere. As pointed out earlier, in faces that are Red objects are more common than any normally lit, the green channel is almost other kind. Any such object needs to be always the best. This image is not normally lighter in the red channel than the other two, lit. The face is much lighter than it should but it should have roughly similar values in be, so the rule doesn’t apply, any more than the green and blue. Blending the green chan- it would if the face were in shadow. nel into the blue is usually the best way to cut I made a third layer by duplicating the sec- down a yellow cast. It brings closer together ond, and applied the blue channel, accepting not only the neutral areas, but also the reds. the default choice of the blue from a merged Not here, though. The green channel is version of the bottom two layers. Choosing much lighter than the blue in the priest’s the bottom layer would have been a mistake, green chasuble, which plays a big role in the because its blue was so dark that it didn’t image. If he were wearing some other color, carry detail in the green chasuble. Once light- I’d use the green for the blend. As matters ened with the ﬁrst blend, it became more stand, I’ll go for the red channel. usable for the second. I chose 60% opacity To get Figure 14.11B, therefore, I created a and, of course, Luminosity mode. That’s new layer and used the Apply Image com- Figure 14.11C. Then into LAB, being sure to mand to blend the red into the blue, picking ﬂatten the image only after the conversion 32% opacity out of a hat. I used Lighten was complete, and you know the rest of the mode, which restricts the blend to areas in drill. Figure 14.11D is the ﬁnal version. which the red in fact starts out lighter than the blue. Here, that comprises the entire A Vote for Fluency image, so Normal mode would have done This picture is not of an actual church the same thing. But we certainly wouldn’t ceremony. Rather, the priest is posing for a want to darken the yellow in some hypothet- professional photographer using top-notch ically bright blue or cyan object, so being in equipment. Everything is as controlled as it Lighten mode is good practice. can be, except that it would take either divine With washed-out originals like this one, if intervention or a good knowledge of the LAB you do a lightening blend to try to regularize language to compensate for the ambient the color, usually you should set the layer lighting. According to the photographer, this blending mode to Color, meaning that Photo- scene featured not just a mix of ﬂuorescent shop is to pick up the detail from the bottom and incandescent lighting, but also flood- layer and use the top one for color only. I lights of two different colors. didn’t do that, because I needed a lighter In short, the picture is a tougher customer layer for the next move. than it looks, because it’s full of casts and The biggest issue I have with the original counter-casts that are disguised by overall is that the face scarcely exists. The whole lightness. Yet two blends that together took picture needs to get darker to make any kind less than a minute sufﬁced to get it on the of credible ﬁx. I don’t mind if the ﬂowers or right track. Figure 14.11C may not be attrac- the rear wall lose a little color in the process, tive itself, but it’s easy to ﬁx in LAB. The orig- because I think the priest is the focus. inal, Figure 14.11A, is another story.
302 Chapter 14 The secret is to think in the foreign lan- with one another (and with LAB correction guage of LAB even while speaking the RGB generally) in a bewildering variety of ways. idiom. Like a new language, you have to In deciding when to use them, remember speak it slowly at ﬁrst. We’ve only done four that certain things work about the same in images in the course of a 20-page chapter. RGB and LAB, and other things are quite With practice, it becomes very fast. different. Those radical differences are the And, like a foreign tongue, knowing a few ones you have to cater to. Trying to create phrases and grammatical constructions gives extra detail with curves or some similar only a little hint of how rich and rewarding command may work slightly better in one the fully realized language can be. By neces- colorspace or another, but you’ll be able to sity, this is a brief overview. We had space for fake it elsewhere if you have to. only one example of each of the four major The color-busting capabilities of the A and categories: using RGB curves ﬁrst to change B channels, though, can’t readily be faked in contrast only and then to change color only; RGB. And the ability to use a particularly and then using RGB channel blending in the good RGB channel to strengthen an image same two ways. These techniques interact can’t be faked in LAB. Accordingly, three ﬁnal pieces of advice. First, when you open a new RGB ﬁle, start by The Bottom Line checking out the channels to see if blending is LAB’s signature is the separation of color from indicated. If you don’t, you may be passing up contrast. It sometimes pays to do the same thing in an opportunity that won’t be available later. RGB: to prepare one version to be used for detail only, Second, if you have decided to work on the and another to be used only for color. Most of the time, doing so adds several steps and doesn’t deliver color in RGB, just try to eliminate the impos- as good quality as LAB would. sible. Don’t worry if the image temporarily There are, however, exceptions, where it’s worth the looks too dull. LAB can give you far more extra time to do some of the contrast-only or color- color than you’ll ever need. only work in RGB. Particularly, blending of RGB chan- A picture does not live up to its full poten- nels can produce certain effects that LAB can’t. This chapter discusses how to identify the situations where tial unless both contrast and color are all they RGB work in the LAB style can create better results can be. Keep the two concepts separate in than LAB itself can. your mind, and you’re well on the way to getting them to that happy point.
