Color management and monitor calibration.

1. Introduction

1.1. Goal
The goal of this article is to shed some light on a subject that is often not understood by novices. For this reason it is kept simple and more difficult aspects are left out. For experienced users that would like to dig deeper i have added some references in section 5.

1.2. Why do we need color management?
For many photographers color management is just some vague principle for purists. While it is commonly known that in the old days, we could change the look of the photo by choosing another film with a specific look. With different development methods, we could adjust the look further. Usually that work was done by the photo lab and we got a photo that looked just like we intended if they did the development right. Nowadays we shoot digital, so there should be little room for errors. We can adjust the colors ourselves with software and can produce the perfect photo with just a slightly different white balance and adjusted saturation. In practice, however, we may come across problems with colors that just don’t look right. The reason for this is that many devices, like your camera, monitor and printer simply can’t deliver all possible colors. So, if we want to prevent wrong colored photo’s, we must get some knowledge about what can go wrong and how to solve it. When sharing photo’s on the net, colors are most influenced by your monitor, as the camera has a larger color range and the printer isn’t used for publishing on the web. This monitor can be calibrated, so the presented colors are as intended. To understand why the calibration is needed, the aspect of color space must be explained first.

1.3. How to read the article
In the next section the concepts of color management and color space are explained and why this can cause wrong colors under certain conditions. This is only a brief explanation and most can be found on the world wide web, as I did too. After that i will continue with a way to calibrate your monitor yourself. For those that are only interested in results – just skip section 2 "Color Management" and jump to section 3 “Monitor Calibration” and section 4 "Summary" after that.


2. Color management

2.1. How is color made and what is color management?
The image sensor in a camera consists of photo sites with a colour filter in front of it (Bayer pattern filters are used most), which make it possible to capture an image of primary colors red, green and blue. With these three primary colors, every other color can be made with mixing different amounts of red, green and blue (hereafter R,G,B) and that is exactly what happens in the normal color photo. Your printer makes colors in the same way, although there usually a CMYK space is used as white paper is the base of an image - instead of a black screen. Color management involves the adjustment of the color spaces of the used equipment for images. Color spaces will be discussed in section 2.3.

2.2. Color depth
The color depth is usually given in a number of bits, say n. In this case the available colors are discretized in 2 to the power n different shades - for each primary color! A popular color depth is 8 bit, which is 256 colors or 0,1,2,...,255. Other used depths are 12 bits which means 4096 shades, 14 bits which is 16384 shades and 16 bits means 65536 shades. A more simple explanation of color depth is: the higher the number, the more different shades can be distinguished. The color spaces which will discussed hereafter are assumed to have 8 bit depth. The idea remains the same with higher color depths, there are just more shades in between when the color depth gets larger. See ref. [13] for examples.

2.3. Color spaces
The range of possible colors that a device can deliver, can be presented in a color space. In fact this is just a map with primary colors (red, green and blue for the RGB color space) and all the mixed colors. The further to the outer lines of the color space, the more saturated the color becomes. The maximum number is 255 for each color (8 bit color depth). For example, the number (255,0,0) means maximum saturated red, no green and no blue in the RGB color space. Because not all devices are capable of reaching all colors, a color space or gamut of colors -that can be delivered with that device- can be defined. In case we connect devices, like our camera to our monitor or printer – it becomes necessary to check whether the color spaces match.

Standardized color spaces are defined by The International Color Consortium (ICC). ICC supports a variety of device-dependent and device-independent color spaces which can be divided into three basic families:

• CIEXYZ based (Wiki CIEXYZ color space)
• RGB based      (Wiki RGB color model)
• CMY based      (Wiki CMYK color model)

The color model, which is a model where colors can be described with number combinations, together with a mapping (a definition of mapping can be found here or one could consider mapping as connected fields in a lookup table) to an absolute color space is called a color gamut. Some color spaces are shown in figure 1a. Obviously this is a three dimensional model, as there are three primary colors. It is possible to calculate the new coordinates in case of a transformation from another color space, see ref. [2] for example. Considering the goal of this article, i will not go into detail here.


Fig. 1a. Three dimensional color spaces. Here the
wired space is Adobe RGB and the solid one sRGB.
Note the visible spectrum is not shown here.




