Seeing: The Reporters Eye
Seeing: Sketch it out
Seeing: Review and summation
Color is Fascinating:
Correctness: By The Numbers
Seeing and Color By the Numbers:
Sample Size Matters:
Memory Colors: Introduction
Color Contrast Types:
The Color Expansion Technique:
The Mechanics of the Color Expansion:
The Bell Curve: Masking for HK Effect
Seeing: The Reporters Eye
Recently, while reading the paper, I came across a human interest story that told of how a teen with cancer was dealing with the pain and isolation that her treatment had brought to her. I read with interest how her family had made changes to everyday schedules, how she had taken her medications, visited her doctors, been violently sick from her medications, and even how badly some of the medications tasted. The reporter went the extra mile and spent multiple intense visits with the family getting close to them and trying to express through words what has to be the most difficult and stressful experience that a family can go through. It was a very moving and very detailed story.
It was the rare story where I actually thought the pictures were less expressive than the thoughts and feelings that the reporter had been able to record from the teen and her family. Her hopes, her dreams, her sadness at the thought of losing her family and friends, down to her fear that her pets would not be properly cared for. While it was a sad story it was also very full-filling as the follow up story told of her remission and the renewed hope for a fairly normal life. Both stories painted me a picture verbally of a part of that person that I would not ever see if I never took the time to listen and understand like the reporter had.
So what brought this story so close to home? It was the details in the text that were deeper than surface facts. Things like how the medicine tasted to the girl, what color and type of pets she had, the color of her poorly tied shoelaces, a description of her missing hair, the telling of her very real fears; all of these built up the understanding of who this person is and why this story was so compelling. During the overview of the story the editor told of how the reporter had spent multiple days getting to know the family and had gained their trust and acceptance. The reporter admitted that they felt deep empathy for the girl and for her family and promised honesty and compassion.
For the story to succeed it was essential that the family trust the reporter to tell the truth and not put words into the mouths of the subjects. I thought to myself, how do I do this with a picture? How do I tell any story? What things do I look for that tell that story? How do I know I have dug deep enough into what I am seeing to know that I am telling the story I want to tell? This is where the “Reporters Eye” comes into play.
I ask myself what things would a reporter do that make the story deeper? They did not just drive by and make a quick one off meeting, they took their time. As we have discussed before, taking time with your corrections gives you the freedom to see the problems rather than guessing that you need to change something. It frees the mind to wander about the images and discover the unique things that you want to enhance. Having read the article I also now believe that there are multiple revisions that have gone into the story to distill out the essential items that they focused upon and used to bring the story to a more personal level. This is a part of the image process I want to address this week. Revision.
Image correction is such a subjective process that what you do today might look funny to you tomorrow. It is difficult to determine what is correct for many different reasons some of which are discussed this week in regards to color and contrast. In order for you to sort out the “keepers” from the “also ran” image corrections; you are going you need to remember to give the image a rest. Typically, I give my work a 24 hour waiting period unless I am under a very tight deadline I always give my corrections a 24 hour rest to give me time to clear the image and impression out of my head and to allow me to see the image with new eyes so to speak. If when I go back to an image after the 24 hours it still looks good to my eyes and I find nothing I would think about changing, then I finish the image with whatever processes I need to finish be it sharpening or re-sizing or saving in a different format.
Seeing: Sketch it out
While I am a half decent photographer and post processor, I find myself lacking when it comes to a very important tool for understanding some of the deeper connections in a subject when you are trying to see it. Sketching. The willingness to sketch a scene is a very powerful tool that you can use to aide in your quest to see things better. The focus and attention you need to give to your subject causes you to start to make an in depth analysis of the subject and instantly opens you to Sharpened Vision.
Depending on your skill level with sketching you might jump into Heightened Vision as your mind and hand work together to extract the details necessary to create the image you are sketching. It is this connection to the details that I want to highlight as a part of the process. Many people who sketch find that they relax as they are doing the process and that also aides the transition from normal vision to Sharpened and Heightened vision. I myself use the layers feature of Photoshop to give me a “sketch pad” and I make editing notes upon it as discussed in the first session of the class. I use the mouse to draw on the layer and mark out areas I want to enhance and pay more attention to. My notes and editing sketch help me to stay focused on what I want to enhance and provide me with a check off of the items so I know I have done everything I set out to do. While I do not always display this layer…in point of fact for almost all of the work I have shown to you over the years I have never shown this process. I find it invaluable for my vision and use it frequently to get at the details of the corrections I want to create.
