Hue Curve Adjustments
Hue curves are powerful, fast-to-use controls that let you make gentle or sharp alterations to the color components of an image by placing control points along sections of the spectrum. Unlike RGB curves, which plot individual color channels vs. image tonality, hue curves plot individual color components relative to user-defined ranges of hue (Figure 4.54).
Figure 4.54 The hue curves UI as seen in Apple Color.
Their power lies in their ability to quickly make alterations to specific subjects in an image based on their hue alone, and the mathematics of curve operations make most adjustments really smooth and seamless, with none of the edge artifacts you can get when you're using HSL Qualifiers to pull a key. This is especially good when you're making huge targeted changes to image saturation (something I do quite a lot).
Most grading applications have, at minimum, three hue curves:
- Hue vs. Hue
- Hue vs. Saturation
- Hue vs. Luma
Some applications also have Saturation vs. Luma (providing curve-based control over saturation in targeted areas of image tonality that's quite powerful); however, the minimum hue curve controls correspond to the three essential color components of the video signal.
Hue Curve Controls Compared
Hue curves are available in many applications, including Apple Color, Assimilate Scratch, FilmLight Baselight, and Quantel Pablo (Quantel was the originator of these curves, and Quantel applications from Paintbox through Pablo have referred to this as the Fettle interface). Figure 4.55 shows a comparison of Hue curves as represented in different color grading applications.
Figure 4.55 Images of hue curves in UIs compared, top to bottom: Excerpts from Quantel's Fettle UI, Assimilate Scratch, and FilmLight Baselight.
While we're on the subject, one of the standout features of Colorista II is a pair of circular HSL controls, which essentially perform the same function (Figure 4.56).
Figure 4.56 Circular HSL controls in Magic Bullet Colorista II.
These unique takes on the hue curve interface have efficiencies built into their UIs:
- The left control combines the functionality of the Hue vs. Hue and Hue vs. Saturation controls; changing the angle of a particular hue handle shifts hue, and changing its distance from the center adjusts saturation (similar, in fact, to the familiar use of color balance controls).
- The right-hand control lets you alter Hue vs. Luma.
The only drawback of this system is that the number of hue handles is fixed, whereas curves let you add as many control points as you like. However, practically speaking, you can do just about anything with the Colorista interface that you can do with curves, except for perhaps a handful of extremely tight multi-hued adjustments.
Let's take a look at each of the three basic hue curve functions using a helpful hue-wheel test pattern to examine their effects.
Using Hue vs. Hue Curves
Hue vs. Hue lets you isolate a range of hues that you want to literally change into other hues. The effect is similar to the hue parameter that most grading applications have, which rotates the hue of the entire image around a 360-degree axis (which causes a familiar "rainbow shift" if you animate it). However, using a curve interface, you limit hue shifts to specific portions of the spectrum.
In the following example, the blue portion of the spectrum is loosely isolated and raised on the curve (Figure 4.57).
Figure 4.57 The blue portion of the hue spectrum isolated and raised on the curve.
The result is that the hue of the blue wedge of the color wheel is altered, in this case changed to green. Because the curve has a smooth falloff, you can see hue transitions from the altered green into the other original colors on the wheel. Although obvious on the test pattern, this isn't always obvious in a real-world image, but it shows that subtle hue shifts are usually more successful than huge ones.
I use this control for making subtle adjustments to skin tone (taking advantage of the narrow range of hues that skin occupies), foliage, and skies (all techniques you'll see in subsequent chapters).
Using Hue vs. Saturation Curves
Hue vs. Saturation is the hue curve I use most frequently. It lets you isolate a range of hues within which to raise or lower saturation. Again, isolating the blue of the spectrum—in this case, lowering the curve—lets us desaturate that particular portion of the color wheel. Again, because this is a curve, the falloff is gentle, although a sharper curve results in a sharper transition from the desaturated to fully saturated areas of the image (Figure 4.58).
Figure 4.58 Lowering the curve to desaturate a portion of the color wheel.
This is a powerful tool for stylistic adjustment (reducing the saturation of everything except the blue of the sky, for instance), addressing quality control violations (reducing the intensity of reds that are peaking too much), or shot matching (eliminating a subtle yellow cast that's otherwise hard to isolate).
Using Hue vs. Luma Curves
Hue vs. Luma lets you isolate a range of hues within which to lighten or darken corresponding features. Isolating the blue portion of the spectrum and lowering this control darkens the blue wedge of the test pattern (Figure 4.59).
