National Academies Press: OpenBook

Video Displays, Work, and Vision (1983)

Chapter: 4. Display Characteristics

« Previous: 3. Radiation Emissions and Their Effects
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 66
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 67
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 68
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 69
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 70
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 71
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 72
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 73
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 74
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 75
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 76
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 77
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 78
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 79
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 80
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 81
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 82
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 83
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 84
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 85
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 86
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 87
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 88
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 89
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 90
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 91
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 92
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 93
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 94
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 95
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 96
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 97
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 98
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 99
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 100
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 101
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 102
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 103
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 104
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 105
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 106
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 107
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 108
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 109
Suggested Citation:"4. Display Characteristics." National Research Council. 1983. Video Displays, Work, and Vision. Washington, DC: The National Academies Press. doi: 10.17226/169.
×
Page 110

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Display Charactensffcs In this chapter we summarize and evaluate the known relation- ships between characteristics of video display devices and observer visual performance, subjective responses, and physio- logical responses. The chapter is divided into major sections on CRT display variables, pertinent display measurement techniques and associated problems, a comparison of flat-panel and CRT display characteristics, and characteristics and relative effec- tiveness of filters. For each of the pertinent display variables, we consider three categories of effects on human users: physiological effects, the effects of display variables on measurable and objective perfor- mance, and known relationships between display parameters and subjective estimates of display quality or related physical symp- toms. Physiological effects are those in which the display param- eter has a known, direct physiological effect on the human visual or other organic system. Physiological effects typically cannot be controlled by a user and are not necessarily recognized by a user. For the second category of effects, representative performance measures include speed and accuracy of performance. In the third category, the reported symptoms include subjective estimates of blurring of characters, headaches, visual fatigue, and musculo- skeletal discomfort. EFFECTS OF CRT DISPLAY VARIABLES Luminance Increases in display luminance have several direct effects on visual physiological and optical responses and visual performance. 66

67 Effects on Visual Acuity In general, increases in display luminance will cause decreases in pupil size, which in turn lead to increases in the optical depth of field and improvement in optical quality. Figure 4.1 illustrates reduction in pupil size as a function of retinal illuminance, assuming a uniformly illuminated retina. This increase in retinal illuminance, which causes a decrease in pupil diameter, directly affects the visual acuity of the normal healthy eye, as shown in Figure 4.2. While the differences are not very great over the normal display operating range, an increase from approximately 1 or 2 milliLamberts (mL) to about 60 or 70 mL causes an increase in visual acuity of approximately 50 per- cent. Thus, displays having higher luminance permit an operator to see finer details on the display. The greatest proportional gain in acuity with increasing luminance takes place between approxi- mately 1 and 10 mL. In general, a positive-contrast display (light characters on a dark background) will have a background luminance of about 1 or 2 mL, and a character luminance of about 25 mL, with a character density of approximately 30 percent. This combination produces a display having an average (adapting) luminance of about 6 or 7 mL. By comparison, a negative-contrast display (dark characters on a light background) will have a background luminance on the order 8 7 - _ 6 cr LL LL ~ J I: J 4 3 `jI Tat_ 2 a, Dark -1 0 1 2 3 4 5 OG R ETI NAL I LLUM I NANCE (trolands) -1 0 1 2 FIGURE 4.1 Diameter of the pupil as a function of retinal illuminance. SOURCE: ten Doesschate and Alpern (1967~.

68 ~ .6 1 .4 1.0 - c' J 0.8 6 - - cn > 0.6 0.4 0.2 o ·f ·l- :~t . it. ·, ·~ 1 1 1 1 1 1 1 _5 _4 _3 - 2 -~' 0 1 2 3 LOG L (mL) - FIGURE 4.2 Relationship between visual acuity and adapting luminance. SOURCE: Hecht (1934~. Of 25 mL and a character luminance of about 1 mL, producing an average (adapting) luminance of about 17 mL. Accordingly, one might expect an increase in relative acuity from 1.4 to 1.6, or approximately 15 percent, for a change from positive to negative contrast. This acuity increase, however, is probably neither important nor real. As suggested by Rupp (1981), the adaptation level is probably not a function of either background luminance or inte- grated luminance, but rather a function of the higher luminance of an irregular surface. Thus, Rupp suggests that the lighter of the two items, either the background or the character, will essentially control the adapting luminance level, thereby negating any effect on pupil size due to positive versus negative contrast. Whether this is actually the case has yet to be demonstrated experimen-

69 tally for VDTs. There is cause for concern over such general- izations because of the lack of direct application of existing literature to VDTs. For example, there is overwhelming evidence that contrast sensitivity, as well as acuity, increases significantly with increases in overall retinal illuminance (see Figure 4.3~; these and other data are based, however, on display fields in which the light and dark elements are approximately equal in area rather than on the unbalanced display typical of a VDT. It is known that people with poorer eyesight benefit more from increased levels of retinal illumination than do people with normal eyesight (Hopkinson and Collins, 1970~. It is also known that maximum acuity is obtained when the surround (the area or sur- face around the display) is equal in luminance to the display (adapting) luminance (Hopkinson and Collins, 1970~. A secondary benefit of higher display luminance is the increase in visual depth of field Cased on a fixed diameter of the "blur circle") as the 1.0 0.5 Zo - ~ 0.05 o - 0.1 ~ 0.01 is o c' 0.005 0.001 Retinal ~ Illuminance — (Trolands) 0.0009~ ~ 0.09 - - // //111 /// / /// 0.9 ~ ~90// ~ 90k l l l 1~1111 l 1 1 1 ,,,1 , , 1 1,,,,1 0.1 0.5 5 10 50 100 SPATIAL FREQUENCY (cycles per degree) FIGURE 4.3 Effect of retinal illuminance on contrast threshold. SOURCE: van Nes and Bouman (1967~.

70 pupil diameter decreases. Assuming that the luminance to which an observer adapts is in fact the space-average luminance of the display, a negative contrast display (higher space-average luminance) would typically yield a pupil diameter of about 4.5 mm while a positive contrast display (lower space~verage luminance) would yield a pupil diameter of about 5.0 mm (see Figure 4.1, above). This difference in pupil diameter corresponds to approxi- mately a 30 percent difference in blur circle diameter (at the 50 percent intensity point), as shown in Figure 4.4. Again, however, application of these data to VDTs in the workplace should be experimentally verified. As with all lenses, aberrations in the eye are greatest in the periphery of the cornea and the lens. Thus, pupil constriction improves the quality of the image formed on the ret me by excluding light that passes through the peripheral portions of the cornea and the lens (i.e., light rays beyond the border of the pupil at its adapted diameter). While pupil constriction is caused by increasing the amount of light in the adapting field, it also occurs synergistically with lens accom- modation (focusing) for near objects. Thus, as the eye focuses on closer objects, such as a VDT at a working distance, the pupil will e.oo _ 7.00 - C' ° 6.00 . _ - ~ 5.00 LL J Cl: Cal J m ,~ 10% Intensity 1~/ } 12% 4.00 3.00 2.00 1.00 I ~ 1 1 1 1 1 1 1 1.00 50% I ntensity ~ 2.00 3.00 4.00 5.00 6.00 7.00 PUPI L DlAM ETER (mm) FIGURE 4.4 Blur circle diameter as a function of eye pupil diameter. SOURCE: Campbell and Gubisch (1966).