Color is Fascinating:
One of the most versatile properties of an object is it’s color. Color helps define shape, denotes some specific objects, and can even infer a specific feeling; it provides context for relationships of distance, color even helps us survive. Without our ability to see in color we would have a tough time figuring out what foods to eat, whether or not a predator is stalking us, what time it is, or where our car is in the parking lot. It is a fundamental human perception.
It is no small wonder that we spend so much time working on enhancing our images so the color looks and feels right. We can spot a fake image in a heartbeat, an enhanced one takes much longer to figure out because it is believable. When we take our “seeing” out for a real world walk around, one of the first things we will be noticing is how often color is used to provide mood and feeling in both classical and modern art media. It is a basic element of art. Color is also used to get our attention, draw our eye and convey instant information to us.
Understanding properties of color and color perception are very important to your ability to make objective judgments about the “correctness” of color within the context of the images you are working on. This week we are going to explore color. We are going to look at it in terms of correctness, contrast, and perception. We are going to push it around a bit and we are going to enhance it to meet the intended goals we have for our images.
Correctness: “By The Numbers”
In the current race to the bottom of the quality mountain with image enhancement; some color enhancement people have adopted a faster, cheaper, good enough; philosophy regarding the way images are handled. They believe that people will accept good enough and that they need to compete commercially, and therefore needing to do it cheaper and faster so they can get on to the next item and keep the revenue flowing; they also have to compete globally for the attention of the viewer and the potential money the viewers may be willing to provide for their services. Thankfully, I am not one of “those” image correction people. In this regard, I believe that we are making a grave mistake in sacrificing image quality for time and money and that image quality will be suffering for years to come because of it.
It has become obvious to me that WE NEED TO DO IT RIGHT! If we miss the color correction then the rest of the image will struggle to make itself believable. No amount of cropping, burning and dodging, masking or sharpening can fix badly color corrected image. In CM 101 we learned a “By the Numbers” (BTN) approach to image correction. using BTN imposes a certain set of standards upon us; if we do something “BTN” we are expecting that everything is 100% correct, that it has been analyzed numerically, and is in fact without mathematical flaw. Perfection. Getting the color “correct” is one of the foundation building blocks for future steps in the image enhancement process we are following and failure to get the color “right” will lead to bigger problems with the color as we later boost and expand color in our images. We all know that type perfection is just not real. It feels good to say it; but like any other analysis it comes down to “What do the numbers show?” The smart answer to that question is “What do you want them to show?”
Remember the following. “Smoking is a leading cause of statistics.” in the United States.
- “43% of all statistics are worthless.”
- “3 out of 4 Americans make up 75% of the population.”
- “Death is 99 percent fatal to laboratory rats.”
- There are lies, Damn Lies and Statistics” – Mark Twain
- “Color correction BTN is always better than a 79.5% guess.” – Greg Groess
I added the last one with my tongue firmly planted in my cheek to give you pause; and to remind you and myself that as we go over the BTN approach to color correction we will need some solid framework to begin our analysis and build our corrections from.
“Seeing” and Color By the Numbers: Building your mental framework.
In the first week of this course we ask questions about the images we are looking at. We asked:
- Are the significant highlight details intact?
- Are the significant shadows details intact?
- Is the color flat? Is there detail in that color?
- Is the color over or under-saturated?
- Is there enough variation in the existing colors?
- Are the colors believable?
This week we are going to address the color aspects of the image and the highlight and shadow as they relate to the color of the image. Frequently in the CM101 class I am asked “How do you know that you have selected the right.<Insert your favorite image attribute here>..” My answer is usually an attempt to make verbal sense of something that I just do; I will tell you now that I do not always know that I have selected the right anything. I test it to see if it is right.