Figure 4.59 Isolating the blue portion of the spectrum.
This is a potentially powerful tool but also a really tricky one, the results of which depend highly on the quality of your application's image processing. Although it's possible to make smooth adjustments using nearly any type of application that uses Hue vs. Hue and Hue vs. Saturation curves, the Hue vs. Luma curve is at a disadvantage because it manipulates the most data-rich component of the signal (luma) relative to the most data-poor component (chroma). As a result, it can exhibit unpleasant artifacts such as macroblocking and edge aliasing.
These artifacts appear almost immediately (Figure 4.60), no matter how small a correction you're making, if you're working on media with 4:2:0 chroma subsampling (such as HDV or H.264-based formats). 4:2:2 chroma subsampled media has a bit more resiliency, allowing small adjustments before unpleasant artifacts appear; although, past a certain point, even 4:2:2 media will suffer.
Figure 4.60 At left, our test pattern downsampled to the 4:2:0 Apple Intermediate Codec, showing harsh artifacts from use of the Hue vs. Luma curve. At right, the same image at its original ProRes 4444 encoding; an identical correction exhibits smoother transitions.
However, as you can see in Figure 4.60, this control comes into its own when you use 4:4:4 chroma-sampled media. With the full range of luma and chroma available to you, you'll be able to use the Hue vs. Luma curve to the best possible advantage.
Controlling Hue Curve Adjustments
Some hue curve interfaces place a series of pre-defined control points on the curve, giving you a rapid starting point, while others present you with a blank slate.
The most important thing to know about controlling hue curve adjustments is that you should "lock off" parts of the curve you don't want to adjust using additional control points. For example, in the following curve, you might want to make a series of adjustments to alter the orange hues of the image, without affecting the neighboring yellows or reds.
Placing control points at the neutral center position at the border of the areas you don't want to affect limits the falloff of your adjustments, preventing your correction from spilling over into hues that you don't want to affect.
Using Hue Curves to Fix Mixed Lighting
So now that we've seen the theory of how hue curves work, let's see some practical examples. One of the biggest bugaboos of many scenes is mixed-lighting scenarios, where you've got light of two different color temperatures competing with one another. You might be able to overcome the problem with a creative use of color-balance adjustments, but more often the solution won't be that simple.
I've found that, especially for subtle mixed color shots, the Hue vs. Hue curve can make a quick fix that works really well. In the following shot, a subtle greenish/yellow cast from a practical fluorescent fixture that the production couldn't eliminate is polluting the cool blue sky-lighting of the scene (the spill can be seen within the red circle of the original image in Figure 4.61).
Figure 4.61 A Hue vs. Hue curve adjustment (shown in Assimilate Scratch) used to shift the green/yellow fluorescent spill toward the cool blue sunlight of the rest of the scene.
The problem is too subtle for HSL Qualification (which is a mixed lighting solution I often use for window exteriors that are a lot more blue than the rest of an interior scene), and trying to fix this with color balance will only make the midtones more magenta than we want.
Another strategy could have been to desaturate the offending greens, but that would have leeched color from the scene; this way, the offending greenishness has been converted to a pleasing sunlight-emulating bluishness. The result is smooth, seamless, and quick.
Using Hue Curves to Selectively Alter Saturation
Here's a practical example of using the Hue vs. Saturation curve to make a stylistic alteration to a real-world image. In Figure 4.62, the original shot as graded has a saturated green pool table with an orange ball. The client wants to bring more attention to the ball, and one way of doing this in a creative way is to both raise the saturation of the narrow portion of oranges that correspond to the ball, and lower the saturation of the green table to make it a subtler, darker shade.
Figure 4.62 A multi-point Hue vs. Saturation curve adjustment (shown in Assimilate Scratch), deemphasizing the green in the table, and heightening the warmth of the orange ball and man's skin tone.
This serves to lend greater weight to the wood tones throughout the image, which are also picking up a bit more saturation because of the increased orange, which is additionally giving the man's skin tone a bit more warmth. More to the point, the orange ball really sticks out after this adjustment.
The Hue vs. Saturation curve is also great for quickly lowering the saturation of elements that are too distracting, such as an overly colorful blue shirt, a vividly painted red car, or an insanely reflective magenta life-jacket (all situations I've fought with).
Overall, it's a useful tool for enhancing color contrast, bringing out colorful elements that can't quite compete with the rest of the scene. For example, if the green grass of a summertime shot isn't grabbing you, you can pump it up using the Hue vs. Sat curve to give it a more vibrancy.