Blending With the A and B 15 The ﬁnal major use of LAB is as powerful as it is counterintuitive— using the A and B channels, which contain no detail, to blend into the L, which does. It’s a radical way to separate objects out for treatment based on their color, and it’s also the easiest way to experiment if you’re not sure how much more color an image needs. mbitious chefs, like aspiring Photoshop authors, think they have to demonstrate their creativity by coming up with dishes that nobody has ever seen before. In both cases, the tricks can be carried to ridiculous extremes. Some years ago, it was fashionable for restaurants to marry up the most ridiculous pairs (or ménages à trois) of ingredients that they could dream up. I’ve had the dubious pleasure of eating liver with blueberry sauce; oset- ra caviar with wasabi on a tortilla; and, most recently, sausage ice cream. None of these combinations sounds even a fraction as ridiculous as the one that’s the subject of this chapter. We have just spent 300 pages emphasizing that the A and B channels contain no contrast information, only color; that they can never be portrayed as white, black, or anything even close because those would represent imaginary colors; that any detail that shows up in them is probably a defect that needs to be blurred away; and that they are, in short, dull, gray blurs. Blending such monstrosities into a contrast-bearing channel like the L sounds roughly as appetizing as mixing 25-year-old Macallan with Diet Coke. Figure 15.1A is the original. Figure 15.1B shows what happens when you try replacing the L with a copy of the A. In Figure 15.1C, the L is replaced by a copy of the B. To make these nearly contrast-free versions, with the L channel open I chose Image: Apply Image, Normal mode, 100% opacity, with the A designated as source for one and the B for the other. Nauseating enough, but check out what comes next. In Figures 15.1D through 15.1G, still at 100% opacity, I changed the blending mode to Overlay. In two of these four variants, I also checked the Invert box in the Apply Image dialog.
