next up previous contents
Next: Linear Scale-Factor Methods Up: Display Devices Previous: Slides and Goldberg Rule


CRT monitors are the most widely used display devices in computer graphics. A CRT (cathode ray tube) is the essential part of all CRT monitors. It contains three electron guns which emit narrow streams of electrons. The streams pass through deflection coils that band the beam up, down, right and left. The inside of the front face of the CRT is coated with three different phosphors, each of them emits light of certain wavelength when struck by electrons. The intensity of light depends on the number of incident electrons. There are various phosphors, but red, green and blue emitting type phosphors are used in common monitors. As all displayable colors are formed from an additive mixture of red, green and blue, it is clear that the color gamut of a CRT monitor is a subset of all visible colors.

We will examine some CRT monitor characteristics next. Just as before, we are interested in the monitor response to various input values. Unfortunately this response is not linear. For a CRT with N displayable levels, the luminance of grade i, tex2html_wrap_inline4971 can usually be approximated by:


where tex2html_wrap_inline4973. The value of tex2html_wrap_inline4975, of course, is slightly different for each color channel.

We have measured tex2html_wrap_inline4971 for the red, green and blue color channels of an SGI color monitor. The results are shown in figure 3.4. It can be seen that default gamma correction is not perfect, but it is close to linear response. For most applications it will be sufficient.

Figure 3.4: Red, green and blue gun response after tex2html_wrap_inline4951-correction

Although it seems in (3.4) that the CRT contrast is infinite (the lowest luminance value is 0), there is no real 0 luminance, i.e. the CRT emits some radiance even for i=0, (if the measured characteristics in the figure 3.4 are observed carefully it can be seen that there is some emittance even for 0 input level (best seen for red and blue guns), and there is always some reflection from surrounding light on the screen [ToHe89]. If you look at the switched off display in a lit room you will see it is far from totally black. This unwanted luminance level that is proportional to the ambient light is added to the image. It means that in (3.4) a constant proportional to the ambient luminance level should be added. The ambient luminance decreases the CRT contrast and the gamut as well. Figure 3.5 illustrates the problem. It shows how gamut changes with the increase of ambient light.

There is a lot of research done on the calibration of the CRT monitor [Durr87], [Hall89]. Actually the characteristic changes as time passes, it depends on room temperature and even on the orientation of the monitor (influence of the Earth magnetic poles [Fitz89]). All of these influences are relatively small, and they are not so important for the common user.

Figure 3.5: Chromatic gamut of CRT for ambient illumination covering a range of five log units. [Layc83]

The CRT contrast depends on the ambient illumination level, on user settings, and on particular device characteristics. For the same CRT the contrast can vary from 20 to 100. The best would be to measure the particular CRT and to work in a room with a constant illumination. Of course, in this case the device settings once set, should not be changed any more. The Goldberg-Gamma should be taken into account for CRTs as well. In dark rooms the Goldberg-Gamma should be set to 1.5, in dim surroundings to 1.2 (default setting for TV). When a CRT is viewed in very bright surroundings the Goldberg-Gamma decreases to 1. Goldberg-Gamma correction can be done simultaneously with display gamma correction. Instead of tex2html_wrap_inline4975 in (3.4), tex2html_wrap_inline4983 should be used. More exact relationships than the Goldberg rule are described in the RLAB color space for luminance and chromatic adaptation [Fair94].

Figure 3.4 shows the measured CIE Y of the red, green and blue color channel. It can be seen on the y axis that the maximum values of Y are not the same. The blue gun emits the lowest maximum intensity, and the green gun the highest. If we are interested in computing an equivalent luminance image from our rgb image, then each of the color channels should be properly weighted. For this particular monitor, the weights can be computed from maximum intensities as:
For our monitor, the weights are tex2html_wrap_inline4985, tex2html_wrap_inline4987 and tex2html_wrap_inline4989.

We have measured chromaticies CIE x and y values for our monitor as well. It is interesting that chromaticies for a particular phosphor were not constant as expected. The measured CIE x,y chromaticies are shown in figure 3.6. We suppose the reason lies in the fact that there are some cross effects among phosphors. There can be also some error caused by the measuring instrument at low intensities. Figure 3.6 shows gamuts for intensities 10, 30, 50, 70,..., 250. The chromaticies change decreases with higher intensities.

Figure 3.6: Change of phosphor chromaticies with increasing intensities

Note that the weighting functions computed in eq. 3.5 using maximum intensities, can be computed from chromaticies as well. Each phosphor has a dominant wavelength, and the CIE photometric curve gives a weighting factor for each of the three phosphors. Assuming our phosphor chromaticies have CIE x,y coordinates and dominant wavelength values as given in table 3.2, corresponding weights are: tex2html_wrap_inline4991, tex2html_wrap_inline4993 and tex2html_wrap_inline4995.


red0.617 0.334 610
green0.278 0.614 545
blue 0.154 0.058 463
Table 3.2: CIE xy coordinates and dominant wavelength of three phosphor primaries


These results are similar to the results obtained using the maximum intensities. Errors are due to non-perfect measuring conditions.

next up previous contents
Next: Linear Scale-Factor Methods Up: Display Devices Previous: Slides and Goldberg Rule