Curvemeister Testing tools:
Within the Curvemeister plug-in, you have some very useful testing tools. You have the shadow and highlight pins, you have a threshold feature to help you find the shadow and highlight points, you have a histogram to tell you how the pixels are distributed in the brightness of the image, and you have the hue clocks to show you what color your selected sample is. The Curvemeister hue clocks are probably the best tool in the Curvemeister tool box for color correction; because by using them you get real time information regarding any sample point in your image and what your curve changes are doing to the values of those sampled pixels. The sample points are the heart and soul of the hue clocks and your ability to use them correctly. Let’s take a quick refresher look at the settings for the hue clock in Curvemeister and why some enhancements require you to make changes to the sample point while you are using it.
Sample Size Matters:
Every time you set a hue clock in Curvemeister you are gathering together a group of pixels for analysis. The number of pixels in your sample determines the accuracy of your adjustments. If you are looking to define a shadow or highlight point you need to make the sample size small enough to cover the highlight area but not so small as to be impossible to adjust. Smaller samples are harder to adjust because they usually involve only 1 pixel or 4 pixels. The chance that the 4 pixels you sample are perfect for the highlight or shadow is very small. When you make a change to the curves based on such a small sample, your error possibility goes way up. You will miss the real important area you want to adjust unless it really is that small. In general we do not want to correct to a specular highlight or pinpoint of a shadow value.
3 by 3 5 by 5 10 by 10
Above you see a zoomed in section of an image with the selection sizes superimposed over the pixel squares. Notice how much variation there is in even a 3X3 sample. Those 9 pixels are averaged to provide the numeric value you see in the hue clock display. Notice also that the the 5X5 sample has even more variation; by the time we get to 10X10 we will have far too many pixels in the sample to give us a reasonable value for any point in the image. We do get an average for the range we have selected and that is useful later in the color enhancement process but that is not going to give us a result we can take forward into the BTN process with any confidence that we are getting the color correct.
Assignment 1: Compare the sample values in an image with the sample size settings 3, 5, and 10 pixels. (Total recommended time 10 – 30 minutes)
Do a basic by the numbers correction at 3 pixels and at 10 pixels. How do the images compare? which would you prefer to take further in the image enhancement process? Post your images and report your results out to the group for discussion.
For general use I set my sample size to 3X3.
For me it is a good compromise in that it is large enough to cover a reasonable area but small enough to not introduce major errors. OK, I have my sample size figured out…now lets place it; but where do we go first? For shadows and highlights the in a BTN correction you need to consider significance. What is significant? Is a dark doorway well behind the subject significant? What about a chin shadow? The highlights in someones hair? Where do we draw the line?
The choice will most often set the overall tone of the image. Unfortunately, these are highly subjective choices in that they always involve the context of the image and will always be changing. On the surface they all share some common things and that is where we want to head in our correction process. We are looking for areas of the image that we believe are, or should be neutral.
A BTN correction introduces neutrality by definition; unless we are adjusting something to a known value RGB.
In any general use image without a grey square or known color area the only known values are Shadows, Highlights, and Natural Neutrals. Selecting a Shadow or Highlight in a BTN correction is not just a selection of an area that is RGB(255,255,255) or RGB (0,0,0) in fact that selection will most likely be a bad choice because it has no definition and no detail. We want to choose a sample point that leaves us with details in the highlights or shadows. Currently, I recommend a value of 235 to 250 for highlights and 7 to 3 for shadows for any BTN correction. These values leave enough room for subtle details to emerge in the final outputs but they are far enough out from the ends of the curve to allow you to adjust the image without getting into trouble. Many times highlight points have a value of 220 to 230 or for shadow points values of 7 to 10 . In these cases we need to step back and look at the overall image luminosity and contrast; additional adjustment might be needed to the middle tonality of the image. We want a highlight value that is at or above RGB(235,235,235) so that we perceive it as bright white with details or a shadow value that is below 7 so that the shadows are perceived as complete.
In order to help you limit the ends of the curve and preserve details in Curvemeister there is a feature called “Shadow and Highlight Target” in the settings area where you can set upper and lower limits for the output by Curvemeister; of the RGB values in any image you work on in Curvemeister. This setting prevents you from creating a file with RGB values less than the lower number and greater than the higher number; regardless of the actual outputs you set in the image. For example: If I set a highlight that is RGB(255,255,255) Curvemeister will limit the outputs to RGB(250,250,250) if that is where I have set the limit. This setting is used to control the shadow and highlight and prevent you from blowing out all the highlight details or blocking up all the shadow details in print or on-screen.
