8/12 UPDATE: In June 2012, I released a full seminar that refines and builds significantly upon the tutorials outlined in 2009’s Skinny Wednesday series. The entire program is now available in the CMstore and includes a 193 page PDF, high-res images for follow-along editing, collection of editing videos, editing tools for Photoshop and Lightroom, skin correction flowcharts, recorded Q&A, and more. Check out more details here!
Basic Color Correction: Black, White, and Gray Points
Many people already know that there are three basic steps for achieving accurate white balance / overall color in a photo:
1) Set black point
2) Set white point
3) Set gray point
The quickest (and easiest) way to do this is by opening up a curves (or levels) adjustment layer in Photoshop. On the left hand side of the adjustment layer panel, you’ll see three eyedroppers: black, white, and gray.
1) Click the black eyedropper on the most significant black point (the darkest “black” spot in the image that should hold detail).*
2) Click the white eyedropper on the most significant white point (the lightest “white” spot in the image that should hold detail).*
3) Click the gray eyedropper on a neutral gray area of the image. There are a number of approaches to finding and using the black and white points.
*NOTE: Remember that your dynamic range is limited to 255 values from 0 (black) to 255 (light). You can maximize the your photo’s detail (especially in the midtones) by choosing endpoints that represent the lightest and darkest areas in which you need to see detail (which may or may not have been the lightest or darkest areas in the scene).
Some people use threshold layers to identify the points, other people use a difference blend mode, and still others just eyeball it (the more you do it, the better you get!).
It’s important that you understand how to evaluate areas that should be neutral black, white, and gray. Every spot on the gradient between true black and true white is a neutral gray, which – by definition – is comprised of equal parts Red, Green, and Blue. When you click your white eyedropper over your chosen white point, Photoshop assigns the RGB values for that spot to 255, 255, 255 (by default; you can change these values); when you click your black eyedropper over your chosen black point, Photoshop assigns that spot to 0, 0, 0 (again, these are default values for black). When you click the gray eyedropper over an area that should be gray, Photoshop takes the R value + G value + B value and averages them, then assigns that average to R, G, and B. For example, imagine that an area that should be gray turns up as R= 57, B =42, G = 51. The sum of the values is 150; divide that by three to get an average of 50. Consequently, to turn the given area gray, Photoshop reassigns that area to RGB values of R=50, G=50, B=50.
I find that selecting the black and white points accurately corrects my white balance 50% of the time or better. Sometimes, however, the darkest point in the image is not neutral black, or the lightest point in the image is not neutral white (perhaps it is a light cream or an extremely light blue part of the sky). Sometimes there simply is no black point or (more often) no white point.
Finding a Neutral Gray
Almost everyone knows that you can use a neutral gray to set the white balance in your photograph … but do you know why it’s so useful? Neutral Gray is considered a “known” color because it exhibits a predictable relationship between points on the R, G, and B curves. We may not know the numerical values for any given gray, but we do know that a neutral gray will always illustrate the same relationship: R=G=B. If the RGB values are not equal in an area that we know to be gray (such as a gray card), we know that the image is contaminated with color.
Unless you use a gray card in your photograph (which we all know is a superior method for correcting white balance), finding a neutral gray in your surroundings can be challenging. Common areas of neutral gray include concrete, gray clothing (or a shadow on white clothing), and rocks or stone. Many people also use the whites of the eyes (which should actually favor red ever so slightly) as the gray point with excellent results. It’s not unusual, however, to find an image totally lacking in neutral gray objects. What to do? Read on….
What if there is no Neutral Gray? Other “Known” Colors
As it turns out, neutral gray (where R=G=B) is not the only “known” color. There are also predictable color relationships for at least three other subjects commonly found in photographs: skin, foliage, and sky. Here are the “rules” for these subjects, which you can evaluate in both RGB and CMYK (with more accurate results in CMYK as a general rule).
Known Color Subject #1: Skin
Skin can best be described as a yellowish red. You already know the color relationships for skin.
1) RGB: When it comes to skin – regardless of the ethnicity, R>G>B
2) CMYK: Y>M>C
As you know, we can oftenget more specific in the CMYK color space. Unless the face is flushed/blushed/sunburned, skin tends to conform to the following:
* Y is greater than (by 0-15 points) or equal to M
* C is 1/3-1/6 the value of Y
I use my own shorthand for this equation: Y>M>c. This tells me that Yellow is greater than Magenta, but Magenta is closer to Yellow than to Cyan. In short, Yellow and Magenta are high and close; Cyan is low and further from the other two.
Known Color Subject #2: Foliage
With very few exceptions, healthy foliage is a yellowish green and is occasionally a greenish yellow. Foliage exhibits a very predictable color pattern, as follows:
1) RGB: G<R<B
2) CMYK: Y<C<M
Again, we can get a little more specific in CMYK. Yellow is the dominant color, but green is also prominent (so magenta is low). The relationship can be expressed as:
* C is midway between Y and M, usually closer to closer to Y – especially with deeper greens.
I remember the equation as Y>C>m, which, signifies that Yellow is greater than Cyan, and Cyan is somewhat closer to Yellow than to Magenta.
Known Color Subject #3: Sky
During the day, skies usually manifest as a bluish cyan or cyanish blue. We express that as follows:
1) RGB: B<G<R
2) CMYK: C<M<Y
Once again, we can narrow down these guidelines a little more in CMYK. On sunny days in particular, yellow should be absent (or close to it), and magenta will be about 1/2 the value of cyan. The relationship can be expressed as:
* M is approximately 1/2 (sometimes slightly less) the value of C
* Y is extremely low (often 0)
My shorthand version of the equation is C>m>y. Cyan is high and further from the other two. Magenta and Yellow are low and closer to each other than to Cyan.
I’ve Got the “Rules.” Now What?
Check your image’s overall color by checking these “known values.” If you know that a gray should be neutral, but you’re seeing R>G=B, then you know there is too much red (or not enough green and blue). Take a look at the grass. Are you seeing Y<C<M? Let’s say instead you’re seeing C<Y<M. If cyan is greater than yellow, you know that you need to bring down the cyan or increase the yellow. In short, I check skin, sky, foliage, and neutrals (or as many of these as are available) every time I correct an image. I may use the evaluations to correct the image globally (say, bringing down the cyan overall to get the grass “correct”), or I may – as with skin color corrections – use these evaluations along with a mask (perhaps I only need to correct the sky, but the rest of the image looks fine).
Try putting the principles into action on your own. In Part 2 of this tutorial, I’ll do a walk-through that puts them into action.
PS – Anyone familiar with Dan Margulis knows that we can refine the rules a little further by evaluating the values in LAB mode – but that’s a topic for another week.