Usually a two dimensional version or projection is used however, as this obviously simplifies the matter, see figure 1b. The gamuts as shown here clearly show the primary colors and triangles. Inside these triangles all the defined colors can be found by mixing these primary colors. Colors that cannot be expressed within a particular model are said to be out of gamut. The numbers on the edges of the full spectrum are the wavelengths of the primary colors.
If we look at the Adobe RGB gamut in figure 1b, we can see the coordinates of the corners of the triangle (most saturated values) are for Red (x,y)=(0.64,0.33), Green (0.21,0.71) and Blue (0.15,0.06) . These represent the colors with numbers R=(255,0,0), G=(0,255,0) and B=(0,0,255). The values in between can be calculated with the calculator in ref. [2].


Fig. 1b. Several color spaces. As can be seen red with value 255 (or 255,0,0) in the
LAB color space is much more saturated than red 255 in the CMYK space. Note that your
monitor is probably sRGB so cannot display the colors outside this space here faithfully.





2.4. One color number, different colors
In fig.2 two colors with a value of (R,G,B)=(88,249,17) are shown. These numbers represent a certain color in a color space. Without knowing which color space, the actual color is not precisely known. As can be seen, the mentioned color coordinates refer to a well saturated green color in Adobe RGB, but a more flat and pale color in sRGB. This clearly illustrates what can go wrong in case the wrong color space is used.    

Fig. 2 Two shades of green in different color spaces, both (R,G,B)=(88,249,17).
Left is the color with these values in Adobe RGB, right in sRGB (source: ref. [10]).





2.5. Color temperature
Now let’s suppose all the requirements to communicate with our equipment have been met optimally. The monitor has the right color space, still the presented colors may be wrong. The reason may be the color temperature. As can be seen from figure 1, the combination of all colors in the center of the color space will result in a certain shade of white. The color is said to have a certain color temperature. Too much blue will increase the color temperature, for instance. As contrast is dependant of the light distribution, the amplitude – or strength- of the colors will also influence the contrast. The curve with all color temperatures in the visible color spectrum is shown in figure 3. The concept of color temperature can be easily understood: imagine a flame. A flame from a match is not that hot – it is yellow. A very hot flame from a plasma torch is towards blue. You can imagine this will be hot. Hotter than the yellow flame, so the more towards blue, the higher the color temperature. This temperature is given in K (Kelvin), which is the same as Celcius with added 273 to the scale.
Most monitors are not made for photo editing. For CAD work and gaming, high contrast is important and for this a color temperature of 9300 K is often chosen. Note that the color space still may be well matched, the color temperature merely specifies the content of certain colors, not the maximum reachable values. The color temperature in art galleries and museums is often 5000 K – this looks slightly yellowish, best color temperature for photo editing is around 6500 K.
Getting the right color temperature in different ambient light with your camera can be achieved by the correct white balance. Also see ref. [14],[15],[16].


Fig. 3. Color gamut with white color temperature.
Note the color space presented is the visible spectrum.



2.6. Methods for calibration
As we have seen in the introduction, for the use with a digital camera and viewing on our monitor (and web), the best starting point is to calibrate the monitor. The following methods can be used (there may be more methods than mentioned here; the links point to pages that show some aspects of the software, the information found here may not be fully complete):

• Adobe Gamma
• SpectraCal's CalPC
• DataColor Spyder
• ColorVision Color Plus
• Monaco Systems Monaco Optix (no longer available, now: Calibrite)
• Gretag Macbeth Color Checker

As an example a calibration with the DataColor Spyder 4 Pro will be showed in section 3.2.


3. Monitor calibration in practice

3.1. Adjusting colors
Instead of using the standard videocard or monitor settings, one could change these settings by adjusting the coarse dials for red, blue green plus a contrast dial. Save for some situations, one just can’t get the presentation right with this method. Monitors on laptops don’t have any hardware dials on them, it can be tuned with software dials sometimes. The right solution should be measuring the output of the monitor by a device of which the characteristics are known (obviously lineairity is preferred, but not needed). This can be compared to the correct output. A pickup device can view the output and this can be compared with a database in software. From the deviations, a correction signal can be found and send to the monitor, so an optimal result can be seen at the monitor. Most calibration software packages can do this automatically with some user input.

3.2. An example of a monitor calibration
Now the working of the DataColor Spyder software will be discussed to give an idea of the method. For a good calibration, it is recommended to leave the monitor on for at least 30 minutes. The reason is that colors may change during the warm up period. The software takes you through some questions, see figure 4. Note some monitors with reflective glass may not be well suited for a good calibration, and monitors exist with bigger color gamuts, for better possible color reproduction.


