Untitled Document

When I visited the recent Curvemeister Challenge (woman and child) I was surprised to note that many contestants selected areas of the woman's white T-shirt for a highlight and that in some cases large parts of the T-shirt were totally blown out. In Dan's books I have learned that first and foremost, you must establish the lightest and darkest points in an image but I don't think Dan ever said that one should automatically click the white eyedropper tool on the lightest white object in the photo nor that the lightest point *had* to be white-white (5c2m2y), Dan was always careful to mention that the chosen highlight must be white (and neutral) but he also implied that it is quite possible that a white object or substance might not be white in a given photograph, like snow at sunrise or at sunset or in the shade. The subjects (woman and child) are obviously standing the shade and the whites of the T-shirt are not generally white.

In Professional Photoshop 3 Dan wrote : "In each uncorrected image, we must find the highlight — the lightest white — if there is one." [p78 PP3] In this image there is no area that should realistically be made into 5c-2m-2y unless you decide to not represent what the original must have been had the photo not been "underexposed (?)".

Artistic license is OK, Dan wrote: "...it must be something that we are willing to represent to the viewer as being white." [p46 PP5] Dan also wrote "A common problem (...) is to look for "the" neutral point of the image. There may not be one. In Figures 3.2 and 3.4, there's nothing lighter than a shadow that we can guarantee is grey or white. So we don't force neutrality into anything." [p31 PP6] In the image there is something that we know *is* white but it is under a bluish-greenish shadow and the lightest point is not and should not be artificially forced into neutrality (unless you are willing to represent the image that way).

In Photoshop Lab, Dan writes: " Full range means that the lightest and darkest significant points of the image get handled appropriately ... we generally have to be sure that the highlight and shadow are also neutral..." and " if we decide that the highlights and shadows should be something other than neutral, we have to scratch our heads to come up with different numbers." [p45 PLC] In the image of woman and child a "realistic" approach (as opposed to artistic) does not, in my opinion, allow a neutral highlight of 5c2m2y anywhere in the image, so I scratched my head and came up with 11c5m3y (+/-). Andre Dumas

___________________________________________________________________________

I'm not sure what the question is, but I am familiar with the image. :-)

You are correct in saying that the shirt in reality was most likely casted when the shot was taken - the trees and sky and environment would influence the actual color. That does not mean that the shirt cannot/should not be made nuetral when color correcting. As Dan has also brought out in the aforementioned books, the eye can play some tricks on nuetrality that cameras don't pick up on.

I personally used the shirt as my highlight, but I would not recommend blowing out the shirt when setting it. It is not uncommon, however, to blow out unimportant areas to enhance contrast. Too many retouchers make the mistake of using Levels to set the highlight *just* before anything gets blown. You never get that contrast back with an S curve.

J Walton

___________________________________________________________________________

Andre writes,

On the assumption that this is the lightest significant area of the image (I haven't looked at it, but I don't need to), this number is not technically correct.

If you were convinced that the shirt was to be white, then you would set the point to be 5c2m2y. You are not convinced of this, as is your right; you believe that it's slightly blue.

Fine, you may be correct that it should be blue and in that case treating it as neutral would create a yellow cast. However, in making it blue you should try not to make it darker than it would be if you had decided it was white. So, if 5c2m2y would be your aimpoint for a white highlight, then a blue highlight should be more like 6c3m0y. Having it as heavy as 11c5m3y flattens the image unnecessarily. You should be able retain your desired color by choosing something lighter, but still blue. And you will gain contrast.

Dan Margulis

___________________________________________________________________________

Thanks Dan, well I *was* convinced that her shirt was white, but I felt that in the shade, the shirt should be slightly bluish, now I looked at it again and I found I could set the lightest point at 5c2m2y on the woman's shoulder and it did improved the image. The T-

shirt retained a 11c9m5y bluish-grey cast in the areas adjoining the highlight and that, I feel, is realistic since all of this is in the shade, at least this is how the camera would see things (?)

If I remove the blue cast on the white shirt the image changes and that might be considered an improvement, but personally I was more concerned with achieving reality than improving the appearance of this image. Many contestants have chosen to make the image warmer by remapping the blue curve (I think), it makes for a more colorful image and I agree with this approach also but when we modify one color and not the others are we not creating something *unreal* ?

On second thought, when an observer enters that scene, his eyes adapt to the shade and the brain remaps everything to lighter shades, would you say? Then would the white shirt now appear whiter (less bluish-gray) than it would to an onlooker standing beyond in the sunlight? This might have been the better approach but by redoing the white balance that way the image takes on a warmer tone which, seems to me, is not realistic.

