Characterization of your Display
Why Do We Still See Color Errors?
A fundamental problem in color reproduction is that different devices exhibit widely different color characteristics. For example, the RGB values (170, 28, 214) will yield different perceived colors (i.e. XYZ values) when displayed on different monitors. The process of device characterization allows us to adjust the RGB values going into different monitors so that the resulting displayed colors all appear the same to the human eye.
To characterize an output device, some device values are chosen and the resulting output is measured, usually with a color measurement device such as a colorimeter or a spectrophotometer. Instead of measuring all possible combinations of digital values (which would yield over 16 million measurements), a representative set is chosen and a model and interpolation are used to estimate the remaining colors. This approach was used to characterize the press that prints the TIP magazine.
Visual characterization uses the human eye as a measurement device. The display in this demonstration was characterized visually. This method is less accurate than a complete measured characterization, but it is much faster and doesn't require any special equipment.
R = [(dr - offset_r) / (255 - offset_r)]gamma_r
G = [(dg - offset_g) / (255 - offset_g)]gamma_g
B = [(db - offset_b) / (255 - offset_b)]gamma_b
X R Y = M * G Z B
The visual matching tasks you just performed were used to estimate the offset and gamma terms for your display. The 3x3 matrix M is related to the display's phosphor primaries and white point. For this demonstration those were assumed to be constant across all displays, sRGB primaries and D65.
Phosphor chromaticities: These are the colors of the pure red, green, and blue primaries. They were assumed to be those used in the sRGB standard, ITU-R BT.709. Phosphor chromaticities do not vary much among displays and this is not a significant source of error in a visual characterization.
Flare: This is mainly due to additional light in the room being reflected off the display. It was assumed to be 3% based on measurements made off several monitors in average room lighting. Actual flare will be higher or lower depending on ambient lighting conditions. For a completely dark room, flare is close to zero.
Development of visual determination of these parameters is an area of active research, and generally involves more visually challenging tasks than the two presented here.
Press CMYK --> Visual Color --> Display RGB
We begin with the digital CMYK image that is to be printed on the press. The press characterization allows us to convert the CMYK digital values to the visual representation of the colors that appear in the printed magazine. This visual representation is then converted to display RGB digital values using the visual display characterization. These RGB values, when displayed on your monitor, yield a color match to the printed version in the magazine.
While this demonstration was intended to show that you can considerably improve color matches between display and print via device characterization, we don't expect that the reproductions will be perfect. Here are some possible reasons why.
1. Although we have tried to capture the nominal color characteristics of the press, there will still be variations in color from one copy of the magazine to another. Furthermore, the demo images were prepared using color measurements off the target in the April issue. If the press characteristics have changed between April and June, this could result in additional color errors.
2. As noted earlier, the visual display characterization you just performed only corrected for two of the parameters in the display characterization equations. The other parameters (i.e. white point, chromaticities of the primaries) were assumed to be constant across all displays, which is not the case. A more careful characterization takes all these parameters into account, but also entails a more laborious process.
3. An accurate color match requires that we know the exact conditions under which you are viewing the print and the display. The color and brightness of the lighting in your room can strongly affect the results. Again, for simplicity, we have made some nominal assumptions on these viewing conditions.
4. It is possible that some colors in the printed image are actually outside the range of colors that can be displayed. However, we don't expect this to be a large source of color errors.
5. Even if we can accurately capture the physical characteristics of the press and your display, there are complex visual effects that must be taken into account. Our visual system adapts differently to emmissive displays than to reflective prints. This subject is an area of active research in the color imaging community.