71 "automatically" constrict to obtain a somewhat sharper image. Thus, there is a significant interrelationship among display lumi- nance, pupil diameter, blur circle, depth of field, and contrast sensitivity (or acuity). Generally, increases in display luminance will improve visual performance and tend to permit greater cancellation of spherical aberrations by the constricted pupil. On the other hand, positive contrast may tend to make the pupil larger, thereby reducing visual acuity (or contrast sensitivity), increasing the blur circle, and permitting greater spherical aberration. Effects on Flicker Threshold Another physiological effect on the visual system resulting from changes in display luminance relates to shifts in the flicker threshold. As illustrated in Figure 4.5, the temporal contrast sensitivity function becomes less sensitive with decreases in retinal illuminance. Thus, as the average (adapting) luminance of a display increases, the eye is more likely to perceive flicker at any particular repetition rate. This effect has been reported in numerous experiments, including those that have included such variables as the wavelength of the light, the wave form of the stimulus, the size and shape of the stimulus, etc. ~ generali- zation from the research of de Lange (1958), which illustrates the relationship between the critical flicker frequency and the Fourier spectrum of the time varying stimulus, is shown in Figure 4.5. In general, de Lange found that the Fourier fundamental of the display could be used to predict the modulation at which flicker is perceived, as a function of repetition rate, irrespective of the wave form of the light. Unfortunately, large-area displays using negative contrast are perceived to flicker at much higher refresh rates than those using positive contrast in a typical VDT environment. Thus, a display with ~ 50 Hz refresh rate that is just at threshold for flicker at 10 cd/m will flicker very noticeably if luminance is increased to 100 cd/m2. This effect is in conformance with the well~stablished Ferry-Porter Law, which suggests that the highest frequency at which flicker is perceived increases linearly with the logarithm of the adapting luminance, or by approximately 10 Hz for each tenfold increase in luminance. The data of Bauer and Cavonius (1980) clearly support this result. Bauer and Cavonius recommend 1 See also the discussion in Chapter 7 of the relationship between pupil size and accommodation in studies of fatigues

72 Retinal Illuminance (Trolands) 1 0.5 o J o ~ 0.1 o 0.05 0.01 0.005 - Waveforms _ rUL - nurL _ J111J1~ 4.3 43 430 /11 ::j 1 1 1 1 1 1111 1 1 1 1 1 11-11 1 5 10 50 100 FREQUENCY (hertz) FIGURE 4.5 Temporal contrast sensitivity function. SOURCE: de Lange (1958~. Reprinted with permission of the Optical Society of America. a repetition rate of 100 Hz for VDTs with negative contrast. This recommendation appears to be reasonable and probably indicates the main reason that manufacturers have been reluctant to use negative-contrast displays in the past: standard television monitors cannot produce that repetition rate. Effects on Visual Task Performance The effect of display luminance on visual task performance has been investigated in a few studies. Snyder and Taylor (1979) demonstrated that increases in character luminance caused significant increases in individual character legibility in several

73 different viewing tasks. Unfortunately, in that particular experiment, the background luminance level was held constant, and therefore the character luminance was totally confounded with the contrast of the displayed image. Supporting evidence, however, for the effect of display luminance on performance is offered by Bauer and Cavonius (1980), who found that a higher- luminance negative~ontrast display yielded both greater subjective preference and improved visual performance than a lower-luminance negative-contrast display. Further research on the subject of the effect of display luminance when separated from the influence of contrast and contrast polarity is needed, however, before this issue can be directly resolved. Luminance Uniformity There is very little research in the literature to provide informa- tion on the minimum requirements for the uniformity of visual displays. No studies are known to provide either thresholds of detection or tolerance limits for large~rea nonuniformities. In general, we simply do not know how much large~rea nonuniform- ity is a reasonable design goal. The case for small-area nonuniformity is similar. Unless one applies basic sine-wave sensitivity data to a given form of small- area nonuniformity distribution and attempts to predict the detectability of nonuniformity, there is currently not even a suggested means for evaluation. Except for an initial study by Riley and Barbato (1978), there little knowledge of the effects of line errors (on or off) or of element errors (on or off) on display legibility and utility. Research efforts to fill these data gaps are obviously needed. Contrast and Contrast Polarity As suggested in the preceding discussion, increases in contrast have been shown to produce significant increases in visual task performance. In addition to the study of Snyder and Taylor, Shurtleff (1982) also demonstrated increases in legibility as a result of increases in character/background contrast. Further, negative-contrast displays have been found to yield greater legibility than positive~ontrast displays (Bauer and Cavonius, 1980; Radl, 1980~. These studies should, however, be viewed carefully, because changes in polarity were also combined with changes in ambient illumination and absolute contrast magnitude. Again, further research is indicated to achieve a complete under- standing of the relationship between display image contrast and

74 the performance of typical workers. In the experiments to date, all the observers used have been young and have had healthy eyes. Since VDT workers often include older workers and workers having some visually limited capability, it is particularly critical that research be conducted with stratified subject populations that include those people representative of typical VDT workers. No physiological effects are known to be pertinent to the variables of contrast or contrast polarity. Further, the only subjective preference data dealing with these variables has been reported by Radl (1980)and by Bauer and Cavonius (1980), who reported a significant preference for the negative-contrast (black on white) display among the several combinations investigated. Whether this preference would exist under other display and illuminance conditions is unknown. Raster Structure Most VDTs produce characters known as in-raster characters. A CRT creates these characters by drawing horizontal lines (scan lines) on the screen. The electron beam that draws these scan lines is turned on or off as required to produce line segments of symbols and characters on the screen. The collection of scan lines is called a raster, and the characters produced within the raster . are Raster characters. Figure 4.6 shows an example of characters produced in this fashion. Stroke characters are those that are produced by a continuous line process so that they do not appear to be composed of a collection of dots. The printing on this page is an example of stroke characters. Note that the in-raster characters shown in the right portion of Figure 4.6 appear continuous because of the close spacing of the scan lines and proper adjustment of the scan line width. In general, stroke-written characters are preferable to characters having a visible dot or element structure. As the spacing between dots or elements increases, the reading time and reading difficulty increase. As Figure 4.7 indicates, reductions in the space between individual dots reduce reading time, and extrapolation of this function to the zero value on the abscissa suggests an adjusted reading time of zero seconds; that is, zero space between dots (i.e., a stroke-written character) causes no elevation in reading time, which is otherwise the result of space between the dots. It must be recognized that most word processing and data processing displays today use either dot-matrix or raster-written characters, either of which can have visible spacing in the ver- tical dimension and, in the case of dot-matrix characters, also in

- ~ FIGURE 4.6 Characters produced on a VDT screen from a raste' structure.

77 A 525-line raster display will typically present visible spaces between raster lines, which cause character line or dot visibility. A well-designed display having 729 or 1,029 lines is likely to have raster lines that are less visible, and a number of such high- quality displays are currently on the market in word processing and data processing systems. These displays can be expected to make possible performance improvements like those shown in Figures 4.8 and 4.9. Any visible raster structure is undesirable. Studies have shown that a visible raster (modulation in excess of 10 percent) is detri- mental to legibility of either alphanumeric characters or cultural objects. Various means can be used to reduce the raster modula- tion, and one or more should always be used to eliminate visible raster or dot structure. Among the techniques to reduce raster loo is o ~ 80 - o C' UJ CC 'A 60 o it CC UJ 40 - J 20 o _ it' 2 4 /~- L = Large Size Targets ( 13) S = Smal I Size Targets ( 1 3 ) 100% = 1 17 Correct Responses 1 029 729 Number of Scan Ll nes 525 1 1 1 1 1 1 1 1 6 8 10 12 14 16 18 20 22 24 o 14 16 RECOGNITION LATENCY (s) FIGURE 4.8 Recognition latency for 525, 729, and 1,029 line television systems. SOURCE: Humes and Bauerschmidt (1968~.

78 100 O Exp. 3, 875 Lines · Exp. 2, 525 Lines ~ Ref. 2, 525 Lines 80 TV _ c' LL cr o ~ _ ~ _ cot cr llJ Cal 60 _ 40 _ 0 2 ,0 o /~/ A t · I I I I . I 4 6 8 10 12 TV RASTER SCAN LINES/SYMBOL HEIGHT FIGURE 4.9 Effects of display raster lines and number of raster lines intersecting target on symbol recognition. SOURCE: Erickson et al. (1968~. visibility that have been studied to date are spot wobble (Beamon and Snyder, 1980) and matching of the scanning spot size to the raster pitch. It is preferable to select a spot size in the dimension perpendicular to the raster such that the spot size and raster pitch are compatible to produce a visually flat field, that is, a field having no visible raster modulation. In all too many cases, the spot size is too small for the raster pitch, resulting in a visible raster and therefore a static noise source to the visual system. Increasing the number of lines or increasing the spot size in the dimension perpendicular to the raster are appropriate solutions and both result in improved legibility.

79 Resolution The resolution of a VDT display refers, conventionally, to the number of elements (dots) or the number of raster lines with which each character is written. It is often also used to include the visual separation of these elements or the number of elements per unit distance on the display. Thus, the term resolution has a variety of meanings, some of which are not well understood or consistently defined. Assuming, in the more traditional case, that resolution refers to the number of resolvable elements per unit distance on the display, there is no known effect of resolution on the eye as mea- sured by any physiological response. The existing data do, how- ever, clearly support an increase in performance with increases in resolution. For example, Erickson and coworkers (1968) have shown that an increase in the number of raster lines used to write each character causes an increase in legibility. That is, 5 raster lines used to write a given character produce poorer performance than do 7 lines or 10 lines. Based on the results of several experi- ments conducted by Erickson and coworkers, it appears that in order to reach reasonable legibility performance on a television- type display, a minimum of 7 raster lines is desirable (Erickson et al., 1968~. Using a dot-matrix display format, Snyder and Maddox (1978) found that increases in character legibility can be achieved when the number of dots in the matrix writing each character is increased from 5 x 7 to 7 x 9 and from 7 x 9 to 9 x 11. Thus, the highest legibility level is achieved with a 9 x 11 dot matrix, which would be equivalent to Erickson's 11-line raster. Only slightly inferior performance is achieved with 9 raster lines, or equiva- lently, a 7 x 9 dot-matrix character. Anecdotal evidence also supports this contention. There are no known subjective preference effects for various character matrix sizes or numbers of raster lines with which individual characters are written. Jitter and Temporal Instability The simple concept of display jitter has seldom been quanti- tatively addressed. Jitter is the time-based variation of position of displayed symbols due to improper or insufficient video deflection voltages (currents). Some manufacturers specify jitter in terms of motion limits using linear dimension units (inches, millimeters) or percent of display diagonal. Fellmann and coworkers (1981) described a procedure for assessing jitter by focusing a micro- photometer on the middle of the "leg" of a letter and recording

80 the luminance variation as a function of time. They noted that a jitter specification in terms of change in luminance versus mean luminance could be used but chose to offer only qualitative assess- ment based on the recorded graphs. The Electronics Industries Association (1957) does not address the problem of measuring jitter in its published standards (EIA RS-170) pertaining to CRT display measurement. A procedure to measure jitter should be developed and acceptable limits determined. Refresh Rate and Persistence As described above in the section "Luminance," the average luminance of the display can significantly influence the percep- tion of flicker of the display. To combat the tendency to perceive flicker at higher display luminance levels, increases in refresh rate are often necessary. In general, the frequency at which flicker will be perceived will range from approximately 30 Hz to as high as 100 Hz, depending on the temporal modulation of the location of the image in the visual field and the average luminance of the display. Flicker is of particular concern from a physiological stand- point because of its occasionally reported ability to photically induce epileptogenic seizures. Careful analysis and several research studies in this area have clearly demonstrated, however, that persons who are sensitive to epileptogenic seizures caused by flickering displays have induced seizures only when the display refresh rate is extremely low, typically in the 8 to 14 Hz region, the typical alpha frequency region. Since most refreshed VDT displays have a minimum refresh rate of 30 Hz (assuming an interlaced display), there appears to be no significant problem of this nature induced by most existing video displays. Of greater general interest is the lack of any demonstrated relationship between visual task performance and apparent flicker at commonly used refresh rates. Although annoyance, headaches, and other negative subjective responses to flickering displays are often reported, it has yet to be demonstrated that a significant deterioration in visual task performance (except as a secondary effect of negative subjective responses) results from perceived flicker. Thus, it is obvious that displays should be designed to avoid any perceivable flicker; however, this is more a matter of operator comfort than one of demonstrable deterioration of visual task performance. Research recently reported by Grandjean and coworkers (1981) has shown a correlation between the subjectively related quality of displays and the oscillation index, a measure dealing with the temporal variability of display luminance at frequencies within

81 and well above perceived flicker rates. Displays demonstrated to have significant temporal modulation well above 100 Hz have been shown to cause subjectively poorer ratings of visual quality. However, since the oscillation index is computed on the basis of the Fourier fundamental and its first 20 harmonics, over which the Fourier coefficients are summed, it is not clear from this research which harmonics contribute most to the perception of image quality. Furthermore, other characteristics of the displays are confounded with the oscillation index measure because of the very nature of display design parameters, and the evidence is therefore merely indicative rather than conclusive. Color Most VDTs use either a white or a green phosphor, although some displays using a white phosphor contain a color filter to change the apparent chromaticity of the display image. Thus, orange or yellow displays have those hues due to the filter rather than to the intrinsic emission of a phosphor. For the most part, the color of the phosphor, all other characteristics being equal, will have no influence on most physiological measures, such as contrast sen- sitivity or acuity (Watanabe et al., 1968~. In general, for foveally fixated images of the VDT-type, the contrast sensitivity is controlled by the displayed luminance (or adapting luminance) rather than by the chromaticity of that luminance, assuming adequate focusing of the image on the retina. The chromaticity of the image does, however, affect the apparent brightness of the image. Booker (1981) and Costanza (1981) have demonstrated that the red and blue portions of the spectrum are perceived to be brighter at equal luminance than either the green portions or achromatic displayed images. Whether this difference in perceived brightness is related to visual task performance has yet to be demonstrated. There is concern, of course, with displays using the blue or red ends of the spectrum because of the chromatic aberration of the eye: that is, multi- colored displays will cause the red and blue ends of the spectrum to be less accurately focused than green or achromatic displayed images because of the chromatic aberration of the optical portions of the eye. There are no known performance differences associated with differences in the color of the phosphor used. Attempts to measure any performance differences have consistently resulted in no differences, as long as the blue and red extreme ends of the spectrum are not considered. Further, differences between achromatic and green phosphors are totally subjective, and users ' preferences for either are reported approximately equally often.

82 Reflection Characteristics Reflections of ambient lighting by a VDT screen <:an cause a significant loss of character contrast. Two distinctly different types of reflections may occur: Secular (mirrorlike) reflection from the front glass surface of a CRT and diffuse (non-mirrorlike) scattering of incident light from the phosphor surface of a CRT. Specular reflection results in the formation of an image of the source that is causing the reflection; diffuse reflection does not. Specular and diffuse reflections from a VDT screen are illustrated in Figure 4.10. There are many techniques that can be applied to VDT screen construction that can reduce the susceptibility of the screen to reflections of ambient light sources; they are discussed below in the section "Filters for VDTs" (other techniques for reducing reflections are discussed in Chapter 9~. Three different VDT screens were measured to determine the specular and diffuse reflection coefficients (see the section below, "Display Measurement Techniques," for details). The measure- ments are shown in Table 4.1. The specular reflection coefficients were determined by dividing the luminance of the reflected image FIGURE 4.10 Specular and diffuse reflections on a typical VDT screen.

83 TABLE 4.1 VDT Screen Specular and Diffuse Reflection Coefficients i VDT Specular Coefficient Diffuse Coefficient (cd/m2/lux) 2 3 0.072 0.042 0.017 0.091 0.040 0.023 by the luminance of the light source causing the reflection. This unitless number indicates how susceptible the VDT screen is to specular reflections from ambient light sources. The smaller the number the lower the luminance of the specular reflection from any given source. Therefore, smaller values indicate that the contrast loss is not as great. The diffuse reflection is a somewhat different entity. It was determined by dividing the luminance of the VDT screen (with no characters displayed) by the illumination falling on the surface of the screen. The result is a number with units of candelas per meter squared per lux of incident light. Although the units are somewhat awkward, the result indicates how much veiling VDT screen luminance can be expected from light falling on the surface. Again, the smaller the number the better the VDT screen. A secondary effect of specular reflections is that they result in an image of the reflecting source. This virtual image appears to be located behind the VDT screen at a depth that depends on the curvature of the screen and the distance from the screen to the source of the reflection; the significance of this effect is discussed in Chapter 5. The performance, physiological, and subjective effects of reflections are the same as those discussed in the section "Contrast and Contrast Polarity," above. A Summary Measure: Modulation Transfer Function The modulation transfer function (hlTF) is probably the most important parameter associated with image quality.2 Depending 2There have been a number of theoretical attempts to develop measures of image sharpness that can be related to visual perception. The modulation transfer function is one such measure; another is acutance.

84 on the technique used to measure MTF, it directly or indirectly includes the characteristics of contrast, resolution, reflectance coefficients plus ambient lighting, and jitter. The MTF describes the amount of signal output that can be achieved for a specific signal input as a function of spatial frequency. For VDTs it describes the contrast obtainable on the CRT as a function of spatial frequency, which directly relates to the sharpness or crispness of the alphanumeric image. Many techniques have been developed to measure the MTF of displays (Schade, 1948; Snyder, 1974; Bedell, 1975; Task and Verona, 1976~. Each of them makes certain assumptions about the characteristics of the CRT and its associated drive electronics. The most important, and usually the most violated, assumption is that the CRT system is linear. In rigorous mathematical treat- ment, the concept of the hlTF exists only for continuous, linear systems and not for systems in which those conditions are not met. Since the input signal level of the CRT is not linearly related to the output luminance level, CRTs are generally nonlinear devices. When used in a VDT, however, in which only two luminance levels are used (off and on), the CRT operating curve (voltage in versus luminance out) consists of only two points. The CRT used in a VDT can therefore be thought of as a linear device with an operating curve that is a straight line connecting those two points. With this in mind, the concept of the AATF can be readily applied. To achieve the most realistic measurement results, the MTF should be measured directly, without normalizing the function at some arbitrary, low spatial frequency (Snyder, 1974; Task and Verona, 1976~. It should also be noted that the MTF describes the image generation capability only along the direction of TV scan lines. Since the image is discontinuous in the direction perpendicular to the scan lines, an MTF for this direction does not rigorously exist. (The scan line spacing and line profile luminance distribution determine the quality of the display in this direction.) When the MTF is measured by only one of the direct-measure techniques (Schade, 1948; Snyder, 1974; Task and Verona, 1976) under the identical illumination conditions in which a VDT will be used, it includes image degradation due to diffuse and specular reflections from ambient light sources. It is, however, possible to measure the MTF of the VDT in a dark room and accurately predict the hSTF that will result under any ambient condition if the diffuse and specular reflection coefficients of the display are known. If the VDT uses a monochrome enhancement filter, the chromatic distribution of both the ambient light sources and the CRT must be known, as well as the spectral transmittance characteristics of the filter. (Reflections and the effects of various enhancement filters are discussed further in Chapter 5.)

85 The theory and concept of the MTF have been widely accepted; the methods by which it is measured, however, and the ways in which it is applied vary considerably. A concept similar to the MTF, the contrast sensitivity function (CSF), has evolved for vision (DePalma and Lowry, 1962; Campbell and Robson, 1968~. The CSF is often erroneously referred to as the MTF of the eye (Routs and Bouma, 1980~. The CSF is the reciprocal of the contrast threshold function (CTF) of the visual system, which is a measure of the threshold contrast required to resolve a sinusoidal grating as a function of spatial frequency. For determining the relative merit of the MTF of a VDT, the CTF is probably a more appropriate function than the CSF for providing a direct comparison. If the hlTF of a VDT is relatively high (> 0.95) throughout the entire range of spatial frequencies to which the eye is sensitive (0 to 60 cycles/degree), the display should be essentially indistinguishable from the ideal alpha- numeric character. In fact, this area between the CTF of the visual system and the MTF of the display is the unified image quality parameter that is designated as the modulation transfer function area (MTFA) (Borough et al., 1967; Snyder, 1974~. An MTF greater than 0.95 is typically not technologically feasible in practical VDT applications. In general, lower MTFs result in a lower image quality, but this reduction in perceived quality is not linear with spatial frequency (Carlson and Cohen, 1978~. The relationship between the MTF and ocular discomfort sometimes associated with long-term viewing of video displays has not been explored. Since the effect of a poor MTF is to produce a blurred or low contrast image, or both, on the retina, it is reasonable to assume that if the image is sufficiently blurred, the eye will try to adjust its accommodation in an attempt to minimize the blur. This might lead to accommodative fatigue and visual complaints; however, no evidence for this has been presented (see Chapter 7~. How poor an MTF would have to be to produce such an effect is unknown. DISPLAY MEASUREMENT: TECHNIQUES AND PROBLEMS Photometric measurements are fundamental to measuring VDT display characteristics (for a detailed treatment of photometry and photometric measurements, see Teele, 1965; Smith, 1966; Klein, 1970~. The basic unit of photometry is the lumen, which is a measure of visible optical power. To obtain lumen measure- ments, a photometer is fitted with a filter so that it has the same wavelength sensitivity as the human eye. For the peak of visual sensitivity (at a wavelength of 555 nary), one watt equals 685 lumens. The conversion factor becomes smaller for both longer

86 wavelengths (toward the red) and shorter wavelengths (toward the blue) in accordance with the so-called photopic sensitivity curve. There are two fundamental photometric parameters for VDTs: luminance and illuminance. All display measurements (with the exception of color) discussed in this section are based on those two parameterse Luminance is the photometric parallel to the sensation of brightness for extended sources; it is a measure of the lumens emitted by a source per solid angle per unit area. The usual CIE (Commission Interr~tionale de l'Ec~airage} unit for luminance is the nit (1 candela/m ~ or candela/m (cd/m~), which this report uses for luminance values (the English unit most commonly used is the foot-Lambert). Illuminance is a measure of the lumens incident on a surface per unit area. There is no directly parallel visual sensation. The combination of illumination and reflection characteristics of non- light-emitting objects determines the luminance of those objects. The standard CIE unit for illuminance is the lux (1 lumen/m,), which is the illuminance unit used in this report (the most com- monly used English unit is the foot-candle). Measurement Techniques Many photometric measuring instruments are commercially available and range from $300 or $400 to more than $10,000. All photometers used to measure luminance have a similar basic construction, consisting of an objective lens, an aperture, a filter, and some type of light-sensing component such as a silicon diode or photomultiplier tube (PMT). The objective lens forms an image of the object being measured at the aperture plane. The aperture is usually a circular hole or a rectangular slit. The light-sensitive element senses only the light that falls on the aperture. The geometry of the instrument is arranged so that the output reading corresponds to the average luminance of the part of the object that is imaged on the aperture. Figure 4.1 1 shows the arrange- ment of these components. Figure 4.12 shows how such a photometer might be mounted and positioned to make measurements on a VDT screen. (It should be noted that normally a photometer such as that shown in Figure 4.12 is located much farther from a VDT when measuring with the lens shown and much closer when measuring with the microscope objective lens.) With the aid of a barium sulfate reflection, the type of photometer shown in Figure 4.12 can also be used to measure illuminance at the VDT screen. Barium sulfate reflects light in a highly diffuse manner and very nearly approximates a perfect diffusing surface. Hence, by positioning the barium sulfate surface at the position at which the illuminance is to be

87 Portion of Character Measured Character Being Measured Circular / Aperture / Objective perture liter Lens Plane FIGURE 4.1 1 Components of a typical photometer used for · — measuring . .umlnance. Sensitive Component measured (as shown in Figure 4.13 for the VDT screen), the luminance of the barium sulfate can be measured and converted to the illuminance at the plane of the material. The illuminance (in lux) con be determined by multiplying the measured luminance in cd/m by the value of A. We note again that illuminance and luminance are two different types of parameters and one in general cannot be calculated from the other unless the physical and geometric __ _. FIGURE 4.12 Photometer mounted to measure VDT screen.

88 - .-- .... ~~ .~ ~~ -.-.-. ~~ .. ~~ ~~.~ ~ _ .. ... ~~: < , :~ (~ ~: <~,,~ -~ .. ~ . . . . FIGURE 4.13 Measurement of illuminance at the VDT screen. conditions of the situation are known. In English units, one foot-candle of illumination reflecting from a perfectly white, diffusing reflector gives rise to one foot-Lambert of luminance, and this fact has caused considerable confusion of these two very different parameters. Measurement of Various Parameters Character Luminance The very term character luminance implies that there is a single luminance that is associated with a VDT character. As can be seen in Figure 4.14 the implication is not correct. A character is typically made up of a series of dots and strokes, each of which has a distribution of luminance values across its extent. In gen- eral, the linear horizontal strokes (e.g., the top of the letter T and the top, middle, and bottom arms of the letter E) are higher in luminance than the short dots that make up the vertical elements of the characters. While a photometer with microscope objective lens can easily measure any aspect of this luminance distribution, the question is what should be measured and labeled as character

g9 FIGURE 4.14 Light distribution that forms VDT characters. luminance. There are several options but at present there is no accepted standard procedure. The three main options include: (1 ) a narrow vertical slit (width of the slit much less than the width of the vertical stroke of character) scan across a vertical character stroke (such as an I) with the vertical dimension of the slit covering several dots; (2) the same as above, except that the vertical height is limited to only one dot or a fraction of one dot; or (3) a small circular aperture with a diameter much smaller than the size of a dot. Each measurement will yield a luminance value, but the most appropriate measurement for determining character luminance is not apparent. Character Contrast To calculate the contrast of a character, it is first necessary to determine the character luminance and the background lumi- nance. The previous discussion on what constitutes character luminance obviously applies to the uncertainty associated with determining contrast. The background luminance is also a problem because it differs depending on the ambient lighting conditions and the location on the screen at which the background reading is made relative to the location of the character or characters. A

9o TABLE 4.2 Modulation Contrast, Differential Contrast, and Contrast Ratio for Two Measures of Background Luminance: Hypothetical Case Background Measurement Luminance Modulation Differential Contrast Location (cd/m2) Contrast(%) Contrast Ratio Far from character 0.5 99 199 200 Near to character 2.0 96 49 50 NOTE: The assumed character luminance is 100 cd/m2. CRT screen typically carries a halo of light around characters due to several light-scattering and stray light sources, which results in a lower value of measured background luminance the farther from the character the measurement is made. Dark-room measure- ments of absolute contrast (calculated by the difference divided by the sum technique) will not be significantly affected by small absolute changes in background luminance due to the location of the measurement. Other values, however, such as differential contrast (difference divided by background) or the luminance ratio (also called contrast ratio, or character luminance divided by background), can change greatly with small changes in absolute background luminance readings. To illustrate this effect, Table 4.2 shows a hypothetical set of data for character luminance and two different background luminances (one measured near the character and one measured farther away). Note that the modulation contrast is relatively insensitive to differences in background luminance compared with either of the other two calculations for contrast. It is apparent from the data in Table 4.2 that if a definition of contrast is used that is~ighly sensitive to background luminance measurement, the specific conditions under which background luminance is measured need to be specified and, preferably, standardized. Blur, Resolution, and MTF The terms blur, resolution, and MTF all relate to the definition or appearance of sharpness of the VDT characterse No accepted standard procedure exists for measuring these parameterse What- ever procedure is used to measure them, it is still necessary to select and preferably standardize an appropriate photometer aperture.

91 Blur is usually based on a determination of how rapidly the luminance changes with distance at the edge of a character (Dainty and Shaw, 1974; Grandjean and Vigliani, 1980~. This is not always easy to determine. Figure 4.15 shows the luminance dis- tribution (measured with a vertical slit and a scanning micro- photometer) across a single dot in the character I. It is difficult to define where the edge of either side of the I begins and ends in such a way that repeatable measures can be achieved. The MTF can be measured in many ways (Schade, 1948; Snyder, 1974; Task and Verona, 1976; Gaskill, 1978 Probably the easiest method for VDT screens is by using the edge-response or line- response techniques described in Gaskill (1978~. These methods are preferred because no special, separate signal generation devices are required; instead, selected VDT characters can serve as the input signals. For example, if the input signal is the narrowest vertical line that the VDT is capable of producing, then the luminance profile shown in Figure 4.15 is the line spread function (LSF) for the VDT. The &1TF of the VDT can then be calculated by taking the normalized Fourier transform of the LSF, as described in Gaskill (1978~. An alternative technique is to measure the edge response of some selected character on a VDT. This technique does not c' a A J Photometer Scan Path VDT Character " I " \ X FIGURE 4.15 Luminance distribution across a single dot in the character I.

92 ~~ - X FIGURE 4.16 Typical edge response (luminance distribution at the edge of a character). require the assumption that the selected character represents the narrowest line possible, as does the LSF technique. It does, however, require an additional mathematical step. The edge response is simply the luminance distribution measured at the edge of a character. A typical edge response may look like that shown in Figure 4.16. The curve is first differentiated, which yields the line spread function, and then the normalized Fourier transform is taken to obtain the MTF (as described above). Reflection Characteristics As described in the section "Filters for VDTs," there are two types of reflections that occur with VDTs: specular and diffuse. Specular reflection occurs from the front surface of the screen and satisfies the optical law that the angle of incidence of a light ray on a surface equals the angle of reflection. Diffuse reflection is a scattering in all directions of the incident light and occurs at the phosphor surface of the VDT. There are no accepted, stan- dardized procedures to measure either of these parameters. The specular reflection coefficients for the data presented in Tables 4.1, 4.5, and 4.6 were obtained using the following pro- cedure. A light box, measuring approximately 20 x 25 cm, was positioned 1.5 m from a VDT screen and tilted at an angle of about 17° from a straight line from the center of the VDT screen (see Figure 4.17~. The photometer was located about 17° on the other side of the straight-line position and focused on the VDT screen. The luminance of the reflection was measured and divided by the luminance of the light box. It should be noted that much of

93 the light from the light box was transmitted through the front surface of the VDT screen and fell on the diffuse phosphor surface. The measurement was not, therefore, a pure specular reflectance coefficient because of the diffusely scattered light in the area measured. It is extremely difficult to avoid this situa- tion entirely. Since the effect depends on the size of the light box used and its distance from the VDT, it is necessary to specify these parameters in order to make meaningful comparisons between VDTs. Determining the diffuse reflection coefficient is somewhat more involved. The phosphor surface acts as a fairly good dif- fusing surface, but it is by no means a perfect diffuser. Conse- quently, the amount of light diffusely reflected depends both on the angle of the photometer with respect to the VDT surface and the angle of the light source with respect to the VDT surface. A geometry should be chosen that is representative of the viewing and illumination conditions typically encountered in VDT use. For the measurements shown in Tables 4.5 and 4.6, the photometer was located directly in front of the VDT screen, simulating an operator's view. The illuminating light source consisted of overhead room lights and a slide projector located about 45° from the straight-line position (see Figure 4.18~. The resulting illumination at the VDT screen was approximately 400 lux, near the mid-range of the 300-500 lux recommended for office work areas. The illumination was measured using the barium sulfate technique described in the section "Measurement Techniques." The diffuse reflectance coefficient was calculated by dividing the resulting screen luminance (using a blank screen) by the incident illumination. VDT Screen - 7° 17° 1,8 ~ _: Photometer Light Box FIGURE 4.17 Geometry for measuring specular reflection.

94 - VDT Screen A \ /~45 bide Projector -lit Ph otometer FIGURE 4.18 Geometry used for measuring diffuse reflectance coefficient. Since individual characters are not measured in determining the reflection characteristics, the choice of photometer aperture is not critical. Standardization As is apparent from the preceding discussion, there are currently no standard procedures for measuring the critical display quality characteristics of VDTs. Standard procedures should be developed so that accurate comparisons of image quality among VDTs can be made. Once this is accomplished, recommended values for these parameters should be established to provide guidelines for deter- mining the image quality of any particular VDT. FLAT-PANEL DISPLAYS Most of the concern about VDTs centers on the cathode-ray tube, currently the most common display device in VDTs. However, a growing proportion of terminals replace the CRT with a flat- panel, solid-state display device. While analysis of the engineer- ing details and advantages of flat-panel displays is beyond the purposes of this report, it should be recognized that these devices may have distinct advantages over the CRT. Flat-panel displays take up less display depth and can therefore be located more

95 conveniently on a display surface of limited size. In addition, since each picture element (pixel) on a flat-panel display is defined by the location of electrodes or similar elements, the picture is geometrically stable and does not move from frame to frame or over time. The following section describes the main variables related to visual perception of the dot-matrix display structure used in flat-panel displays and indicates what is known about the relationship between these variables and visual task performance. Since the basic physiological effects of the various display variables (luminance, contrast, flicker, etc.) are the same for flat-panel displays as for CRTs, this section is limited to a discussion of the relationships between display design variables and performance. Dot-Matrix Display Variables Character Size Effects In general, the size of the display should be adequate to present the necessary information for the task required of the operator. There are no standard requirements for screen size, but there are requirements for sizes of the information elements presented on the screen. Thus, as the task requires a greater amount of infor- mation, the screen generally increases in size to accommodate the larger amount of information, each element of which must be presented at a suitable size. It has been shown that the proper character size varies with the nature of the task. For example, visual search for individual characters is improved as the size of the character gets larger. Snyder and Maddox (1978) showed that increases in individual dot sizes up to 1.50 mm improved random search time (Figure 4.19~. They also showed that a dot size of 1.50 mm is greater than opti- mum for continuous reading of text (Figure 4.20~. In essence, if the characters are too large, reading time is reduced because of the increase in necessary eye movements and numbers of visual fixations. On the other hand, larger characters are desirable in a search-type display, simply because peripheral characters can apparently be more easily located in peripheral vision. A mean- ingful standard at the present time appears to be that offered by the proposed German TCA specification, which requires a mini- mum character size of 2.6 mm or 18 min of arc, whichever is greater. There is a large variability among observers and among tasks in viewing distance, and the character size must be compatible with the viewing distance. The above specification takes this into account. It generally assumes that the minimum viewing distance

96 20 - I Cal 16 UJ U) L' 12 CC o 111 8 UJ Z 4 LL oL I I I I I 0.50 0.75 1.00 1.25 1.50 1.75 ELEMENT Sl ZE (mm) FIGURE 4.19 Effect of element size on random search time SOURCE: Snyder and Maddox (1978~. 4 - UJ ~ Z lo_ In Cal a. m Q z _ ~ On Z ~ 1 Lo. Z . O _ 0.50 0.75 ~ _ - - 1.00 1.25 1.50 1.75 ELEMENT SIZE (mm) FIGURE 4.20 Effect of element size on reading time. SOURCE: Snyder and Maddox (1978).

97 will be 50 cm, although longer viewing distances are feasible for some particular tasks and workplace geometries. If a longer viewing distance can be anticipated, then the greater character size necessary to meet the 18 minutes of arc requirement is justified. Character sizes should not be obtained at the cost of a reduc- tion in image quality (see below). For example, it can be demon- strated that increasing the character size by increasing the pitch of the raster lines in a CRT display is detrimental because the visibility of raster lines significantly reduces performance even though the character size is increased. Character Formation In general, as for CRT displays, stroke-written characters are preferred to characters having visible dot or element structure (see discussion in the section "Raster Structured. Where dot-matrix characters are used, there are known trade-offs among the design characteristics of the characters. In particular, the dot size and shape interact in a unique fashion. In all cases, vertical elongation of dots should be avoided; dots approaching a square aspect ratio are most desirable. This has been shown in several experiments, such as that summarized in Figure 4.21. When the shape of the dot is combined with what is known regarding spacing between elements (see discussion in the section "Raster Structure"), it is apparent that the best display design has square dots that are essentially adjacent to one another. The size of the dot matrix or the number of video lines in a raster display used to form the individual characters is equally important. Figure 4.22 shows that the optimum matrix size depends on the character font used. As discussed above, a 7 x 9 character is generally more legible than a 5 x 7 character and a 9 x 11 character is more legible than a 7 x 9 one. Also, the more dots or elements available to form the character the better the individual legibility of the character, although diminishing returns appear to be reached beyond matrix sizes of 9 x 11. Similar results have been obtained with raster scan displays. In recent years, there has been some interest in describing the intermittence of characters using a measure called percent active area, which is simply the proportion of the total display space which is illuminated by the dots. The percent active area is the square of the ratio of the dot diameter to the center-to-center dot spacing, multiplied by 100. As shown in Figure 4.23, increases in the percent active area lead to reductions in character recogni- tion error rates. Under adverse reading conditions, active areas in excess of 50 percent are desirable, while under normal conditions

98 4 Ue! ZIn U) — LL en 2 A: u. m ~ Z LL _ ~ 1—c In 45 it 1 ~ _ o _ ~ Square Horizontally El ongated Vertically Elongated ELEMENT SHAPE FIGURE 4.21 Effect of element shape on reading time. SOURCE: Snyder and Maddox (1978~. active areas in excess of 30 percent appear to achieve asymptotic performance. In general, the percent active area measure is not critical if good design practice is followed by reducing space between character dots. Contrast Maximum legibility can be achieved for contextual (e.g., text) displays when the modulation is at least 75 percent.3 Very little gain is achieved beyond 75 percent. If the information presented on the display is noncontextual (e.g., isolated numerals or letters), Modulation, one of several quantitative measures of contrast, is equal to the difference between character and background luminance divided by the sum of the two.

99 70 60 50 In O 40 fir LL ~ 30 UJ 20 10 o Huddleston O Lincol n/M itre ~3 Maximum Dot Maximum Angle 1~ 5x7 7x9 2 3 9 x 11 7 x 9(=5x7) 9 x 11 (=5x7) SIZE FIGURE 4.22 Interaction of font and character/matrix size in determining single character legibility. SOURCE: Snyder and Maddox (1978~. a modulation of at least 90 percent is required to avoid reductions in legibility. It should be noted that these values of modulation are referenced to the display as viewed by the operator and therefore take into account the ambient illuminance and any reflections. Glare reduction, good workplace illumination design, and high intrinsic contrast of the display are all necessary to achieve an acceptable contrast level. Characters with sharper edges, or less blur, are generally more legible than those with greater blur or reduced sharpness. Unfor- tunately, adequate measures of sharpness and their relationship to legibility are not well established. It has been shown that reduc- tions in object identification occur when the blur exceeds one-half of the width of the individual item, but these data apply to vari- ous cultural objects rather than to alphanumeric characters. Studies should be performed in which text reading and legibility, rather than object recognition, are the primary tasks.

100 Decreased Unstressed Luminance Condition A Condition B Decreased Lumi nance & Increased Reading Dist Condition C Background 1.2 0.12 0.12 0.97 (fL) Luminance Reading Dist 18 t8 24 24 (inches) Contrast 7.5 7.5 7.5 3.2 20 UJ Cry CC UJ Z 10 CC o cr: 111 5 o 10 Decreased Co ntrast & Increased Reading Dist Condition D \ ,Condition D ,,L, If \~/ \ Condition C ~ Condition B ,< \ ·~ O ~ 20 30 40 50 60 70 80 MEASURED PERCENTAGE OF ACTIVE AREA FIGURE 4.23 Effect of percent active area on character recognition. SOURCE: Stein (1980). Font The legibility of displayed alphanumeric information is greatly dependent on the character style or font. Legibility also interacts with the size of the matrix and the overall character size. As illustrated in Figure 4.22, the Huddleston font is the most legible of those studied for 5 x 7 characters, but the Huddleston and

101 Lincoln/Mitre fonts are equally legible for either 7 x 9 or 9 x 11 matrix sizes. Since there is absolutely no standardization of fonts across existing systems, care should be taken by designers and users to select fonts that give optimum legibility rather than unique character designs. Generalizations from existing literature pertaining to stroke-written characters (e.g., printed text) appear reasonable and should be followed until more directly related data are generated. Luminance Uniformity Uniformity considerations are similar to those discussed previously in this chapter for CRT displays. Information Density Research relating the minimum, maximum, and optimum densities of information in the vertical and horizontal dimensions is urgently needed. Currently, word processing and data processing displays range from a few lines through a more typical 24 lines per display height to a full page of approximately 60 lines. The displays vary in physical size, and the characters also vary in size. It is clear that full-page displays are desirable for formatting purposes, but they are often very difficult to read because of the resulting small character size. Similarly, it is obvious that large character sizes on partial page displays produce legible characters but that formatting is a difficult and often tiring task. There are no useful guidelines from the literature to suggest optimum levels of display information density, and we strongly recommend research in this area. Dot-Matrix Display Quality Measures While image quality measures have been researched in some depth for CRT displays, very little attention has been given to suitable measures of image quality for flat-panel displays. Although it may at first seem reasonable to assume that such measures should be approximately the same, the very nature of the differences between the two displays suggests that the metrics designed to accommodate continuous information, as is the case with the CRT, cannot often be used to describe information that is presented discretely. This section summarizes briefly the only research done to date that has attempted to summarize image quality for dot-matrix displays.

102 TABLE 4.3 Pool of Predictor Variables Vertical Horizontal Description VFREQ HFREQ Fundamental spatial frequency (cyc/deg) VFLOG HFLOG Base 10 log of fundamental spatial frequency VSQR HSQR Square of (fundamental spatial frequency minus 14.0) VMOD HMOD Modulation of fundamental spatial frequency VDIV HDIV Fundamental spatial frequency divided by modulation VLOG HLOG Base 10 log of VDIV and HDIV VMTFA HMTFA Pseudo-modulation-transfer-function area VMLOG HMLOG Base 10 log of VMTFA and HMTFA MCROS HCROS Spatial frequency at which modulation curve crosses the threshold curve VRANG HRANG Crossover frequency minus fundamental frequency SOURCE: SnyderandMaddox(1978). In a three-year research program, Snyder and Maddox (1978) summarized the best possible prediction of image quality and visual task performance from a variety of geometric and photometric variables that were measured from flat-panel displays. The pool of predictor variables is shown in Table 4.3. These variables were all measured physically from a variety of flat-panel displays from which human visual task performance data were collected. The data pertained to two visual tasks, a reading task and a visual search task for randomly appearing alphanumerics. The predictor variables shown in Table 4.3 were then entered into a linear stepwise multiple regression equation, to obtain the best prediction equation for both the reading and the visual search tasks. The resulting prediction equations are shown in Table 4.4. From this table it can be seen that the prediction equation predicts reading time to an accuracy of approximately 53 percent of the total variance among display types, and the equation for search time predicts approximately 50 percent of the variability among different displays. It would appear that these predictability proportions can be improved with further research, but it is also clear that it is necessary to make careful and detailed measurements of displays to achieve this level of predictability. Further research is clearly indicated to obtain a greater understanding of the relationship between visual task performance and the design of flat-panel displays.

103 TABLE 4.4 Extended Predictive Equations Task Metric and Related Information Reading Time Search Time Adjusted Reading Time (s) = 5.74 + 0.3111 (HFREQ) + 2.479(HMOD) + 4.365(HLOG) - 14.973(HFLOG) + 1.112(VMLOG) Correlation Coefficient R = .72 R2 = .525 Asymptotic R2 = .637 Search Time (s) = 7.27 + 0.027(HDIV) + 2.159(HLOG) + 5.916(VFLOG) - 0.339(VMTFA) - 0.054(VRANG) + 5.487(VMLOG) Correlation Coefficient R = .71 R2 = .500 Asymptotic R2 = .575 SOURCE: Snyder and Maddox (1978). Advantages and Disadvantages of Flat- Panel Displays Compared With CRTs A flat-panel display is usually only 1 to 2 in. deep, while the CRT used in most terminals is on the order of 12 to 18 in. deep. Thus, for a given desk size, a flat-panel display can be located farther from an operator than a CRT display and may therefore be helpful in preventing problems with accommodation. The flat-panel display is also usually lighter weight and can therefore be moved more readily. A flat-panel display has a fixed image location, which does not vary with voltage irregularities, and it does not have deflection circuit inadequacies and some of the other ills that plague CRT displays. It has been suggested by some that the better image stability of flat displays may help significantly reduce ocular discomfort reported by users of CRT VDTs; however, there has been no research directly addressing this suggestion. Greater contrast can be obtained on some flat-panel displays in compari- son with CRTs. This is often desirable in an environment that has high ambient illumination. The major disadvantage of a flat-panel display is its extremely high cost relative to a CRT. At the present time, the few flat- panel displays that would meet the requirements of current data processing and word processing terminals cost in excess of $3,000 prohibitive compared with the cost of typical CRT displays. Thus, it may be some time before widespread use of flat-panel displays is seen in the VDT environment.

104 FILTERS FOR VDTs The contrast-reducing effects of reflections can be partially controlled by the use of various optical and physical techniques. If these techniques are not used and if the ambient lighting conditions cannot be properly controlled, it may be advisable to use a filter over the screen. Many types of filters are available, ranging from less than 35 to more than $100 and having an equal range of effectiveness. The purpose of these filters is to improve the legibility of the display by improving the contrast or reducing glare: in most cases "glare" refers to specular reflections from the front surface of the VDT. Both diffuse and specular reflec- tions from VDT screens were discussed in the section "Reflection Characteristics." This section describes several types of filters that are currently available and discusses their effectiveness. Kinds of Filters Circular Polarizer with Antireflection Coating A circular polarizer filter with antireflection coating can be used to reduce both specular and diffuse reflections. It is the most expensive filter available and probably one of the most effective. The outside surface of this type of filter is coated with several layers of optically transparent materials to form what is called an antireflection coating. The effect of the coating is to signifi- cantly reduce specular reflections from the surface of the filter. The rest of the filter package consists of substrate material (typically glass) sandwiched around the more delicate components, a linear polarizer and a quarter-wave plate. The linear polarizer and the quarter-wave plate together form what is commonly known as a circular polarizer. The circular polarizer converts unpolarized incident light to circularly polarized light. The light is changed from right-handed circularly polarized light to left- handed circularly polarized light (or vice versa) on reflection from the VDT screen. Because of the optical physics of the circular polarizer, the light is blocked from getting back through the filter in much the same way that light is blocked by crossed linear polarizers. This type of filter reduces specular reflections in two ways: by reducing specular reflections from the filter itself through the use of the antireflection coating and by eliminating specular reflections from the underlying VDT screen through use of the circular polarizer. Diffuse reflections are reduced primarily by the light attenuation effects of the polarizer material, which allows only about 35 percent of the incident unpolarized light to

105 pass through the filter to the phosphor surface of the VDT screen. The light is diffusely scattered by the phosphor surface, thus losing most of its polarization characteristics; and it is again reduced to about 35 percent as it passes back through the filter toward the user. This process results in an improvement of the display contrast since the ambient incident light (illumination) is attenuated twice by the filter (once as it arrives at the screen and again as it diffusely reflects through the filter toward the operator), while the VDT character luminance is attenuated only once as it passes through the filter to the operator. Neutral Density Filters A neutral density filter is probably the simplest of the contrast enhancement filters. It typically consists of a neutrally tinted plastic that allows the passage of some percentage (usually 15-25 percent) of the light that falls on it. This filter is most effective in reducing diffuse reflections. Light from ambient sources is attenuated twice as it passes through the filter to the VDT phosphor surface and is reflected from the phosphor surface through the filter toward the operator. Since the light from the MDT characters passes through the filter only once, the display contrast is improved. Specular reflections may not be reduced by this type of filter unless the surface of the filter is treated with an antireflection coating (as discussed above) or with a matte finish coating that blurs the specular reflections. A filter that is apparently not commonly available but that would appear to be both effective and inexpensive is a neutral density filter formed into a spherical concave shape. Because such a shape is opposite in direction to the curvature of the VDT screen, the edges of the filter would have to be located a short distance from the screen. If the radius of curvature of the screen were approximately equal to the operator's viewing distance, alla the screen were tilted somewhat below the operator's eye level, reflection sources would be limited to the operator's chest and abdominal areas. And if those areas were kept somewhat dark, for example by an operator's wearing dark clothing, specular reflec- tions should not be a problem. Diffuse reflections would be reduced as they are with any neutral density filter. A filter based on the physical curvature of the filter material is described in U.S. Patent 3,744,893 entitled "Viewing Device with Filter Means for Optimizing Image Quality" issued to Chandler (1973~. As described, the filter was intended for use with a film viewing device but could be adapted to VDTs.