Well now we have a problem; it should be obvious to everyone but I’ll spell it out in any case. What happens if our image has no natural neutral ? In theory every image should have a shadow and a highlight but what happens when we have no neutral? Here is an image without an obvious neutral. we’ll soon see what happens.
Colors have contrast; in fact they have multiple contrast properties that influence your perception of the color you are seeing. We humans need color contrast in order to survive. It is the ability to see the colors around us that enables us to find food, understand spatial relationships and even remember where we parked our cars. We also need to understand that our perception of color is tricky and can be effected by color contrast properties. We need to understand color contrast so that we can make better choices in our color correction and color expansion processes.
Color Contrast Types:
The contrast of saturation:
The term refers to the contrast between pure intense colors and dull diluted or grayed colors. Dull colors would appear to be duller when it is placed next to pure intense colors, and pure intense colors would appear move vivid when it is next to a dull color.
The contrast of extension:
Also known as the Contrast of Proportion. The contrast is formed by assigning proportional field sizes in relation to the visual weight of a color. Different amounts of one color are needed to balance another. The contrast of extension is used to refer to contrast between the proportion of one area of color to another.
The contrast of complements:
The contrast is formed by the juxtaposition of color wheel or perceptual opposites.
The contrast of light and dark:
The contrast is formed by the juxtaposition of light and dark values. This could be a monochromatic composition.
It is the contrast between two colors that are almost complimentary, but not exactly. It is contrast between a color and another color that is to the right of left of its compliment. Satisfaction to the eye requires harmonic balance of the colors. If the colors of something you’re looking at were not balanced, the eye would tinge colorless, gray, or pure colors with the compliment of the colors next to it. Therefore, the reality of the color is affected by it’s surrounding color, and would appear that is has shifted towards it’s surrounding color’s compliment. In other words, a color would look different than what it really is. This effect gives a feeling of excitement and lively vibration of colors of changing intensities.
The contrast of hue:
The contrast of warm and cool:
Of the contrast types above the ones we can have some control over during image processing are contrast of saturation, hue, and light and dark. For the most part the others are compositional or environmental contrasts that we have recorded but our changes will naturally enhance them as we remove color casts and separate the color hues by expanding specific colors ranges.
When we look at a sample point using Curvemeister we actually see a “Range” on the channel curve. This is the “color worm” that is displayed by CM. It shows you the upper and lower values in the sample you have selected. The points in between (the worm) are the range of the sample.
One of the strongest analysis tools in the CM tool box is the data shown by the color worm for this range. It defines the color we have sampled and it allows us access to the specific hue we have selected within the image on each color channel displayed. We can use range data for many different things. We can use it to add tonal contrast if we use the Master channel in RGB, we can use it to increase tonal contrast in LAB or CMYK using the L or K channels respectively; we can also expand the range of any given color by using LAB and a contrast pin that we select from the right click menu or better yet a contrast pin that we have build manually.
So now that we know how to build a contrast pin just what are we going to use it for? Since we are talking about color we are going to use the contrast pins we built to expand the range of a selected color to enhance the difference between various hues in an image and increase the saturation at the same time. This is a very powerful technique that can make a noticeable difference in your image very quickly. There are others ways to accomplish the same process but they usually involve Photoshop actions that you have very little control over and you may have to run several times to get just the right colors selected. Using this technique you will be able to define the colors you want to expand and keep control over the rest of the image.
The Color Expansion Technique:
Contrast Pins in Curvemeister allow you to “stretch” the curve line and separate the pixel values numerically. If you think of the curve line as a flexible cord you could say that the curve only has so much “length”. We take advantage of that length in many ways. We use the curve to make “s” Curves for overall increases in contrast, we move the endpoints to make darker masks, we add lizard tails to the ends of the curve to recover lost shadow and highlight details. When we increase the contrast of the image by making the curve steeper we are in fact shortening the curve and increasing the “slope” of the line.
In Geometry the slope of a line can be defined as “rise over run” meaning that as the line runs from left to right the height of the line increases; the actual equation is: m = Change (y) / Change(x) When we increase the slope we change the ratio of rise over run. The net effect of this is to move the points along the line closer together and increase the difference in the values between any two sample points. Let’s start with a better understanding of a contrast enhancement first then we’ll move onto the color contrast part of the process.