Andre Dumas

___________________________________________________________________________

Andre,

Seems to me you're asking how we should see, that is, like the camera's capture of the color cast(s) or our eyes which can adjust and/or ignore the cast(s). That depends on the intent of the image, what it's seen next to and your own preferences.

With regard to the adjustment I usually approach the white highlight setting in one of 3 ways; (1) preserve the cast (2) take out half the cast or (3)

neutralize the cast.

Lee Clawson

___________________________________________________________________________

There may seem to be only two variables, but really there are three:

1) What the camera saw;

2) What the human observer saw;

3) What the human observer remembers seeing upon sitting down to color-correct the file some weeks later.

The most obvious example of where #2 and #3 can be different is in the case of walking outdoors in a place of great natural beauty. Any trees that are in the area are perceived as the same color by methods #1 and #2. However, upon returning home (method #3) the human remembers that the colors were greener.

We humans always see white colors as white. However, as we have more time to recollect what we saw, it may seem to us that what we saw had a warm or a cold cast. Warm casts are generally preferred, but there are some exceptions.

So, yes, if we saw the scene everything would definitely be lighter and more neutral, but it has been a long time since we saw it, so we are free to revise reality within Photoshop.

Dan Margulis

___________________________________________________________________________

--- Lee Clawson wrote:

Thanks Lee, I guess you're right, I am really asking "how should we see". In the full size image of the woman and child (Curvemeister Challenge) I can see the napped texture of the woman's shirt and I can see it's a white stretchy cotton that is tinged with the pink of her shoulder showing through the stretched fabric. We don't see these things when most of the shirt is remapped to something close to 5c2m2y as some did. Does it matter?

I agree with you up to a certain point, everything "depends on the intent of the image, what it's seen next to and your own preferences."

Also,I thought that perhaps Dan's thoughts on the subject were being misinterpreted as I misinterpreted them in the past when I put more importance on establishing a white point and less importance on what it actually did to the image.

Andre Dumas

___________________________________________________________________________

This debate remembers me that old dilemma: to be faithful to the original or to improve it. The first used to be the option of retouchers dealing with other's images. The former, in general, is the way of the photographer who has created the image.

For years photographers struggled with sets of filters, like the 81 series. Now, with Photoshop, it doesn't make sense to pay respect to those long time feared color casts.

But this seems to be a new tendency. More and more I have to keep, or even introduce, some flaws to make an image to look real.

--

Cícero Rodrigues

___________________________________________________________________________

They presumably shot contone negatives and then contacted through all three to produce the black, thus not getting any image except where all three channels were relatively heavy.

This created a skeleton black, there being no other option. GCR isn't possible without bringing a computer into play.

If you'd like to experiment for yourself, take any RGB image, and apply the red to the green in Screen mode, and apply the blue to the result in Screen mode. The final image after the three blends should be a highly acceptable black channel.

Dan Margulis

___________________________________________________________________________

Cicero, thanks for your comments, yours is also my point of view.

I think that artistic license (used intentionally or unintentionally) is entirely at the artist's discretion and is neither good nor bad. With certain images like "Woman and Child" *I* preferred doing it "as it was in reality" for reasons that I mentioned. With other photos I may lean toward sacrificing reality for the sake of "improving" the image.

As Lee Clawson pointed out "it (really) depends on the intent of the image, (...) and your own preferences."

Andre Dumas

___________________________________________________________________________

This might be of interest to photographers (and others).

Recently it has struck me that a dogmatic approach to the ritual of setting a white point has not resulted in achieving the best corrections of my images so I have returned to "The Source" (Professional Photoshop 6) to see if Dan had "misled" me on the subject and I have found something that is quite interesting about specular highlights and catchlights, I wonder how many of us have mistakenly set a specular highlight to 5c2m2y ? I have. I would not do it for a reflection on a piece of chromed metal or glass but how many other very light areas of an image should be classified as speculars and not treated as highlights?

Today I was doing an exercise with Camera Raw, correcting the photo of a small brook showing water cascading against, over and around stones, a portion of the effervescent water catching the sunlight. The author corrected most of the brilliance in the water to something close to 5c3m3y with the result that his "final corrected image" had lost most of the interesting shadow details the original had. If most of that brilliance had been treated as speculars the final image would have been much better.

On page 24 of Professional Photoshop 6 Dan writes "The highlight (...) cannot be a reflection or a light source." . Much of the glare in the cascading water *was* a reflection of the sunlight and should have been treated as specular highlights.