A Figure 15.1 Left, the original. Bottom left, its L channel is replaced by a copy of the A, and bottom right by the B. Oppo- site page: four images, in a random order, show blends of the A or B into the L not in Normal but in Overlay mode. In two of the versions, the A or B was inverted during the blend; the other two are standard overlays. The A and B themselves never changed in any of the seven images, so the colors should always be the same. All variation is in darkness. Can you pick out the images made with overlays of the A, the B, the inverted A, and the inverted B? (Hint: refer to the two images below to see how darkness is being allocated in the A and B channels.) B C Before discussing how such a weird con- where they intersect the greenery. In the A cept works, let’s ﬁrst admit that something and B, all three objects are the same because quite signiﬁcant, and potentially quite useful, all are neutral, and they break away from the is happening, and it’s something that would colorful background naturally. be difﬁcult to achieve outside of LAB. • Nor can you base a selection of the green- ery on any of the RGB channels. The darkest, When Zero Equals Fifty Percent most neutral areas would be inseparable We saw in Chapter 9 how selections and from the darkest greens. They would also be masks that would be too complex to make in more selected than brighter greens. That’s the RGB can often be found easily in LAB. And so opposite of what’s needed, and the opposite it is here. The sky, the lake, the man’s face, and of the way the A and B channels do things. the greenery are each clearly deﬁned in both • Even if you could make the selections in the A and B—but not in any RGB channel. RGB, you still couldn’t use them for the types Even if they were, we couldn’t make the kind of trickery shown in this chapter. Those tricks of mask that the special purposes of this require that neutral colors fall in the center of chapter require, for three reasons: the blending channel—that is, values of 0A • In RGB, the man’s jacket, pants, and hair or 0B—whereas in RGB channels neutrals fall are three different tonalities in each channel, wherever they like. and would have to be selected out manually The type of blending we’re about to discuss
Blending With the A and B 305 D E F G offers extraordinary ﬂexibility to remake the model that where the overlaying channel is image by adjusting the darkness of important lighter than 50% gray, it lightens the under- objects. It can, in effect, completely change lying one; where darker it darkens; and where the way the image is lit. it’s exactly 50% gray, nothing happens. The An inﬁnite number of variations can be other related blending modes would create conjured up. I like Figure 15.1D in principle, results that were similar but not identical. but think that it’s gone overboard. So, I would Consequently they don’t pertain to the basic use a lower opacity when making the Overlay concept. The text of this chapter will use blend. In Figure 15.1G, I’m ﬁne with what Overlay only, but there’s a box on Page 307 for happened to the lake, but the greenery is those interested in the other modes. much too light for my taste. I would therefore The A and B work so well in Overlay mode blend the original L channel back into the precisely because values of 0A and 0B—neu- new L using Darken mode, maybe at 100% trality, in the AB language—correspond to a opacity and maybe not. 50% gray, not a white. Therefore, anything Overlay is one of several modes using the neutral won’t change. The man’s grayish
A clothing is almost identical in all four of the overlay variants of Figure 15.1. Where there’s color, it’s an- other story. This B channel has more decisive color breaks than the A. Anything more blue than it is yellow is B–negative, darker than 50% gray for overlay purposes. The sky is blue, therefore dark, and the lake even bluer, there- fore darker. Anything more yellow than blue is B–positive, lighter than B 50%. The face is somewhat more yellow than blue, and the greener y is way more yellow than blue. So, when the B is applied as an overlay to the L , colors don’t change, but darknesses do. The greenery gets much lighter, the face slightly lighter, the sky slightly darker, and the lake much darker. Do you see which of the four variants was made this way? By checking Invert when doing the overlay, we reverse the effect. Which of the four has the lightest lake and the C darkest greenery? That one was done by overlaying the inverted B onto the L. The two done by overlaying the A onto the L are harder to tell apart because nothing is either strongly magenta or Figure 15.2 Overlay mode in action. Top, the original. Center, in LAB, the original is applied to itself in Overlay mode. Bottom, the L from the top version is married to the AB from the middle one.
