Color management

Learning goal

You will be able to explain why the same photo looks different in color on a phone, a computer, a web page and a print, by following how a file's colors are translated from device to device. Remembering that the color numbers in a file don't describe a color on their own, and that a profile gives them their meaning, you will see how a color management system converts these numbers from the source profile to the target device's profile. You will know that every screen and every printer and paper pair has its own device profile, and tell apart assigning a profile to a file and converting it to another profile. You will compare the rendering intents that decide how out-of-gamut colors, which the target device can't show, are handled in conversion, and learn how saturated skin in warm light can break down as it spills out of gamut and how to manage this.

The question

After the monitor was calibrated, the wall print began to match the screen. But when the baker opens the photo on their phone, they see a completely different frame: the orange light from the oven's mouth is almost fluorescent, the baker's cheek leans toward red. When your friend from the club uploads the same file to the bakery's web page, this time the opposite happens: the colors are faded, dull, as if a thin gray veil had been drawn over them. On the calibrated screen on the computer, on the other hand, the frame is measured and warm.

The file is the same, the pixels' numbers are the same; the only thing that changes is the device showing it. Your friend asks which one is the real photo. Why do colors change from device to device? Can a file's color stay the same on every device that opens it?

Context

In the color profiles lesson we saw a color space as the scale that defines which colors the color numbers in a file actually correspond to, and discussed the difference between sRGB and a wider space and the importance of embedding a profile (Lesson 11.9). In the monitor calibration lesson, too, you saw the screen being measured and tied to a profile (Lesson 25.1). This lesson brings those pieces together: how are a file's profile, the screen's profile and the printer's profile connected to each other inside a color management system?

Color management is a topic most photographers find complicated and ignore. Yet its basic idea is simple: every device speaks color in its own language, and color management translates between these languages. When the translation is done correctly, the same photo looks close on different devices; when the translation is skipped or done wrong, colors shift.

Simple explanation

How does color management translate?

Each pixel in a file carries a number for each of the three color channels. But a number such as "two hundred in the red channel" doesn't describe a particular color on its own; the file's color profile says which red that number corresponds to. The same number means a measured red in a narrow space, and a much more saturated red in a wide space.

A device profile is a profile, created by measurement, that defines which colors a screen or a printer and paper pair produces and how. The screen profile created by calibration and the printer profile the print shop prepares for each paper are both device profiles. Every device shows the same number as a different color according to its own makeup; a phone screen tends toward saturated colors, while a paper may print a more limited range of colors.

A color management system is the software arrangement that makes color look as alike as possible from device to device by translating a file's color numbers from the file's profile to the target device's profile. The translation is done via a shared, device-independent scale: first the file's numbers are translated into the real colors on this shared scale, then these colors are converted into the numbers of the target device, for example the screen or the printer. A color-managed application does this translation. An application that doesn't do color management, or a device that doesn't know its screen, on the other hand, sends the file's numbers as they are; a file saved in a wide space looks faded in such a place, and a file in a narrow space looks too saturated on a wide-gamut screen. The fluorescent orange on the baker's phone and the gray veil on the web page are two faces of the translation being skipped somewhere.

What is the difference between assigning and converting?

The assign profile operation is giving a file a new profile without changing its color numbers, so that the same numbers are read as different colors. Converting to a profile is the opposite: recalculating the numbers according to the new profile so as to preserve the colors' appearance as much as possible. This difference is the most often confused point in color management.

In a file that has no profile at all, or that comes with the wrong profile, assigning is a correction: if it is known which space the file was actually saved in, the right profile is assigned and the colors fall into place. But assigning another profile to a file that has the right profile breaks the colors; because the same numbers are read on a new scale, skin shifts to orange or gray. When you need to move from a wide space to sRGB to send a file to the internet, the operation to do is not assigning but converting. The right rule is this: if you want to preserve the file's appearance, convert; if you know the file is wrongly labeled, assign.

This distinction is useful in printing too. The print shop often says which space the file should be sent in, or shares its own paper's profile. Rather than converting the file to the printer profile yourself, sending it in the requested working space with the profile embedded makes sure the conversion is done in one place, in the print shop's color management system. Converting the same file twice can accumulate small losses at each step.

How are out-of-gamut colors managed?

An out-of-gamut color is a color present in the source file but outside the colors the target device or paper can produce. A very saturated orange edited in a wide space can't be carried over as it is when converted to sRGB or to a paper; the color management system has to place it somewhere inside the target.

A rendering intent is the choice that decides how out-of-gamut colors will be placed inside the target gamut in conversion. The perceptual intent brings out-of-gamut colors inside by compressing all the colors a little together; the relationships and transitions between colors are preserved, but the whole frame may look a little less saturated. The relative colorimetric intent, on the other hand, doesn't touch the colors inside the gamut and cuts only those left outside to the nearest boundary color; most colors are preserved exactly, but transitions can flatten in out-of-gamut areas. Which is better depends on the photo; in a frame full of saturated colors the perceptual intent, and with measured colors that stay inside the gamut the relative colorimetric intent, often gives a better result.

