Possibly the most powerful adjustment, Curves is used primarily when we have tonal issues, and it can correct color issues as well.
Curves can lighten or darken all color channels at once or individually. If we darken only the red channel, for example, less red light is allowed to our eyes and the image looks darker and more cyan. If we darken the red, green, and blue channels together, the image will be darker but won’t develop a color cast.
Note that the Properties panel displays a histogram in a square with a 4x4 grid. Running diagonally from the lower left to the upper right is the eponymous curve. Yes, it’s a straight line for the moment.
If you’re careful to position that point horizontally in the center, then you’re predominantly affecting midtones, with highlights and shadows experiencing less effect. How do you know this? The horizontal axis is “input.” Note the grayscale running along the bottom of the grid from black to white. There’s one running vertically on the left side, too, for “output.” When you drag a point upward from the center, you should read the curve like a graph: that midtone input is now output as a lighter color, with black and white, if present, not affected at all.
In this image, there are no fully white or black pixels. The histogram’s peaks and valleys stop well short of either end. The image’s pixels possess fewer tones than they might. The ice and snow should contain some pure white pixels, and there likely should be some totally black pixels too. Thus, this image lacks contrast. It also has a pronounced blue cast. We’ll be fixing both, eventually. You might want to read a little bit about “Histograms” (page 226).
Note the sliders along the bottom of the histogram grid in the Properties panel. These make it easier to grab the curve’s endpoints, which control black and white in the image.
This time, Photoshop is trying to help you notice if you are losing detail at either end of the tonal range. This is called “clipping,” and it is usually avoided. If you did what’s shown in the figure, option/Alt-dragging the white-point slider quite far to the left, you will clip the highlights severely. That is, any pixels that turn white are getting clipped to white.
When you option/Alt-drag the white-point slider inward, the Output reads 255 (the level or tone for white; 0 is the level for black). In my example illustration, the Input reads 171, a tone just a bit brighter than midtones. This means that any pixel with a level or tone of 171 or higher is being clipped to white. All those pixels in the histogram above and to the right of the white-point slider will lose all detail.
We want some of those pixels to glisten white, but not that many! So we’d back off until only a few pixels were clipped. If you see colors other than white, that means the clipping is occurring on only one or two channels and not all three. Blue would indicate that only the blue channel is getting clipped, and cyan that both the blue and green channels are. Since this image has a blue problem, we see blue when we “back off” on the white slider adjustment.
If this image were more neutral, we could option/Alt-drag the black- and white-point sliders inward until a few pixels are clipped: a few white pixels showing on a black field if we’re dragging the black slider, and black on white for the white slider. This act is called a “monochromatic contrast enhancement” since it doesn’t affect color. In this case, the image isn’t neutral but rather strongly blue. So this is not the answer here.
Since the steps above didn’t help us cure all that ills this image, reset () the adjustment again. We will do similar steps, but for each color channel. If we can improve the contrast on those consistently (the same amount of clipping), we should have a color-corrected result.
This process can help an enormous number of images, at least if they lack contrast. If you return to the RGB curve in the Properties panel, you’ll see each channel’s curve and the black monochromatic curve too. You can use that now to add or remove a bit of light by dragging up or down in the middle, while still benefiting from your previous labor.
We have actually stretched each channel’s histogram to reach black and white. The Curves interface is a touch misleading about that. To see the actual resulting histogram for the image, choose Window > Histogram to open the Histogram panel.
You may have noticed the little Auto button to the right of the Channel menu in the Properties panel. Photoshop’s default behavior for that button uses a sophisticated but monochromatic algorithm to improve the image’s brightness and contrast. Sadly, it does nothing for this image’s color cast.
You’re not going to like this next step, but here goes: