of Effects Illumination and Viewing Angle on the Modeling of Flicker Perception in CRT Displays by Shane Sidebottom
Thesis submitted to the Faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE in Industrial and Systems Engineering
______________________________ R. J. Beaton, Chairman
____________________________ A. M. Prestrude
____________________________ J. Deighan
April 1997 Keywords: Flicker, Critical Fusion Frequency, Illuminance, Luminance, CRT
Effects of Illumination and Viewing Angle on the Modeling of Flicker Perception in CRT Displays by Shane Sidebottom Dr. Robert J. Beaton, Chairman Industrial and Systems Engineering Department (ABSTRACT)
This study evaluated the usefulness of a psychophysical model as part of a new ANSI/HFES 100 standard for CRT flicker. A graph based flicker prediction method developed from Farrell, 1987 was evaluated. The Farrell model is based on phosphor persistence, screen luminance, display size, and viewing distance. The graph based method assumes a worse case scenario (i.e. a white display screen shown on a display with P4 phosphor). While the Farrell model requires photometric measurements to be taken using special equipment, the graph based method require a knowledge of the display size, viewing distance, screen luminance, and refresh rate. Ten participants viewed different display sizes from different eccentricities under different levels of illumination and luminance. In each condition the display's refresh rate was manipulated using the Method of Limits to determine the critical flicker frequency (CFF). An Analysis of Variance was used to detirmine significant effects on CFF. CFF increased with increasing luminance and display size. Adequate illumination significantly increased CFF. A viewing eccentricity of 30 degrees (measured horizontally from the center of the screen) produced the highest CFF values. Under the conditions of 30 degrees eccentricity and 250 to 500 lux illumination, observed 50% CFF threshold values exceeded the 90% CFF threshold values predicted by the graph based method. This study demonstates that when tested under the same conditions it was developed under, the Farrell method successfully predicts flicker perception; however, when tested under conditions representative of real world working conditions, the Farrell model fails to predict flicker perception. New parameters for the model are suggested.
i
Acknowlegements I would like to thank everyone who has helped me throughout my graduate career at Virginia Tech. I would especially like to express my gratitude to the members of my committee: Dr. Beaton, Dr. Prestrude, and John Deighan. I thank them for all their help. I would especially like to thank Dr. Beaton for all his advice, encouragement and patience. From my heart I thank my family. Without them this work would have been impossible.
ii
TABLE OF CONTENTS INTRODUCTION .............................................................................................................1 LITERATURE REVIEW ...................................................................................................3 Flicker...........................................................................................................................3 Factors from Psychophysics.........................................................................................3 Factors from CRT Research.........................................................................................6 Individual Variability ..................................................................................................7 Luminance .................................................................................................................8 Dark Adaptation.........................................................................................................8 Refresh Rate .............................................................................................................9 Screen Pattern...........................................................................................................9 Display Size and Angle of View.................................................................................9 Eccentricity ..............................................................................................................10 Illuminance ..............................................................................................................11 Phosphor Persistence .............................................................................................11 Farrell's Equation .......................................................................................................13 METHOD .......................................................................................................................18 Participants.................................................................................................................18 Equipment ..................................................................................................................18 Experimental Conditions.............................................................................................18 Procedure ...................................................................................................................19 Experimental Design ..................................................................................................20 RESULTS ......................................................................................................................21 REFINING THE FARRELL MODEL...............................................................................30 DISCUSSION ................................................................................................................32 REFERENCES ..............................................................................................................34 Appendix A. Setup and Calibration of Test Equipment .................................................37 Fox Video Generator Programming.........................................................................37 Confirmation of the Test Images..............................................................................39 Appendix B. Pre-Screening Instructions........................................................................41 Appendix C. Questionnaire ...........................................................................................42 Appendix D. Informed Consent......................................................................................43 Appendix E. Participant Instructions..............................................................................45