Skin is especially sensitive in this choice. In the warm light coming from the oven's mouth, the saturated red-orange tones on the baker's cheek can spill outside the paper's gamut. With the relative colorimetric intent this area is cut at the boundary, and a flat patch of color can replace the fine transitions on the cheek. The reddish and brown undertones of dark skin can fade and lean toward gray when converted to a narrow space; the pink tones of light skin, on the other hand, can look flushed on an overly saturated screen. Turning on the gamut warning to see which areas of skin are out of gamut, and comparing the two rendering intents on the face, makes it possible to preserve the skin's detail and natural tone.

Visual evidence

You look at the file's profile in the editing program: it was edited in a wide working space. When you open the copy you uploaded to the web page, you see the profile wasn't embedded; the page took the numbers to be sRGB and the colors faded. This time you convert the file to sRGB and save it with the profile embedded. On the page, the colors come close to those on the calibrated screen.

You go back to the fluorescent orange on the phone. The phone's vivid color mode is on; when you switch it to natural mode, the orange comes down to a measured warmth. This time it was the screen's setting, not the file, that needed touching.

For the print, when you turn on the gamut warning, you see that the most saturated orange area on the baker's cheek is outside the paper's gamut. With the relative colorimetric intent a small, flat patch forms on the cheek; with the perceptual intent the transition is preserved and the skin calms down a little without losing its warm brown. You send it to print with the perceptual intent. The same file now looks very close on all three devices.

Common mistakes

Summary

A file's color numbers take their meaning from its profile, and a color management system translates these numbers from the file's profile to the screen's or printer's device profile; when the translation is skipped, the same file looks too saturated on one device and faded on another. The assign profile operation gives a new scale without changing the numbers and is used only on wrongly labeled files; to preserve appearance, the file is converted to a profile. Out-of-gamut colors the target can't produce are brought inside the gamut with a perceptual or relative colorimetric rendering intent, and on saturated skin in warm light, which one preserves detail and natural tone is checked with the gamut warning.

Mini assignment

In this assignment you will compare how the same photo looks on different devices and in different conversions.

1. Open a photo with vivid colors, preferably one with someone's face in the frame too, on a phone, a computer and a web page, and write down the differences.
2. Check the file's profile; if the profile isn't embedded, convert it to sRGB, save it again with the profile embedded, and compare how it looks on the web page.
3. Deliberately assign a wrong profile to one copy, and convert another copy to the same profile; put the two results side by side and explain the difference.
4. Turn on the gamut warning, and if you have a printer or paper profile, try the two rendering intents; note how the out-of-gamut areas on the face and the skin's natural tone change.

Check yourself

• Can you describe how a color management system translates a file's numbers?
• Can you say what the difference between assigning a profile and converting to a profile is, and which is used when?
• On out-of-gamut skin, did you check which rendering intent preserves detail and natural tone?

Glossary

Mini quiz

1. Which red does the number "two hundred in the red channel" in a file correspond to?

• A) Always the same red
• B) A different red in a narrow or wide space, depending on the file's profile
• C) The printer's red
• D) The screen's brightest red

2. What operation is done when moving a file edited in a wide space to sRGB for the internet?

• A) The sRGB profile is assigned
• B) The profile is deleted
• C) The rendering intent is turned off
• D) The file is converted to sRGB and the profile embedded

3. In a frame full of saturated colors, which rendering intent is usually more suitable for bringing out-of-gamut colors inside while preserving transitions?

• A) The perceptual intent
• B) The relative colorimetric intent
• C) Assigning a profile
• D) Neither; out-of-gamut colors are deleted

4. With the relative colorimetric intent, a flat patch of color forms on the cheek of the baker in the oven's light. What is done?

• A) Leaving the patch as it is
• B) Assigning another profile to the cheek
• C) Seeing the area with the gamut warning, comparing it with the perceptual intent, and choosing the intent that preserves the skin's transition
• D) Turning the photo into black and white

Answers

1. B — Color numbers don't describe a color on their own; the file's profile says which color it is.
2. D — To preserve appearance, the numbers are recalculated for the new space and the profile is added to the file.
3. A — The perceptual intent compresses all colors together; the transitions of out-of-gamut colors are preserved.
4. C — On out-of-gamut skin the intents give different results; comparing them on the face makes it possible to preserve detail and tone.

Summary card

• Color numbers take their meaning from the profile.
• Color management translates from the file profile to the device profile.
• Convert to preserve appearance; assign to fix a wrong label.
• Send to the internet a file converted to sRGB with the profile embedded.
• On out-of-gamut skin, compare the perceptual and relative colorimetric intents on the face.

Bridge to the next lesson

Once colors fell into place across devices, the baker wanted something bigger: a print a meter wide for the shop's back wall. The print shop asked how many pixels the file was and how many centimeters the print would be. Your friend's file was six thousand pixels wide; they didn't know whether this would be enough for a meter. Someone also asked from what distance the print would be viewed. How big can you print a photo? The next lesson deals with resolution, viewing distance and print resolution.