Blending With the A and B 307 strongly green. Nevertheless, the face is Constantly tweaking both ends of a curve slightly more magenta than green and the is annoying. It’s easier to have something shrubbery more green than magenta. The excessively colorful like Figure 15.2C on the two objects move in different directions after top layer, and adjust the opacity slider to the overlay, unlike the version done with taste. I almost always use this method when the B, where both face and greenery were dealing with face shots. more yellow than blue. If the A is overlaid This summarizes the two main reasons to uninverted to the L, the face gets lighter and do the improbable by using the A or B as a the greenery darker; inverting does the oppo- blending channel. We can create different site. Personally, I think that both versions effects in the L by altering the darkness of cer- done by overlaying the A channel are better tain interest objects, and/or we can enhance than the original. color by overlaying the AB onto themselves on a separate layer and then dialing down the The Easy Way to Brighter Color opacity if need be. The rest of the chapter will To hammer home how the mode works, rather than apply a single channel to another, I’ve applied Figure 15.2A to itself in Overlay The Overlay Sextuplets mode, 100% opacity, to produce Figure 15.2B. Overlay is one of six blending modes in which 50% Applying an entire image to another is not gray has no effect on the underlying image, lighter a composite process, but rather a channel-by- areas lighten, and darker ones darken. All six are channel one. Since this is an LAB ﬁle, the L is computed in different ways. Some try to protect extreme values from blowing out, whereas others overlaying the L, the A the A, and the B the B. permit it. Some have relatively strong impacts and Figure 15.2B’s shortcomings are the fault of others don’t. the L. Things that used to be light got even If you just stick with Overlay for AB blending, you’ll be lighter, blowing out all the detail on the white ﬁne. The key is knowing when this type of blending is clothing. Things that used to be dark got appropriate, not which of the sextuplets to use. Or, darker, plugging up the motorcyclist’s helmet when you set up the blend, you can toggle back and and jacket, among other things. forth to one of the other modes and see if you like the To correct this, I replaced this L with the result better. original from Figure 15.2A. In Figure 15.2C, For insatiably curious people, here’s the executive consequently, the only change is in the A and summary of what the other options have to offer. B channels. And, garish though it may be, it •Vivid Light, Linear Light, and Pin Light are, as far as I carries a considerable temptation. know, useless in this type of blending. Overlaying the A and B onto themselves •Hard Light and Overlay give identical results when can’t introduce a cast, since all values of 0A applying the A or B to themselves, but not when or 0B are unchanged. Instead, it drives all applying to the L. Hard Light extracts extra detail colors apart, much like the curves featured in from the highlights and shadows, if that’s what you’re after. Chapter 1—with one important exception. AB curves are powerful but not particu- •Soft Light has less of an impact than Overlay. It slightly favors darkening rather than lightening. larly amenable to experimentation. I think When applying the A or B to itself, Soft Light accents the colors of Figure 15.2A are basically right subtler colors more than brilliant ones. Also, while it yet lukewarm, particularly in the faces and does increase the intensity of warm colors, it accentu- hair of the people on the right side. But I don’t ates cold colors more. You may ﬁnd these attributes know how far to go with them. I’d like to be desirable in certain images. able to do some ﬁne-tuning.
308 Chapter 15 show examples of both uses. There will be no In RGB, we’d use the Hue/Saturation or use of curves, ﬁlters, or selections. Selective Color command to lighten the sky. Our appetizer will be Figure 15.3, which is It’s somewhat difﬁcult (we would have to se- to appear in a book titled Palm Trees: A Story lect both Blues and Cyans in the dialog, as the in Photographs. That settles any question bottom part of the sky will not lighten enough about what the important part of the picture if only Blues are selected); it doesn’t retain a is. Unfortunately, the lighting was unco- good gradation from the top to the bottom of operative. The sky is an impressive deep the sky; and it takes longer than the 10 sec- blue, particularly at the top of the image. That onds needed to produce Figure 15.4 in LAB. would be great if the book were about skies, We’ve already seen how to do this one, in but it isn’t going to cut it when we’re sup- Figure 15.1D, one of the variants of our ﬁrst posed to be concentrating on the trees. image. We have only to decide that every- thing in this image that’s more blue than it is yellow needs to be light- ened, and the bluer, the lighter. The B channel can do this, pro- vided we invert it when we apply it to the L in Overlay mode. If we left it uninverted, it would darken all blues, the opposite of what’s needed. Note the slight darkening of the path, the people, the beach, and the greenery, all of which have a slight bias toward yellow as opposed to blue. To me, these things are all desirable in this image’s context; if you disagree, they could be removed or lessened by blending Figure 15.3 into 15.4 in Lighten mode. If you feel that Figure 15.4 isn’t colorful enough, you might wish to hit it with AB curves to make parts of it more yellow. Using AB overlays as in Figure 15.2 wouldn’t be helpful. Given how light the sky now is, it can’t be made any bluer. The inabil- ity to print good blues is a sad fact of life in CMYK. Our next example deals with an- other gamut issue, one that requires both of the types of AB blending that we’ve been discussing. Figure 15.3 The image is to appear in a book about palm trees, but the sky is so dark that it detracts from them.