iii
LIST OF FIGURES Figure 1. Farrell CFF curves for the ANSI/HFES 100-199x standard. ............................2 Figure 2. Flicker sensitivity and retinal illumination.........................................................5 Figure 3. Flicker sensitivity and target size. ....................................................................6 Figure 4. Cumulative percentage CFFs for different observers ......................................8 Figure 5.Eccentricity and visual angle...........................................................................11 Figure 6. CFF for different CRT phosphors...................................................................12 Figure 7. Flicker thresholds as a function of the logarithm of energy observed............15 Figure 8. Farrell's regression equation based on the logarithm of energy observed. ...16 Figure 9. Main Effect of size on CFF.............................................................................23 Figure 10. Main effect of illumination on CFF ...............................................................24 Figure 11. Main effect of luminance on CFF .................................................................25 Figure 12. Main Effect of eccentricity on CFF ...............................................................26 Figure 13. The size x illumination interaction. ...............................................................27 Figure 14. The size x luminance interaction..................................................................28 Figure 15. The luminance x eccentricity interaction. .....................................................29 Figure 16. New CFF curves compared with Farrells. ...................................................31 Figure 17. Luminance and refresh rate before and after FORTH subroutines..............37 Figure 18. Oscilloscope reading at 55.5 Hz ..................................................................39 Figure 19. Oscilloscope reading at 74.5 Hz. .................................................................40 Figure 20. Oscilloscope reading for 94.5 Hz. ................................................................40
LIST OF TABLES Table 1. Farrell m and n parameters for 30 visual angle. ............................................14 Table 2. Farrell m and n parameters for different display sizes. ..................................15 Table 3. ANOVA Summary Table..................................................................................22 Table 4. New m and n values for 30 eccentricity .........................................................30
iv
Flicker Perception
1
INTRODUCTION Our time often has been called the information age. Millions of people earn their livelihood by spending a considerable part of each day processing information on Video Display Terminals (VDTs). Despite major advances in other display technologies, the Cathode Ray Tube (CRT) is the predominate technology. There is a good reason for this; as a display for desktop computer systems, the CRT has many advantages and few disadvantages. One disadvantage of the CRT is flicker. Flicker is the perception of extremely rapid changes in screen luminance - the screen appears to flash rapidly. Flicker is annoying and has been associated with visual discomfort. The most common complaint among VDT users is eyestrain (Sheedy, 1992). Flicker is believed to be a contributing factor to eyestrain. It is believed that flicker interferes with saccadic eye movements, causing the excessive muscular activity that leads to visual fatigue and eyestrain (Isensee and Bennett, 1983). Flicker also has been shown to affect reading speed adversely (Cushman, 1986). The American National Standard for Human Factors Engineering of Visual Display Terminal Workstations (ANSI/HFES 100-1988) specifies that displays shall be flicker free for 90% of the intended user population. To this end, ANSI/HFES 100 specifies a test procedure for evaluating CRTs. Unfortunately, this procedure can be difficult to perform, even more difficult to verify, and may not be representative of actual working conditions (Jones, 1996). A revision to the ANSI/HFES 100 standard is underway (ANSI/HFES 100-199x), which may allow the prediction of flicker perception in a quicker, easier, and more accurate manner. The pending ANSI/HFES 100-199x standard may use a psychophysical modeling equation developed by Farrell (1987) in two separate methods for predicting flicker: by graph and by measurement and analysis. The first method plots Critical Flicker Frequency (CFF) curves for a range of angular viewing dimensions on a graph of luminance vs. refresh rate (see Figure 1). To predict whether a display will appear to flicker, one only has to calculate the degrees of viewing angle of the situation (display size and viewing distance). If the refresh rate and luminance of the display lead to a coordinate above the CFF curve for the angular viewing dimensions calculated, the display is predicted to be flicker free. The graph-based method is a conservative test for flicker; it may predict that a display will flicker, when no flicker is actually observed. The second method is based on analysis of photometric measurements. It involves measuring the amplitude of the fundamental temporal refresh frequency, and then multiplying this value by the amplitude modulation to obtain an observed energy value. By using Farrell's equation to generate a predicted energy value at which CFF occurs and comparing it with the measured value, one can determine if a display will be flicker free. The purpose of this study was to examine the utility of the Farrell equation as part of the new ANSI/HFES 100-199x method for flicker evaluation. Research is needed to investigate the accuracy of the method developed by Farrell in predicting flicker perception under a variety of conditions. Different conditions of luminance,
Flicker Perception
2
illumination, viewing angle, and display size, were used to test how the method applies under different working environments.
Figure 1. Farrell's flicker prediction graph proposed for use in the ANSI/HFES 100-199x standard.
