Recently, a reader posted a question on the Chroma-Tech Web site (www.chroma-tech.com) asking for an explanation of rendering intents - one of the more difficult to understand aspects of color management.

Most common desktop publishing software applications such as Adobe Photoshop, Illustrator, InDesign and Quark Inc.’s QuarkXPress, along with many RIPs, allow you to select rendering intents to be used when conversions using ICC profiles are performed.

Rendering intents are even used when we view images on our monitors. And wherever a rendering intent can be used, a rendering intent has to be used.

It’s not surprising, then, that questions are posed about rendering intents. What are rendering intents and which one do you use? The answer to that question is the same as the answer for just about every question we can think of: that depends.

Works with profiles

The use of rendering intents goes hand-in-hand with the use of ICC profiles. ICC profiles are used to convert images and graphics from one color space to another, such as from RGB to CMYK or from one CMYK space to another CMYK space.

One of the objects an ICC profile defines is the color gamut of the device. Gamut means all of the color that a device can capture, display or output. RGB devices almost always have a larger gamut than CMYK devices (see Figure 1, which illustrates the difference between a theoretical RGB and a CMYK color space.)

When you convert an image from one space to another, there may be some colors that one device is capable of reproducing that the other device is not.

In Figure 1, the three dots labeled 1, 2 and 3 represent three pixels whose colors are in-gamut while the three dots labeled 4, 5 and 6 represent three pixels whose colors are out-of-gamut.

A color that exists in the source color space but that does not exist in the destination color space will need to have its value changed into that of a color that is available in the destination space.

For example, let’s assume that I have a photograph of a flower that is vibrant red in color. Let’s also assume that my press cannot reproduce a red that saturated.

The red color of the flower (pixels 4, 5 and 6 in figure 1) will need to be adjusted or changed to a red that I will be able to print. The rules that govern how this adjustment takes place when using ICC profiles for conversion are known as rendering intents.

Four choices

Rendering intents come in four versions: perceptual, saturation and two flavors of colorimetric, relative and absolute.

Saturation is the least-used intent (see Figure 2). Notice that pixels 4, 5 and 6 happen to be in a straight line out from the center of the color space.

In the saturation rendering intent, because each out-of-gamut pixel is treated individually without regard to any surrounding pixel, the closest color to each of these three pixels that are out-of-gamut happens to be the same color.

Notice also that the three pixels that are in-gamut are also changed. These pixels are remapped toward the outside edges of the CMYK color space, effectively increasing the saturation of the image. Using the saturation rendering intent will bring all of the colors into gamut but will increase the saturation of all in-gamut colors during the conversion. Although this is useful when working with cartoons, comics, business graphics or when trying to improve a weak image, it will seldom be used in your day-to-day workflow.

Perceptual rendering

Next, let’s examine perceptual rendering intents, one of the most commonly used versions (see Figure 3).

When using this intent the entire gamut of the image is compressed to fit within the gamut of the destination device.

This means that all of the colors that were actually in-gamut will need to be adjusted to, in effect, make room for the out-of-gamut colors. Moreover, unlike what happened in the saturation rendering intents, with perceptual rendering intents all pixels are treated with respect for each other.

As a result, out-of-gamut pixels may not be moved to the closest reproducible color. For example, pixel 4 is moved well past the nearest reproducible color at the edge of the CMYK color space to allow room for pixels 5 and 6.

You may think that this adjustment, combined with the fact that pixels 1, 2 and 3 - pixels that could be accurately be reproduced - are changed, makes every color wrong.

That assumption, in fact, is correct. Therefore, don’t perceptual rendering intents make things even worse? The answer is no. That’s because all of the colors have been adjusted proportionally, even those that the destination device could accurately reproduce. The chance that a viewer will notice that all of the colors have been modified is minimized. This is the most common rendering intent that is used when converting from RGB to CMYK color space.

Colorimetric intents

This leaves us with the colorimetric intents, both relative and absolute (see Figure 4).

When using the colorimetric intent, all of the colors that are in-gamut (1, 2 and 3) are not compressed but are simply adjusted for accuracy.

Only the colors that are out-of-gamut are compressed, and each color is modified individually without regard for any other colors - much like the out-of-gamut colors were adjusted in saturation rendering intents.

Due to the fact that this disproportionate color adjustment can be noticeable, especially in blends and gradients, its use is normally limited to conversions between color spaces of similar sizes, such as a CMYK-to-CMYK conversion.

If you use the colorimetric rendering intent when converting to a smaller color space from a much larger colors pace, the resulting conversion could introduce banding, posterization and other artifacts into the image.

Relative or absolute?

If you opt to use the colorimetric rendering intent, you still must determine whether to use either the relative or absolute versions. These two rendering intents treat all colors the same way and only differ in their treatment of the white point.

When using relative colorimetric, the white point of the source color space is mapped, or changed, to the paper white of the destination. The result: White in the original image effectively remains white in the final output.

When using absolute colorimetric, the white point of the destination becomes the white point represented by the source profile.

To clarify further, imagine that your source profile was made from your press printing on newsprint stock. The white point in this profile will be the color of the newsprint stock itself, which is certainly not a pure white.

Subsequent printout

If the absolute colorimetric rendering intent is used to convert an image from this color space to, say, your proofing color space, the subsequent printout will have the newsprint paper color simulated over the entire image on the output. Although this would give us the most accurate proof, it is important that this image is never actually printed on the newsprint stock itself, because the simulated color of the newsprint stock would be printed on top of the newsprint stock, significantly muddying the image.

Therefore, the absolute colorimetric rendering intent is only used to simulate an output on your monitor or on your proofing system and should never be used for final output conversion.

Wrapping it up

This wraps up the lengthy answer to the simple question of “What is a rendering intent and which one do I use?”

Take comfort in the knowledge that once you have your head wrapped around the above concepts, you will understand what is possibly one of the most complicated aspects of color management.s

Editor’s note: In the next article, Nate will describe how to use Photoshop to ensure  you are selecting the best rendering intent for each image you’re converting.

John Nate is a color specialist for Chromaticity Inc., an integrator of color technologies for the newspaper and other industries. Nate can be reached via phone at 616.361.7773 or via e-mail at jnate@chromaticity.com.