Blending With the A and B 309 Of Blues and Butterﬂies the changes are entirely the result of AB Intense blues are hard to deal with no matter blends. Here are the steps. what the output destination, desktop or • The flowers start out much more ma- press. Other dark, rich colors are generally genta than they are green, and slightly more achievable. Pastel colors are problematic, blue than yellow. The leaves are moderately especially on press. A patch of solid magenta more green than magenta and sharply more ink, for example, like the one printed way yellow than blue. Overlaying the L with either back in Figure 1.11, is probably so intense as the B or an inverted copy of the A will darken to be out of your monitor’s gamut. Yet lighter the ﬂowers and lighten the leaves. A respon- magentas fall ﬂat in CMYK, victims of a paper sible person would probably try both ways that isn’t particularly white. If that paper and possibly combine them. Me, I thought in is covered up by a lot of ink, it won’t matter. terms of a gross move to the ﬂowers and a But if the desired color is light, ink coverage must be light also. A lot of that ugly, contaminating paper will show through. This problem is a common one in images of flowers, for which light magenta or pink is a favorite color. Experienced CMYK practitioners cheat by making the color darker and, often, redder, in an effort to put more ink on paper. Even if gamut were not a consid- eration, I think that the ﬂowers in Figure 15.5A are too light, particu- larly in comparison to the dark green leaves. I ﬁnd the colors weak in both, and the beautiful butterﬂy that should be a focus of attention is scarcely there at all. Figure 15.5B is more like what I think the picture should be—and Figure 15.4 Applying the B channel, inverted, to the L in Overlay mode lightens all areas that are more blue than yellow.
A B
Blending With the A and B 311 slight one to the leaves, so I made a duplicate it turned out, I thought the layer was ﬁne as it layer and applied the inverted A to the L in was and left it at full opacity throughout. Overlay mode at 100% opacity (Figure 15.5C). • On a new layer, I overlaid the A and B on The reason for the layer was to leave room for themselves, sharply increasing color intensity a possible change in opacity or mode later. As everywhere, but not changing anything of C D Figure 15.5 Opposite page: the original and a version corrected only with AB blends. This page: top left, an inverted overlay of the A channel into the L. Top right, the A is then overlaid onto itself at 100% opacity and the B onto itself at 75%. Bottom left, the opacity of version D is reduced to 60%. Bottom right, two further overlays of the B channel onto itself. The Blend If sliders at right then restricted the brightening effect to the butterﬂy. E F
312 Chapter 15 value 0A0B. I thought there was a danger of ﬁrst overlay had already put more distance the ﬂowers becoming too blue and the leaves between the butterfly and neutrality) and too green, so I overlaid the B at only 75% then still another, winding up with around opacity, instead of the 100% I used for the A. 25B in Figure 15.5F, where we had opened I had now reached Figure 15.5D. with 3B. There was no point in doing anything • Overlaying the AB channels onto them- to the A channel because the butterﬂy was selves at such high opacities usually creates at 0A throughout and therefore would be something too vivid, as it did here. That’s not unaffected by any overlays. a problem, but an advantage. We’re not look- • Finally, Blending Options (review Chap- ing for magic numbers or secret formulas. ter 9, if necessary) excluded everything but We’re trying to make the picture look good in the butterﬂy. I needed only one slider (A, This a trial-and-error way. We always do the over- Layer) to do it. It could have been done in the laying on a separate layer, allowing us to B as well, by bringing the left-hand slider move the opacity slider until we get the look close to the center point to exclude the ﬂow- we want. In Figure 15.5E, I used 60%. ers, which Figure 15.5F shows are more blue • The butterﬂy is a special case. In the orig- than yellow. Moving the right-hand B slider inal, it’s only the wispiest of yellows. No curve toward the center would exclude the green- could make it as yellow as I want without ery, which has more of a yellow component also making anything neutral (such as the than the butterﬂy does. Reminder: the reason black stripes) yellow as well. Fortunately, it’s to use the This Layer slider is that this top easy to separate insect