Flicker Perception
3
LITERATURE REVIEW Flicker Flicker is a visual perception phenomenon - if it is not perceived by an observer, it does not exist. "The phenomenon of flicker is due to the ability of the observer to detect changes in the luminance level when they occur at a rate below that at which the integrating capability of the eye eliminates the sensation of luminance change (Sherr, 1993)." The perception of flicker is produced when the intensity of a light source varies too slowly to appear to be of steady brightness. The frequency at which a light source must oscillate before it appears flicker free is called the Critical Fusion Frequency or Critical Flicker Frequency (CFF). For example, if the CFF of a display is 67 Hz for a specific observer, it means that when the temporal frequency is at or above 67 Hz the observer will not perceive flicker. When comparing situations (i.e. luminance, viewing distance, visual target size, frequency, illumination, etc. of the visual stimuli), higher CFF values mean that a situation has caused an increased sensitivity to flicker in the observer. The physical characteristic of visual stimuli underlying flicker is temporal frequency. With regard to CRTs, temporal frequency refers to refresh rate. A CRT works by continually redrawing or refreshing an image on the phosphor surface of a display. To do this requires system resources, such as bandwidth, which are not infinite. There is a tradeoff. Bandwidth can either be used for temporal (refresh rate) or spatial (resolution) information (Snyder 1980). Because of this tradeoff, it is important to know the minimum rate at which an image must be rescanned (refresh rate) so that the image remains flicker -free and system resources are conserved. Flicker is a predictable psychophysical response based on the temporal frequency of the image displayed. The electron gun at the back of the CRT scans successive lines of individual phosphor elements on the inside the tube. As the electron gun passes over a phosphor element, it excites the phosphor to emit light for a fraction of a second. After being energized, the light given off by the phosphor decreases or decays. Different phosphors have different decay characteristics; nevertheless, this periodic increase and decrease in light energy follows a complex wave form that can be modeled mathematically. Factors from Psychophysics Flicker research has been an important part of psychophysics for over 75 years. Many different methods were used to measure flicker perception. One of the very first and most common methods used a constant light source and an optical chopper disk. The chopper disk cutouts let light though in a periodic manner. Rotation of the disk caused the light to flash on and off rapidly. Later, more advanced electronic means were developed such as glow tubes and light emitting diodes (Brown, 1965). From this research many valuable insights have been gained; however, the results are not generalizable directly to CRT displays. Also, most of the experiments performed in
Flicker Perception
4
psychophysics used visual targets (such as disks or squares) of relatively small visual angles. Temporal frequency was a main factor of interest in psychophysics flicker research. It was discovered that the visual system can perceive flicker only if the frequency of the stimuli is below a certain temporal threshold. That threshold depends on the viewing conditions and also varies from observer to observer. The relationship between temporal frequency and flicker is expressed in the Talbot-Plateau Law, which states that if a light source oscillates at a high enough rate, it is perceived to match a steady light source of the same time-averaged luminance (Brown, 1965). The visual system integrates temporal information. Research in psychophysics has shown that visual target size affects flicker perception. At frequencies higher than about 20 Hz, CFF is known to increase with increasing target stimulus size. Under 20 Hz, sensitivity decreases with increasing target size (Keesey, 1972). According to the Granit-Harper Law, CFF increases linearly with the logarithm of the retinal area subtended by the target (Granit and Harper, 1930). For 1 targets at very low luminance levels (under 3 cd/m2) flicker is more easily perceived when the image falls on fovea, rather than in the periphery (Lythgoe and Tansley, 1929). This may appear to suggest that the fovea is more sensitive to flicker. However, the periphery is more sensitive to flicker when larger targets (such as flickering fields or CRT screens) are used. Roehig (1959) best demonstrated this by showing that CFF increases only when the outer circumference of a visual target is increased, while CFF remains the same when the inner circumference of a ring-shaped target of the same size is decreased. CFF is highest for a light-adapted eye. For a visual target falling on the fovea, dark adaptation decreases the observer's flicker sensitivity. Flicker sensitivity for peripheral targets also decreases with dark adaptation provided the target is relatively bright. A dim peripheral target causes flicker sensitivity to decrease for the first 5-10 minutes of dark adaptation and then increase until dark adaptation is complete (Brown, 1965). This occurs because dim targets allow for more complete dark adaptation to be achieved. When a large and bright visual target (such as a CRT) is viewed in a dark room, regardless of its location in the retina, only partial dark adaptation can occur and consequently flicker sensitivity will decrease. Hecht and Verrijp (1933) showed that CFF increases as luminance is increased. More precisely, DeLange (1958) demonstrated that for a visual stimulus with a fixed temporal frequency, sensitivity to flicker increases as the amplitude of the luminance modulation is increased. Flicker perception depends on the depth of modulation (modulation amplitude) and not only time-averaged luminance. Kelly (1961) demonstrated that at high frequencies flicker detection is dependent on the absolute amplitude of modulation and not the time-averaged luminance level at which the modulation is occurring around (see Figure 2). Flicker sensitivity curves of different levels of retinal illumination (time-averaged luminance) all " . . . form a common envelope. . ." at high frequencies (Cornsweet, 1970). Cornsweet (1970) concludes that the effect of illumination on flicker perception is minimal; " . . . if a light is sinusoidally
Flicker Perception
5
modulated just strongly enough to appear to flicker, and then a steady light is added to it, it will still be just at the threshold for flicker (Cornsweet, 1970)." Research by Kelly (1959) demonstrated that at high frequencies, flicker sensitivity curves for different sizes of visual stimuli have similar shapes (see Figure 3). This indicates that visual system responds to the temporal characteristics of a stimulus in a consistent fashion. Above frequencies higher that 20 Hz, flicker sensitivity increases as target size increases for targets with similar surrounds. Dark surrounds decrease flicker sensitivity.