The Image above shows the areas effected by a straight contrast move. Notice that as the curve line approaches the the pivot point <not necessarily the center of the curve > the amount of the adjustment is reduced. The red area is on the highlight side of the image and the green is the shadow. My adjustment has effected the highlights more than the shadows and it has done little for the mid-tone contrast of the image. Overall the image is improved but there is more work to be done. We will need a secondary adjustment to increase the contrast in the mid-tones.
In order to make the most of this next adjustment we need to use the histogram. The histogram is the Grey area that looks like a mountain peak in the background of the curve window. If you do not have them turned on in CM you might want to go turn them on now. The histogram shows us how many pixels are effected by our adjustments but not exactly where in the image the pixels are located. This is still useful information, we will after all be using it to adjust the mid-tone contrast in our image after we have set a highlight and shadow point. As you can see in the image above the curve line does not cross at the center of the histogram; we will need to make a decision. Because we have adjusted the ends of the curve, we have shifted the overall contrast around and now we need to determine where the “area of importance” in our mid-tones is located and add an adjustment to the curve to make the mid-tones look as good as the highlights and shadows. Start by placing your mouse pointer in the mid-tones of the image and move it around until you find the area or areas where the you want the mid-tones to have increased contrast the “Worm” covers the crossing area of the curve line as shown below. This is the area we want to work on next.
You can right click at this point and select either “mark” or “Contrast Pin” from the fly out menu in CM. If you choose “mark” you will have to manually select and link the curve points so that you can adjust the contrast (Building a manual Contrast Pin). If you use the “Contrast Pin” you can place the mouse on the curve frame edge and rotate the curve points to adjust the contrast.
If you choose to use the “mark” feature of Curvemeister you have a bit more adjustment room in where you place the two points you want to use as the contrast pin and your choice is select-able. Start by Clicking on the Curve line above and below the mark you have placed on the curve shown above. You should have 2 control points on the curve now. To make them a linked contrast pin you need to click on one control point and then Ctrl-click on the other one to link them as shown below. Notice I have extended the region I have selected under the linked pins. I made the decision to do this based on the histogram information and on my assessment of the image. If I chose the ends of the “mark” I had placed on the curve I would be ignoring some valuable mid-tone data I want to enhance for the image. Remember the Sample size? This is an area where it matters.
After the contrast adjustment notice that the area near the center of the curve has been expanded and that the mid-tone contrast areas have had a bigger change introduced.
We are going to use this type of expansion to help us adjust the variation in the colors of our images and separate different values of the same hue creating more color variation and apparent detail in the image. To take advantage of this we are going to need use the LAB color space. Since we know that LAB offers the advantage of having color separate from the luminance information in the image structure; LAB gives us some very powerful color tools that we can use to help our image look better overall.
The Mechanics of the Color Expansion:
The LAB color space has 3 channels. L for Luminance, A Channel for Magenta (Red) to Green color and B channel for Yellow to Blue color. We are going to be using the A and B channels for our work so let’s take a quick look at what is going on in the A and B channels.
In the image below you can see how the colors are located in the A and B color spaces.
Just from the histogram data displayed above I can see that both Green and Blue are almost absent from the image since all the pixel data rests to the yellow magenta sides of the of the vertical line I have drawn on the channels to divide them into colors. The curves show me that I will not be able to expand on the green very much and that finding any blue in the image will be nearly impossible. So, how do we expand the colors we can work on? The first thing we must be certain of is that the image has been corrected for color balance BTN before we attempt this color Expansion step.
If you have not done a By the Numbers correction or you are not certain the image is neutral you should go back and verify that process before you move on to this step.
Failing to ensure that the image is color correct will introduce wild colors and odd color casts as we increase saturation and expand the colors across the specific hues. One of the fastest ways to do this is the Saturation slider in LAB. Use the slider to increase the saturation to an extremely large value and see if the color of the image holds together.
Image before BTN Correction Supersaturated Image before BTN
BTN Color Correction Supersaturated Image after BTN
Let’s look a the images above and do some quick assessments.