Fig. 4. Starting up the calibration software




Next the monitor type must be given, see figure 5.


Fig. 5. Screenshot DataColor software.



After this, the pickup device can be put on the monitor, while it is connected to the computer through the USB port. The software presents the pickup device a large number of different colors and the response will be stored in a table. Also a number of different shades of white will be processed for optimal white balance. A screenshot of that can be seen in figure 6.


Fig. 6. DataColor screenshot. The spyder sensor measures the color as presented by the monitor.
The color was green while taking the screenshot, of course all colors are tested.
Note the progress bar under the image, total time is around 15 minutes.




After all colors are done, the pickup device can be disconnected and the correction table is complete. After correction the output on the monitor should now be optimal. For reference, the measured color space is shown together with standardized color spaces sRGB, NTSC and Adobe RGB.
The calibration process must be repeated after a certain period, usually a few weeks to 6 months as the equipment can slightly change during time.


Fig. 7. DataColor measurement results.
Note the colorspace is compared to RGB, it is close to sRGB.




To give an example of the situation for my monitor two screenshots are shown, the first before calibration, the second after that. The image is a generic background as can be found on InterfaceLift and chosen for a good visibility of the effect of a calibration.


Fig. 8. Windows background before calibration - note the cold colors.




Fig. 9. Windows background after calibration.





4. Summary

Color management might look complicated at first, but hopefully this short article clears up the matter a bit. A certain color can be given by coordinates (R,G,B for RGB color space). As the size of the colorspaces are different and 255 is the highest number (for 8 bit depth), the presented colors may be misinterpreted in case the accompaning color space is not known. In practice, RGB and sRGB are often used. So, check what color spaces suit your equipment. After that, only a good white balance is needed to result in an image that is just like intended.


5. References

For collecting information i used several sources as mentioned here. Of course more can be found, consider this as a starting point.
[1]   http://en.wikipedia.org/wiki/CIE_1931_color_space
[2]   http://www.easyrgb.com/index.php?X=MATH&H=19#text19
[3]   http://en.wikipedia.org/wiki/Color_space
[4]   http://en.wikipedia.org/wiki/Absolute_color_space
[5]   http://www.cambridgeincolour.com/tutorials/color-management1.htm
[6]   http://fstoppers.com/adobergb-vs-srgb
[7]   http://www.photographybay.com/2007/04/12/the-mystery-behind-color-spaces-adobe-rgb-vs-srgb/
[8]   http://www.cs.rit.edu/~ncs/color/t_convert.html
[9]   http://dpbestflow.org/color/color-space-and-color-profiles
[10] http://www.drycreekphoto.com/Learn/color_management.htm
[11] https://babelcolor.com/tutorials.htm#TUTORIALS_RGBspaces
[12] http://www.brucelindbloom.com/
[13] http://www.photoshopessentials.com/essentials/16-bit/
[14] http://www.slrlounge.com/tutorial-understanding-white-balance-color-temperatures-in-8-simple-steps
[15] http://www.cambridgeincolour.com/tutorials/white-balance.htm
[16] http://www.exposureguide.com/white-balance.htm


6. Acknowledgements

I would like to thank the Dyxum AMAE team for the proofreading and all suggestions for improvements.

Edited by MichelvA - 16 January 2023 at 07:45

Great work Michele , thanks .

Many thanks, Michele. Very interesting work.

Best wishes.

Thanks Michele - an interesting topic well written. But I do have a suggestion...

When I am presented with a graph without labelled axes my brain seizes up. (And I don't think I'm alone in this.) Could you either label the axes in the graphs (especially Fig 1b) or explain what they are in the text?


Cheers
Tony

Here you go and thanks - all.

I've decided to do more printing so could anyone recommend one of the screen calibration tools eg Color Monkey, Spider *** or whichever works as reading the Amazon reviews they all seem to have their problems.
I recall David Kilpatrick saying that as they all give different corrections none of them were any good !
All advice appreciated
Many thanks
RT

Hi there, I just calibrated my Asus laptop (ips-screen) and my Dell monitor with the Datacolor Spyder 4 but next to each other the display doesn't look alike with the Asus being a tad warmer. Both profiles are loaded. I thought after calibration they would be the same. Am I missing something?

Edited by KoBra64 - 22 November 2014 at 12:08