from background layer has longer ranges in the areas that later. Therefore, having already made one interest us. Its flowers are very blue, and overlay of the B to itself, I made another therefore are further away from the center (more than doubling the effect, because the point of the slider, making it easier for us to ﬁnd the exact point where they begin. Doing it with the A, however, is even easier. Fighting Blend-Induced Grain The butterﬂy is found at dead center because The A and B channels can be grainy enough to create it measures close to 0A throughout. Every- artifacting in certain types of blending. The usual thing else in this picture has a distinct color. danger zones are relatively dark areas that are being drastically lightened, such as the sky in Figure 15.4, So, just bringing both sliders in close to the or where there are repeated applications of the same center point sufﬁces to isolate the butterﬂy. channel, as in the butterﬂy of Figure 15.5F. The best weapon against this effect is the Surface Blur The Art of Selective Lightening ﬁlter (Photoshop CS2 and later), applied before the When a picture is too dark, lighten it. When a blend. If Surface Blur is not available and the purpose picture is too light, darken it. of the blend is to enhance color, using another blur- Simple advice, but effective. If you intend ring or noise-reduction ﬁlter afterward should work. If to follow it literally, you don’t need this chap- the blend will be to the L channel, though, caution is ter. But sometimes the rule, obvious though needed. The advantage of Surface Blur is that it has a limited effect on edges. If you Gaussian blur a channel it seems to be, isn’t quite right. Sometimes before blending it into the L, it may cause haloing the picture is too dark and we only need to around some objects. The Dust & Scratches ﬁlter can lighten parts of it. be a better choice. The most notorious such situation occurs For further discussion of Photoshop’s blurring options, when some knucklehead shoots a picture of see Chapter 5. a person whom the sun is directly behind. Any time the light is to the rear of the subject,
Figure 15.6 Images that are partially in shadow can be treated effectively with overlay blends. however, will do. The Image: Adjustments> stay the same. If we overlay the B, everything Shadow/Highlight command (Photoshop will lighten except the jeans, which will get CS and later) is a good way of handling darker. Which should we use? such atrocities. Different considerations apply when a face is in partial shadow, as in Figure 15.6. Before Don’t Believe What You See discussing other options, here’s the obvious When overlaying the A and B onto themselves to pep one. Figure 15.7A is the Shadow/Highlight up colors, don’t be deterred by what seems to be a loss command’s default settings applied to the of contrast. The idea of this blend is to go well beyond RGB original digital ﬁle. the intensity that you really want, and then dial it back with the opacity slider. If you are using 100% opacity Let’s now set that version aside for later for these blends, you may be creating colors well comparison, and start again with Figure 15.6. beyond what your monitor is capable of displaying Certainly the face is too dark. Being red, it’s accurately. If such out-of-gamut colors are there (and positive in both A and B, so either channel there’s no way to know other than to look for the could be overlaid onto the L to lighten it. undesirable effect), detail will be lost when they are The shirt and the background are green, so converted into the monitor’s version of RGB, just as it they’re positive in the B but negative in the would be if the ﬁle were converted into CMYK. A. The jeans are negative in the B and near When you reduce the opacity of the out-of-gamut neutral in the A . If we overlay the A onto layer, bringing the colors back to something more monitor-friendly, the missing contrast will return. the L, the face will lighten, the shirt and the background will darken, and the jeans will