- The Before BTN Image looks pretty good overall but upon supersaturation you notice that the backs and necks of the swans have a lot of blue in them that we really would not want to enhance as a part of our Color Expansion process. Note that the color of the water is mottled as well and blue and yellow are mixed in the scene
- When we do a BTN correction on the swans to get the color right we found that there was a blue cast over the image and by adjusting the red and blue channels we were able to make the image neutral. In the Supersaturated image the swans are more neutral in the backs with white whites and the toned areas of the necks are consistent and while way over saturated they look normal.
Color Expansion Walk Through:
The first step in the color expansion process is to set up an Anchor Point. The anchor point serves as a point of reference and allows us to return the color balance the next steps are going to destroy. Open the image in CM, switch to LAB mode, and take a good look at your image. A good Anchor Point is usually found in any area of the image that you do not want the color expansion to affect. Neutrals are a good target for setting the Anchor Point.
The fact is that you have to decide what colors you want to do expansion on and what colors and tones you want kept relatively unchanged. Once you have settled on the Anchor Point; Alt-click on the image and set a hue clock for your anchor. Do not move this point after you have begun or you will lose your starting point. The first values shown on the hue clock are your targets to return that specific sample point back to after your changes.
Once you have your Anchor Point you need to find an area of the image that contains the specific color you want to expand and add variation to. Foliage, grasses or textured surfaces are all good candidates for this process. You should be looking for colors that seem to blend together to hide detail or colors that you feel could just plain look better if more varied. During our seeing exercises we learned that the details of the colors we see are sometimes hidden in the cameras ability to differentiate certain colors. We are going to fix that problem here.
Remember the discussion on Sample Size? This is where we might need to make a change to our sample size. You can set the sample size for any given Hue clock from the hue clock options menu. Set your sample point and then click on the little arrow in the hue clock. You will get a menu of options to set for that hue clock and one of the choices is “size”; set the size to 11. You now need to choose a single channel in LAB either A or B is fine but, you should display it rather large on your screen; we may be working near the center of the curve and we need to be able to see what we are doing.
Move the sample point around the image slowly and watch the color worm on the selected channel. In most areas of the image your color worm will be quite small but in a few areas you will see a quick jump in the worm where it gets rather wide and then shrinks back down. This is one of your possible targets…What has passed under your mouse is an area of the image where there is a large difference between one color and similar values of the same hue. These areas are prime targets for color expansion. Try to find that point again and be patient. Hitting the exact spot is not always possible. When you have a moderate length color worm displayed on your channel you can right click and select “Contrast Pin” this will set a linked contrast pin in the channel you are working with and allow you to adjust the contrast of the colors within the sample region.
A word of caution, Contrast pins that cross the neutral in LAB are a really bad idea. While they help you increase saturation they do not focus the adjustment on a single hue family. If you move the neutral you introduce a color cast.
Place the mouse pointer on the edged of the curve frame and it should change to a bent double arrow. If it does not try the top or bottom edge closest to your linked pin. Left click on the frame edge and slowly move the mouse to the right. The linked pins will separate and the curve line between them will become an “S” curve like the examples above. The end result is a greater difference between the sample points and greater variation in the color. This move typically increases saturation as well since the color curve is steeper overall. It also results in an image with wild “Man from Mars” colors. We need to get back to a more normal place. We now need to use the anchor point.
With the contrast pin still selected and linked; you can use the arrow keys to move the linked pins vertically until you have returned the channel value back to the original value displayed. Sometimes you have to hunt around a bit to get the values back but take your time and return to the anchor value this will keep the colors in balance. You can repeat the process on the other LAB Color channel if you need to make further color adjustments. This is usually a subjective choice but almost always your image will benefit from it.
Perception and Memory Colors:
In a recent color correction discussion someone asked if I thought the green grass in an image was ”Green enough” I responded with a question of my own…”How green is the grass in your dreams?” It was a thought provoking answer to the question because many people in the last few years are following the trend in color enhancement for brighter and bolder colors. Push the color harder…Make it Pop…have you ever heard that? I have..The trouble is that when people do push the color they are making those colors into something they never were; they are adding something odd to the image that leaves it looking “Photo-shopped” This can be an artistic choice I suppose; but for the purpose of this class I would like us to consider that over saturation of a color is just as bad as completely missing the color of an 18% grey card. We are going to push the color around to be sure; but we are not going to look for color that is outrageous or false. We want real looking, interesting color that draws us in and we can believe in; memory colors will lead the way.
