By the numbers is a phrase associated by many with Dan Margulis's method of relying on numerical means, when possible, for determining the color correction adjustments made on an image. It means correcting an image based on numeric readouts instead of the visual appearance of the image. Many of the features of Curvemeister are intended to make the application of Dan's techniques easier and faster.

In general, correction by the numbers means that first, shadow, highlight, and neutral values are set using data from the image, or visual cues based on that data. Then neutrals are set, again relying on numeric values, rather than the appearance on screen of the image.

That said, Curvemeister, in an attempt to please everyone, includes several features that do not contribute to its by the numbers roots, including the options to use a master channel, and the ability to display a histogram in the curve interface.

In Photoshop and Curvemeister, a channel is a block of image data, in the form of 8 or 16 bit integers, that contains an image color or brightness value. All channels taken together describe the appearance of the image.

The meaning of the data values in a channel depends on the color space. For example, an RGB image is composed of three channels: red, green, and blue, and possibly a fourth channel containing transparency information. Other color spaces, such as Lab and CMYK

Each channel has its own curve in the curve window . In Curvemeister, a channel may be easily viewed as if it were a grayscale image, using the tabs along the lower edge of the image preview window , or by typing ctrl-1 to view channel 1, ctrl-2 to view channel 2, etc.

In this document, channel information for a single brightness value - normally corresponding to a single pixel - is represented as follows, for each color space:

Color Model is a widely used term for the numeric model used to represent the brightness and color of each point, or pixel, in an image.

Color Space, as used in with Photoshop, is a slightly more refined term, and means a color model together with a profile that characterizes the exact appearance of colors within the color model.

Because Photoshop tends to hide this distinction, the terms Color Model and Color Space are used interchangeably in this manual. For some people, particularly for those whose work involves specialized use of profiles, the difference is important.

For example, the RGB color space, commonly used for computer monitors, represents each brightness value as a combination of red, green, and blue brightness values. Lab represents brightness and color information independently. HSB describes colors in yet another way.

Each color space can have strategic advantages that depend on the particular image you are working on

In Curvemeister, you may quickly choose which color space you want using the color space radio buttons.

Color spaces are discussed in more detail in The Color Space Matrix.

A systematic undesirable color. A color cast degrades image quality by reducing the saturation of the image's colors and giving it an overall drab look. Next to setting a good shadow and highlight value, removing the color cast is the most important way to improve your image's appearance.

Most photographs begin life with a color cast, however slight it may be, and therefore you have an opportunity to improve an image simply by removing the cast. The easiest way to spot a color cast is to move the cursor over a known neutral object, and read the RGB value. If the values are different you have a color cast, and the image will probably be improved by adding a neutral point at or near that location.

The process of using a desired color value, called a Color Pin, together with an existing image color value, to create curve control points. This is normally accomplished by dragging a pin from the Pin Palette to an appropriately colored object in the Image Window. For example, a flesh colored pin may be dragged to a person's face to correct the flesh tone.

Shadow, highlight, and neutral points may also be defined by dragging pins.

See also the discussion of Color Pinning in the manual.

This an older term for Pinned Sample.

An area in the image window for which a measuring location has been defined, usually by alt-clicking on the Image Window. Eyedropper sample points are used for monitoring the image values at a particular point in the image, or for constraining image values by adding one or more control points to the curves.

An eyedropper sample may be converted to a pinned sample using the menu in the sample palette.. Constraint points reside in the image window, and are similar to Photoshop's eyedropper sample points.

GCR stands for Gray Component Removal, and refers to the amount of red, green, and blue that will be converted to the K channel when converting to the Wide Gamut CMYK color space. GCR thus controls how heavy the black channel will be. If the original or native image is Lab or HSB, an RGB conversion is performed first, and GCR is calculated from the resulting RGB values.

In Curvemeister the GCR value indicates the maximum percentage value that may appear in the K channel when the image is converted to wgCMYK.

GCR may be varied from 0% to 100%, in 25% increments. There is also an option to use the black generation curve that Photoshop uses for its normal CMYK conversion, as specified in Photoshop's Color Settings dialog.

A value of 0% effectively eliminates the K channel from the calculation, with the result that wgCMYK is operationally equivalent to RGB. A value of 100% maximizes the value of the K channel, and therefore guarantees that at least one of the CMY channels is zero.