314 Chapter 15 When I assigned this image as a competi- better, because it would make the greenery tive exercise in one of my advanced classes, it darker, which would contrast with the light- seemed so clear to me that the A would be ening of the face. This theory is exploded by better that I didn’t bother to investigate the Figure 15.7C, whose real attraction is how it alternative. Now that my own entry into the lightens the leaves to bring them out from the contest has gotten thrashed, something that background. The pattern in the shirt is also irritates me no end when it occurs, I take the more pronounced, because its green areas view that when in doubt we should try both are getting lighter while the more neutral ways and trash the one we don’t like. To that parts are not. end, Figure 15.7B is the A applied to the L in For these reasons, I favor it over Figure Overlay mode, 100% opacity. Figure 15.7C is 15.7B, but either one will eventually be better the same thing, substituting the B for the A. than the Shadow/Highlight version of Figure In judging these two, forget that the face 15.7A. The two LAB versions can’t be directly is lighter in one version. That happened compared to Figure 15.7A yet, though. When because the B is more positive than the A. If an image is too dark, it’s usually too gray also. you want the two versions to match, you can When operating in RGB, S/H therefore, quite apply the A a second time, or cut the opacity correctly, adds saturation to all colors by of the version produced with the B. default. The color enhancement step in LAB My thought was that Figure 15.7B would be has yet to come. Figure 15.7 Below, the Shadow/Highlight command applied in RGB to Figure 15.6. Opposite page: top, the A channel is applied in Overlay mode to the L of a fresh copy of Figure 15.6. Bottom, the B channel is applied instead. A
B C
316 Chapter 15 Because this picture is so full of shadow, it ﬁle would then be calling for an imaginary doesn’t need the color boost we saw in Figure color, which would convert to something 15.5, but it does have to get a bit lighter and other than white when we went to RGB or brighter. I found a revoltingly lazy way to CMYK. And if the deepest areas of the shad- make it happen: I blended Figure 15.7B into ows lose detail, it isn’t such a big deal. 15.7C in Overlay mode, opacity 30%, produc- In comparing Figure 15.8 to Figure 15.7A, ing Figure 15.8. the S / H version, note that the faces are Overlaying a ﬁle onto itself is the same as equally dark—that is, if you average both the overlaying each channel onto itself. Doing so lit and shadowy areas, which are more pro- to the A and B increases color intensity. Over- nounced in the LAB version. On the whole, laying the L onto itself is rarely recommended however, Figure 15.8 seems darker, but much because of its tendency to blow out highlights more contrasty. Its portrayal of lights and and plug shadows. In Figure 15.6, neither is shadows is more realistic. much of a factor. The lightest areas are the various reﬂections, none of which are neu- Separating Greens Conventionally… tral. Doing these overlays in RGB might blow In the foregoing example, the pepped-up these areas out to blank white. It can’t happen color could also have been gotten by curves, in the L, because even if though perhaps not as quickly. The blend the overlay forces 100L, it into the L can’t readily be duplicated. can’t force 0A0B. The LAB To help understand why, let’s end with an Figure 15.8 Figure 15.7B is overlaid onto Figure 15.7C at 30% opacity.
image somewhat similar to the one of the bow hunter we just worked on. This time, instead of competing with the Shadow/Highlight command, the opponent is a version done in the famil- iar LAB way with curves. Figure 15.9, taken deep in a forest in Switzerland, is typical of images taken deep in forests. It’s simulta- neously too dark and too light, as the camera fails to make the adjustments that come so naturally to the human visual system. The colors are too muted for the same reason. Above all, the greens are too consistent because the camera lacks the sense of simultaneous contrast that we humans use to break colors apart from those of neighboring objects. This, as much as a canyon, is the classic type of image on which LAB cor- rection works much better than any alternative. The curves-only version is straightforward. The pic- ture consists only of light and dark areas, with a whole range of tones miss- Figure 15.9 Either curves or AB blends to the L channel can lighten this forest, but ing in action. The water the results of the two methods will show aesthetic differences. never gets darker than 90L; the forest never sures close to 0A0B everywhere, so we con- lighter than 65 L. These ﬁndings call for an clude that the picture has no inherent cast L curve in the shape of an inverted S, steep that would cause us to move the center in the lightest and darkest parts, ﬂatter in points of the curves as was done in Chapter 4. the middle. Straight lines in the AB curves would have As for color, this picture lacks prominent sufﬁced, but I made two slight adjustments. I interest objects that need to be driven apart, thought that the foreground leaves were so we have no need for the Command–click threatening to get too yellow, so I placed a trickery of Chapter 12. Also, the water mea- locking point at 0A and then twisted the