So what are memory colors?
A good working definition is “colors that a majority of people can agree are appropriate for any given object.”
There can be a range of hues within that definition that are acceptable but there are certainly colors that make no sense being in or on certain places. An example would be green hair. Unless it has been altered by something; green hair is just not a correct hue for hair; nor is blue. As a general rule we can pretty well state that “cool” colors are not acceptable for hair. Having just typed that I now face the question of what to do with Grandma’s blue hair? Wait…that is an artificial dye color and so I would not use that as a sample for checking the hair color…Whew..I though I worked myself into a corner.
An interesting thought on memory colors…
In a recent test; fifty observers chose their memory colors from an array of 931 Munsell color chips. The variability of the judgments were shown and their means were compared with the average chromaticities of the corresponding natural objects. The ten mean memory colors were all significantly different from the natural colors. Each memory color tended to be more characteristic of the dominant chromatic attribute of the object in question; grass was more green, bricks more red, etc. In most cases, saturation and lightness increased in memory.
For now understand that your memory of a color is an important tool in the process but that you can be fooled by your perceptions and memories. So how do we trust our perception?
If fifty people all chose colors that are too saturated and too light for known objects how do we know we are not making the same memory mistake? Well the fact of the matter is we don’t; because we are fooled. Our memory and our eyes natural ability to make something look normal even when it is not causes us a problem. While we need to use memory colors to help us prevent obvious color correction mistakes and find color casts they cannot help us when it comes to saturation and brightness.
When it comes to color and saturation you need to understand the “HK” effect. The Helmholtz-Kohlrausch effect (HK) might better be called chromatic luminance, since “white” or achromatic luminance is used as the standard of comparison. The idea at it’s most basic is this; the more saturated a given hue; the brighter <more luminent> it appears to our eyes. For some colors we perceive the brightness to be 2 to 3 times as great as it actually is.
As you can see an image with a HK problem needs some help to restore the colors. Notice that while the Blue hue clock on the right seems to be brighter the actual value is lower for brightness, 34 on the left and 15 on the right. The increased saturation is fooling us to think it is brighter overall. We don’t want the colors to distract us from the main subject of the image just because the perception of them is eye catching. There are multiple ways to tackle this but for this class we are going to introduce a different kind of mask. We are going to make a mask that allows the saturation and color contrast we have built into our image while protecting the brightest colors from looking too bright because of over saturation.
The Bell Curve: Masking for HK Effect
Curvemeister allows you to make some very interesting curves, often when we make wild and woolly curves they are going to be used in masks. The ability to create a mask in any form we can conceive is one of the best features of Curvemeister. You are limited only by your imagination in what kinds of tasks you can complete using a mask. For purpose of this process we need to go back to the definition of the HK effect and see what we can do to reduce its result.
The Helmholtz-Kohlrausch effect (HK) might better be called chromatic luminance, since “white” or achromatic luminance is used as the standard of comparison. The idea at it’s most basic is this; the more saturated a given hue; the brighter <more luminent> it appears to our eyes. For some colors we perceive the brightness to be 2 to 3 times as great as it actually is.
OK, so where can this take us? Another way of looking at this is that we perceive brighter things to be more saturated than darker things. We need a mask that allows us to saturate and expand on the shadows and highlights while holding the mid-tones in check as they become over saturated far too quickly. Would an inverted K or L channel help us prevent the HK effect? Possibly but remember that the inverted mid-tones will have an effect via the mask as well. We need to create a mask that blocks the mid-tones from getting the full effect of our change. What we really need is a Luminance mask that blocks the mid tones and little else. Enter the Bell Curve…
Using the Curve shown I have created a mask shown below on the left, that does some pretty interesting things. As you look a the left image below you can see that in comparing the left to the right I will be protecting the mid-tones of the image from the majority of the changes because they are the black areas of the mask. Note that the highlights and the shadows are open for adjustment and will be passing the color enhancements through to the final image. The really interesting thing about this curve is the adjust-ability of the middle of the curve even in this extreme shape. Please Watch this video for a greater overview of the Bell curve and how you can shape it to your needs in HK control of saturation.
This has been a very important session. There is a large amount of information for you to absorb. Please ask questions and be patient with yourselves as you explore this material.