So from now on I'm going to be much more deliberate and careful in deciding what is a highlight and what should be treated as a specular.

Andre Dumas

___________________________________________________________________________

On 12/12/05, colorman042000 wrote:

Most dogmatic approaches bring predictably average results. You always have to think if you want to get the *optimum* result.

Everyone has set a specular too heavy. Sometimes it gets noticed, and quite often it doesn't. Remember that for many images the difference between 4/2/2/ and 0/0/0 is *very* difficult to detect by looking at overall contrast. It's not the best but it's also not going to jump out at you unless you have critical detail in the highlights (think snow scene).

Umm... no. I have a hard time imagining an image where scum dot in a specular highlight would affect shadow detail to that extent.

I agree that it would be better. For most people a contrast move implies 1/4 and 3/4 moves, but you must remember that the first place to look to for contrast is highlight and shadow. If the highlight is full or the shadow weak all the *traditional* contrast moves in the world won't get it back – in fact it can make things worse.

Good idea! Eventually you will get a feel for how to treat certain types of images and the evaluation process won't take long.

-----

J Walton

___________________________________________________________________________

Andre writes,

Correct selection of highlight in this fashion is arguably the most crucial decision that we face in color correction, right up there with control of color casts as the most important factor in the image's success.

The correct philosophy actually goes further than just speculars. The highlight is the lightest SIGNIFICANT area of the image. Significant means at a minimum that you wish to retain detail in it. People often make the mistake of choosing highlights in obviously white areas that are nevertheless too small to make retention of detail a big issue. Sometimes in crowd images there are people who are clearly wearing white shirts that are not specular highlights, but the areas of white are small enough that they should be ignored and treated as if they were.

In my lectures one of the most effective images is an outdoor scene over which a Canadian flag is flying. It is so obvious that this is the whitest area of the image, and that it is not a specular, that almost everyone chooses it for the highlight. Yet the flag will print only a couple of centimeters wide. If its white areas are blown out altogether, nobody cares. And if the flag is ignored, the next lightest area of the image is much darker, so that there is a massive gain in contrast if the other point is chosen.

Also, in inferior printing conditions, you have to become more aggressive,

because contrast is at a premium. So, if you were of two minds about the flag, for example, it might be better to choose it as the highlight if printing on fine coated paper, and yet it would be clearly right to let it blow out in a newspaper.

Dan Margulis

P.S. I agree with J that sometimes the impact is small, but it's very image-specific. If you are mistakenly setting a specular to read 4c2m2y and the next lightest area, which should properly be the highlight, to 6c4m4y then the gain will be small. But if that next lightest area would ordinarily be 12c9m9y, then the gain by taking it to 4c2m2y will be enormous.

___________________________________________________________________________

Thanks for your comments J. Initially I went along with the author and lowered the exposure control (in Camera Raw) to his -1.15, thinking that I would regain the lost shadow details later on with Photoshop's Shadow/Highlight, but the exposure move did seriously limit the quality of the adjustments that I could make with the Shadow/Highlight control. You probably would agree that a -1.15 exposure adjustment in an image that was well balanced initially is a major move.

By treating much of the "sunlit churning water" as specular highlights, further adjustments with S/H resulted in a much better image.

Andre Dumas

___________________________________________________________________________

As Dan says, what you do with an image depends on the image and what effect you're trying to make, which may not coincide with what the camera/scanner/film recorded.

One reason I like the Curvemeister plug-in from Mike Russell (see Dan's LAB book p. 281-282 and the Curvemeister.com web site) is that you have access to threshold sliders that retain a grayed-out image while showing you break-through highlights in color. This enables me to much more accurately decide what is a maxed-out color plate, what is a spectral highlight, what is a tonal limit, and what I might wish to target as my endpoint areas in the image. This function works in RGB, but you have one-click access to CMYK and LAB. Neat tool.

Mike Davis

mldavis2 AT sbcglobal DOT net

___________________________________________________________________________

There was an interesting thread a few weeks ago about setting highlights and when to allow these to blow out. I am thinking now about the shadow point, in black and white pictures.

If I set a shadow point of say 90% in an area where I want to retain detail, is there any reason to limit the ink in an area where there isn't any detail? Can areas of solid black be 100%, irrespective of the paper used?

Thanks.

Toby Macklin

___________________________________________________________________________

Toby,

Rather than "irrespective" I would respect the amount of ink the paper can absorb. After that I generally have 2 things I'm looking at; (1) the black areas of the image (2) the shadow areas and the relative contrast in that area. Shadow areas can run from (a light shadow) 75% on up to (darker areas) 92%. After 90% I assume that unless I have very good contrast in the darker areas the remaining changes will be perceived by most people as black.