318 Chapter 15 Figure 15.10 A version corrected only by curves. upper par t of the A cur ve to intensif y greens. I did not wish to simply steepen the entire cur ve f ur ther because I felt that the foreground rocks were getting too red and did not need more magenta. In the B , I did use a straight line, but I made an executive decision to shift it very slightly to the left of the center point, to add a hint of blue to the water. As nothing in the image is known to be truly neutral, shifting the image toward blue breaks no rules. These curves produce Figure 15.10. On to its curveless competitor. …and Unconventionally Starting again from scratch with Figure 15.9, the forest can be lightened by over- laying one of the color channels onto the L. As all natural greenery is A–nega- tive and B –positive, we could blend with an in- verted copy of the A or an uninverted copy of the B. I chose to use the B, mostly because the greenery was more positive in the B than it was negative in the A , which makes the moves more obvious and easier to control. Also, as the fore- ground rocks are somewhat red, an A overlay would
Blending With the A and B 319 darken them and a B blend lighten them. The fall, however. So, I took the liberty of adjust- latter option seems right in the context of a ing the highlights with the Shadow/Highlight picture that’s getting lighter overall. command, completing the journey to Figure To create Figure 15.11, I ﬁrst blended the B 15.11 and also our discussion of this weird channel into the L, Overlay mode, at 100% method of blending. opacity (Figure 15.12A). The version done with curves only has a As this did not lighten the foreground different flavor than the one done with leaves as much as I wanted, I repeated the blends. It carries better detailing in the tran- move, overlaying the B onto the L again, this sition to the deep shadows. It’s a more time at 50% opacity (Figure 15.12B). Hav ing lightened the image enough, it was time to turn to color. I overlaid the A onto itself at 100% (Figure 15.12C). By the application of such a drastic move to the A without doing anything to the B , the picture had become unbalanced. But overlaying the B onto itself at 100% seemed to make the image too yellow. As a compromise, I did B to the B at 80% (Figure 15.12D). Still thinking that con- trast had to be added between the lightest and darkest areas of the forest, but constrained by my earlier promise not to use curves, I applied the L to itself in O verlay mode (Figure 15.12E). I did not object to the ensuing loss of detail in the shadows, think ing that it helped add depth by emphasizing the relative lightness of the foreground branches. The move blew out the water- Figure 15.11 This alternate correction employed no curves, but rather a series of blends.
A B C Figure 15.12 The progression of steps from Figure 15.9 to Figure 15.11. Top left, the B is overlaid into the L at 100% opacity. Top center, the B is overlaid again into the L, this time at 50%. Top right, the A is overlaid onto itself at 100% opacity. Bottom left, the B is over- laid onto itself at 80% opacity. Bottom center, the L is overlaid onto itself at 35% opacity. The objects that we are trying to lighten or darken are invariably the most colorful ones. We don’t want to restrict the effect only to the most colorful areas, however. We simply want it to be less in areas that are similar to the target object but not as strongly colored. D E Note that between Figures 15.12A and 15.12 B the foreground branches, our target, are lightened conservative take on the image. The blend more strongly than anything else, yet other version is dramatic. We seem to see a deeper green areas are lightened somewhat. picture, with more distance between the fore- In the A and B, the more colorful the ob- ground leaves and the deep shadows at rear. ject, the lighter or darker it is with respect to a medium gray, which represents a neutral. Whites and Blacks Are Both Grays The key is that all neutrals are of the same To review why overlay blending with the AB tonality. It seems ridiculous to group the wa- works (and why it doesn’t work in RGB or terfall with the darkest part of the shadows, CMYK ), let’s rehash the channel structure but that’s how it is in LAB, as Figure 15.13A one last time. demonstrates.