106 Notch or Color Filters Notch or color filters are designed to allow transmission of a high percentage of incident light of some specified wavelengths (typically in the green portion of the spectrum) and a high absorption at other wavelengths. The principle of this type of filter is essentially the same as that of a neutral density filter, but in notch or color filters the bandpass (color) is tuned to the VDT screen color. A green filter placed over a VDT with a green phosphor will allow most of the display luminance to pass through the filter to the operator, while ambient illumination, which is usually broadband white, is largely absorbed by the filter (except for the green portion). This process reduces the ambient light that causes diffuse reflections on the VDT screen, thus improving contrast. As is the case with neutral density filters, control of specular reflections with this type of filter depends on the surface treatment of the filter. Directional Filters Directional filters use geometric or optical means to prevent ambient light from reaching the VDT or to prevent reflections from reaching the user. One type of directional filter is com- posed of a thin sheet of material with tiny, opaque, imbedded slats that are perpendicular to the surface of the sheet. The slats act as a miniature Venetian blind, allowing light to travel only in certain directions. When the slats are oriented toward the oper- ator, light from the VDT can pass to the operator but light from overhead cannot reach the VDT screen. This process reduces contrast loss due to diffuse reflections. Specular reflections would have to be reduced by surface treatment of the filter, as described above. Evaluation of Filters General Comments Some general characteristics of filters should be noted. First, antireflection coatings tend to be somewhat delicate and will typically degrade with time, use, and cleaning. Second, plastics used in filters are softer than glass, and they also become scratched and degrade with time, thus reducing the effectiveness of the filter.

107 TABLE 4.5 Effect of Several Filters on Contrast and Luminance of VDT Characters for a Smooth-Finish VDT Screen Dark Room Specular Reflectiona Diffuse Reflections Filter Contrast Luminance Contrast Luminance Contrast Luminance (% ~ (cd/ ~) (% ) (Cdl m2) (% ~ (cd/ m2) None 98.8 115.0 26.1 330.0 66.4 178.0 1 97.6 28.7 17.4 115.0 65.9 45.8 2 98.1 36.3 24.5 116.0 67.1 56.4 3 96.9 21.9 16.6 92.7 69.2 32.8 4 98.3 41.3 21.1 139.0 70.9 58.1 5 98.2 36.9 28.9 99.2 81.6 50.6 6 99.2 41.7 34.9 95.8 81.3 58.1 7 99.0 34.2 11.8 195.0 80.6 47.9 a Illumination at screen, 266 lux; luminance of specular reflection source, 2,950 cd/ m2 b Illumination at screen, 413 lux; luminance of specular reflection source, none. Third, matte-surface treatments are not very effective in dealing with specular reflections in terms of their effect on contrast, although they do reduce the sharpness of specular reflections. Unfortunately, matte finishes reduce the sharpness of the display characters as well, and this effect increases the farther from the VDT surface the filter is located. Some loss in character sharpness may be helpful, however, in reducing the dot structure of characters (see data on filters 1, 2, 3, and 4 in Tables 4.5 and 4.6~. Fourth, VDT screens are convex, curved surfaces and are therefore susceptible to specular reflections that are visible to the operator over a very wide range of angles (see Figure 4.24a). If a flat or concave filter is placed over the screen, the angles over which specular reflections may occur are drastically reduced and therefore more easily controlled (see Figure 4.24b), a subtle but signficant advantage for such filters. Effectiveness of Filters Because the effectiveness of a particular filter depends on many variables and combinations of variables, it is not possible to fully discuss the issue of effectiveness in this report. For a limited comparison of the effectiveness of several filters and filter types, we measured the effects of seven filters on two different types of CRT screens:

108 TABLE 4.6 Effect of Several Filters on Contrast and Luminance of DOT Characters for a Matte-Finish VDT Screen Specular Reflectiona Diffuse Reflections Dark Room Plus Room Lights Plus Room Lights Filter Contrast Lurn~nance Contrast Lurn~nance Contrast Luminance (% ) (cd/ m2) (% ) (cd/ m2) (% ~ (cd/ m2) None 99.5 130.0 35.2 248.0 78.9 145.0 1 98.0 33.5 17.5 110.0 71.0 38.3 2 98.4 43.1 24.0 107.0 77.6 48.7 3 97.4 25.6 16.1 94.7 78.7 28.7 4 98.5 46.2 23.4 128.0 81.7 51.0 5 98.4 43.1 30.7 90.3 85.5 43.8 6 98.6 46.9 36.5 88.9 87.0 49.2 7 98.2 37.3 11.3 185.0 83.1 40.7 ~- a Illum~nation at screen, 293 lux; lurn~nance of specular reflection source, 2,950 cd/ m2 b illurn~nation at screen, 428 lux; luminance of specular reflection source, none. 1. Amber filter with matte finish (curved) 2. Gray filter with matte finish (curved) 3. Green filter with matte finish (curved) 4. Neutral filter with matte finish (curved) 5. Circular polarizer with antireflection coating (flat)-- manufacturer A 6. Circular polarizer with antireflection coating (flat,~- manufacturer B 7. Green filter with smooth finish (flat) The filters were measured under three conditions: in total darkness, in the presence of a specular source, and in the presence of a diffuse reflection source. The specular reflection source was a light box with a luminance of approximately 2,950 cd/m2 positioned approximately 1.5 m from the VDT screen and approxi- mately 17° off axis (see Figure 4.25~. Measurements under the specular reflection condition were taken with the room lights on, thus this condition was not one of a pure specular reflection. The diffuse reflection condition was achieved with a combination of normal room lights and a slide projector located off to one side to provide nonspecularly reflecting illumination (specular and diffuse reflections on a typical VDT screen are shown in Figure 4.10~. The illumination at the plane of the screen (which results in loss of contrast due to diffuse reflection) was measured under each of the three conditions. The measurements are shown in Tables 4.5 and 4~6.

109 A B VDT ~- Screen < ~ "I Ad_ //: VDT I' / Screen -I N Flat Filter - - - - - ~ Observer Observer = Angle for which specular reflections occur for curved VDT screen ¢= Angle for which specular reflections occur for flat filter NOTE: ~ < ~ FIGURE 4.24 Specular reflection angles for a curved VDT screen (a) and for a flat VDT filter (b). The contrast and the luminance of the VDT characters were measured without a filter on two different types of VDT screens. Table 4.5 shows the measurements made on a VDT screen with a smooth surface; Table 4.6 shows the measurements on a screen with a matte surface. There are several items worthy of special note in the data of Tables 4.5 and 4.6. Displays with both smooth and matte finishes have extremely high contrast in a dark room; it is the ambient

110 VDT Screen_ l / Filter 't 17° ~ 17° - - Photometer - `~ ~ Light Box al FIGURE 4.25 Top view of geometry used to test filter effectiveness against specular reflections. environmental lighting that causes a loss of contrast. All filters reduce the luminance of the display characters. This means that when a filter is used, the VAT must be operated at a higher beam current to achieve the same character luminance as when no filter is used. The increased beam current causes the phosphor to age (become less efficient) more rapidly and reduces the lifetime of a CRT. For the smooth-finish screen (Table 4.5), the circular polarizer filters improved contrast under the specular reflection condition; the improvement, however, was only moderate. For the matte- finish screen (Table 4.6), under the specular reflection condition, none of the filters resulted in a significant improvement over the no-filter condition. For both the smooth-finish and matte-finish screens, several filters improved the contrast under the diffuse reflection condition compared with the no-filter condition, but again the improvement was only moderate. Filters 1, 2, and 3 not only did not improve contrast for either the smooth-finish or matte-finish screens, but resulted in poorer contrast under several conditions. Since the phosphor used in both VDTs was a P-4 white phosphor, these results indicate that color filters might not be expected to improve contrast unless the filter color is matched to that of the phosphor. Filter 7 (green, flat, smooth finish), however, performed well under the diffuse reflec- tion condition, but very poorly under the specular reflection condition. In general, filters are more effective in reducing diffuse reflections than in reducing specular reflections. This is unfortunate because specular reflections cause the greater loss of contrast and probably contribute more to problems encountered in viewing VDTs.

Next: 5. Lighting and Reflections »
Video Displays, Work, and Vision Get This Book
×
Buy Paperback | $60.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Along with the widespread use of computers have come growing fears that working in front of video display terminals (VDTs) can irritate and even damage the eyes. Separating scientific fact from popular opinion, this report takes a critical look at the link between VDT use and eye discomfort and disease as well as at changes in visual performance and oculomotor function. Drawing on information from ergonomics, illuminating engineering, and industrial and organizational psychology, the report gives practical advice on optimal workstation design to improve the comfort, performance, and job satisfaction of VDT users.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!