Lee Clawson

___________________________________________________________________________

On 12/28/05, Toby Macklin wrote:

Sure! Even on newsprint solid black text, logos, and boxes are 100K, so you can certainly do that with images. If it's a large area that's going 100K in your image you might worry about offset on cheaper paper.

Just remember that when you push areas with no detail you are also pushing dark 3/4 tones so you have to make sure those won't plug if they are important. But you definitely want to use your entire range of black on a grayscale image.

-----

J Walton

___________________________________________________________________________

Toby, can you rephrase your question differently? My initial reaction is that one cannot have two shadow points in *one* image but maybe I'm not grasping the meaning of your question, but it seems interesting(?)

Andre Dumas

___________________________________________________________________________

Toby,

What I'm thinking about is photographs (not text or line work) on any papers that absorb so much ink that the dark areas lose contrast. Specifically the shadows and the blacks start to look too close to one another. Newsprint papers can do this as well as a lot of uncoated transparent / translucent papers.

Lee Clawson

2/\V/\7 Studio

___________________________________________________________________________

Hi Andre -

I guess I'm thinking about the other end of the thread earlier about either setting a highlight point (eg. 4c, 2m, 2y in CMYK) or letting the highlight blow out to the paper.

In this case I'm thinking about black and white pictures. My printer tells me that on glossy paper the range his press can hold a dot is from 2% to 97%. On uncoated paper he reckons more like 90% for the top limit. So, what I've done before is limit the black in photographs I've prepared for coated paper to 90% - nothing in the picture more than 90%. Now I'm thinking this may not be a good idea, since there may be areas in the photo that I want to be just black, without any variation, and the blacker the better - and that what I've done before gives me unnecessarily flat pictures.

Lee suggested in his reply to my question a 'shadow' range of I think 75 - 92% with anything more than that being 'black'. Is this is kind of like having two 'shadow' points in one image, as you wrote?

Thanks for asking me to clarify this - it's useful for me to think about it some more - and little questions are popping into my mind as I write! I'd welcome any input you have.

Toby Macklin

___________________________________________________________________________

"Shadow point" is a term used in Dan's books. It means the darkest possible neutral area with significant detail. and is distinct from any pure black that may be present in the final image.

Mike Russell

www.curvemeister.com

___________________________________________________________________________

Hello Toby,

I'm risking the following but I'm not sure that this is what you are looking for: in the very deep shade of an image that has significant dark areas, I set essential details to something between 78% to 90% black, then areas that I want to have perceptible details (necessary for depth) I set those to somewhere between 90% to 95%. Everything else I set to 97% which on my inkjet is *almost* black-black.

In such a photo my shadow point would be about 98%, this for the very darkest areas in the image. Such areas show no significant details but are significant simply because they give significance to lighter areas. So my shadow point for the darkest black would be 98% and I mitigate the other, lighter, areas by adjusting the curve at intermediate points.

Imagine a photo where 20% (or more) of its area shows foliages in deep shade. I find that using a setting of 90% for the blackest black (instead of 97%) will make such an image look *very* washed out. I have tried 95% and have found it acceptable.

I realize that you are talking offset printing and uncoated paper, I believe Lee has some experience in commercial printing so his figures may be more appropriate than mine. My expertise is limited to inkjets and coated papers.

Andre Dumas

___________________________________________________________________________

Target values for highlights and shadow points can be different for every image. But a good starting point for say commerical offset lithography for a grayscale image would be 5% highlight and 95% shadow.

A good starting point for say newsprint using offset lithography for a grayscale image might be 2% highlight and 87% shadow.

A good starting point for say commerical offset lithography for a cmyk image would be 5% cyan, 3% magenta and yellow. Of course you would follow the guide line to find the white image area with the near neutral white that has details ( the diffuse highlight).

The shadow point target values are is tricky because of different papers and press combinations and the amount of UCR used in the neutral shadow point. The rule of thumb here is to find and use the blackest neutral shadow point. Then place your shadow target values that create a neutral black.

For example a full range cmyk value might be 95% cyan, 85% magenta and yellow and then an 85% black. That would total 350%.

Having said all of that there is no magic numbers where you always use one set of values so that you will always get a great shadow point and then be totally guaranteed you will attain superior white to black contrast in an image.

But if you follow these guide lines as a starting point you will then be able to methodically fine tune your target values for your images and your imaging system.

Jim Rich