A lighter GCR value is generally more useful for correcting an image, since it provides more control of shadow values, makes it easier to eliminate color casts, and provides more overall control of color.

A heavier GCR value makes it easier to make large changes to the CMY color channels without causing an overall color cast.

Highlight and shadow are the darkest and lightest parts of the image that should contain important detail, but contain no color information. For example, in a portrait white clothing usually contains the highlight , and a deep shadow under the eaves of a house would be a good example of a shadow.

The key to a highlight or shadow is the presence of important detail. A bright glinting reflection from a metal object is not a suitable as a highlight point, because although it may appear white, it does not contain any detail. Similarly, a pure black unlit window may not be a valid shadow point, again because it contains no detail.

Most images have both a highlight and shadow, but not all. For best results, it is important for you to distinguish these two situations, and several of the tutorials are devoted to this subject.

Histogram is a general term for a graph or diagram showing a graphic tally of pixel values in an image. It is often used to assess the overall spread of data in an image.

For example, an underexposed image will show a histogram whose data is, generally speaking, crowded on the dark side of the graph. Histograms can be misleading, however, and it is always better to look at the image itself, rather than relying on the histogram.

In Curvemeister the histogram is displayed inside of the curve area, sharing the horizontal axis with the graph portion of the curve interface, and showing the relative number of pixels at each value.

Because most scenes consist of three dimensional illuminated objects, interpreting a histogram is usually less reliable than looking at, and measuring, the image itself.

As an example of how a histogram can be deceiving, consider two images, one of a a red disk and the other a red sphere, both against black backgrounds. A front lit red disk against a black background will show a peak in the red histogram, centered around the red value at the center of the disk. The sphere, if lit from the front so that the edges fall off, will show an inverted shape with two peaks.

On balance, histograms are like vice grips. Many people use them and like them, but the availability of histograms has probably done more harm than good. The moral is to always be careful of the relationship of the histogram to the image, and rely more on the image than the histogram, particular for selecting the highlight and shadow points.

An optional curve that, as its name implies, controls image data data from all the channels. For example, the master curve of RGB specifies a multiplier for each of the R, G, and B channels. The display of the Master Curve is controlled using the View>Curves>Master Curve menu item.

The master curve is available for RGB and wgCMYK modes, and functions as a global data modifier that affects data from all the channels equally. It provides the quickest way to alter the overall brightness while in RGB or wgCMYK, particularly if the changes are reasonably subtle.

The master curve does present the risk of causing a pesky color shift in your image as a result of overusing the master curve. For this reason, Curves provides you with the option of working without the master curve. Yet the master curve remains the quickest way to alter the overall brightness of the image, particularly if the changes are reasonable subtle.

For all intents and purposes, the L channel of Lab, and the B channel of HSB function as a master channel, but without the problem of a potential color cast.

A brightness value near 50% gray, or a portion of the image with such a brightness value.

There are five general terms used to describe image brightness values: shadow, quartertone, midtone, three-quartertone, and highlight.

The midtone corresponds to the middle value of a well-exposed, average image. Foliage, dark blue sky, and the well-know 18% gray cards are normally represented as midtones.

A type of Constraint Point that is intended to force that point on the image to be neutral. This is done by adding curve points to the curves for the current working space.

Note: in Lab mode only one neutral point is used.

In Curvemeister, Pinning is short for Color Pinning.

Pinning also describes the operation of adding multiple control points to a curve, or a control point to more than one curve.

Pinning is used to automatically lock down its current position, or to match some other requirement. Examples are locking down the upper half of a curve, or making the color value neutral, or matching a color that you select using the color picker.

A Pinned Sample is an eyedropper sample point that has been set to force a particular color value by automatically creating curve points.

Highlight, Shadow, and Neutral points are examples of constraint points, but this term is most often applied to a sample created using a Color Pinning operation. You may also use the commands in the image window right-click menu to pin a particular image color to match any selected color, or Photoshop's foreground or background colors.

Because constraint points are retained when you switch color spaces, they are a good way to leverage your work across color spaces. For example, you may compare the effect of setting a particular neutral point in RGB mode versus Lab mode, and choose the more pleasant result before clicking the OK button.

See the discussion of Pinned Samples in the manual.