Figure 2. Flicker sensitivity for different levels of average retinal illumination in trolands. A 60 target with blurred edges was used (Kelly, 1961).
Flicker Perception
6
Figure 3. Flicker sensitivity for different levels of target size, (a) 2 flickering target with a steady surround, (b) 65 target, and (c) 4 target with a dark surround (Kelly, 1959) Stimulus intensity was constant at 1000 trolands.
Factors from CRT Research Research on CRT flicker has verified the effects of many of the factors known from psychophysics and has identified other factors that influence flicker perception. Factors such as illumination, luminance, temporal frequency, and retinal target size and location are directly applicable to CRT induced flicker.
Flicker Perception
7
Individual Variability One factor that is not often examined in the flicker research conducted in psychophysics is individual variability. Some people are more sensitive to flicker than others (Rogowitz 1986). To develop an accurate flicker standard for the general population, the nature of the distribution of flicker sensitivity in the population must be considered. Individual differences may be attributable to many factors, such as gender, age, or personality. Young people are more sensitive to flicker than older people (Kim and Mayer, 1994). Women are more sensitive to flicker than men are (Maxwell, 1992). Personality has been cited by psychologists as affecting flicker perception (Amir and Ali, 1989). The factors that lead to individual variation in flicker sensitivity may be of little real use to those designing displays for the average user; however, knowledge of the extent of individual variability is useful. Within any given population there exists variability in flicker sensitivity. Research by Bauer, Bonacker, and Cavonius (1983) found that normal observers vary greatly in terms of flicker sensitivity. Thirty-one subjects viewed a CRT that subtended 30 x 30 of visual angle. The average screen luminance was set at 80 and 320 cd/m2. Mean CFF for the 80 cd/m2 condition was 72 Hz with a coefficient of determination (r2) of 0.97. CFF was 87 Hz for a 95th percentile observer. In the 320 cd/m2 condition, mean CFF was 78 Hz with a 0.98 r2 value. CFF was 95 Hz for a 95th percentile observer (see Figure 4). CFF values followed a Gaussian distribution. Rogowitz (1986) also supports the finding that flicker sensitivity follows a normal distribution. The visual stimulus used was a modulator glow tube with a square wave current drive. This was called a standard lamp by Rogowitz and was used in flicker matching experiments with CRT displays. The lamp subtended 2 of visual angle and had a luminance of 85 cd/m2 at 100 Hz (luminance modulation over time of 89.9%). Rogowitz measured the flicker sensitivity of over 100 observers and found that it followed a Gaussian distribution. The average CFF was 45.3 Hz with = 4.38. Because of the large number of observers tested, Rogowitz (1986) gives a good indication of the nature of the distribution of CFF.
Flicker Perception
8
Figure 4. Cumulative percentage CFFs for different luminance values as frequency increases (Bauer, Bonaker, and Cavonius, 1983). Luminance Average luminance would seem to be the key predictive characteristic of flicker perception: as luminance of a CRT increases, so does perceived flicker. Bauer, et al. (1983) showed that as the average luminance of a CRT is increased, flicker becomes more apparent to the observer. With regard to CRTs, average luminance is the timeaveraged amplitude of the waveform of light emitted. Based on research, Bauer, et al. estimated that for luminance ranges between 80 and 320 cd/m2 with refresh rates between 50 and 95 Hz " . . . CFF increases linearly with log luminance, at a rate of 11.6 Hz per log unit." While average luminance may be useful, it is not the key predictive factor of flicker perception. Research shows, " . . . that CFF can indeed be approximated using the log of mean luminance. However, the key to understanding flicker perception is found in mean luminance plus the 'contrast effect' or modulation amplitude (Maxwell, 1992)." Dark Adaptation Research involving flicker from CRTs has shown that a completely dark environment decreases flicker sensitivity. This corresponds with what is known from research in psychophysics; partial dark adaptation decreases flicker sensitivity to large, relatively bright targets. (Brown, 1965). Flicker has often been tested for in dark rooms (ANSI/HFES 100-1988). Some people make the assumption that if a CRT is viewed in a dark room, flicker sensitivity will be increased. Research has shown that is not the
Flicker Perception
9
case. Jones (1996) demonstrated that average flicker sensitivity significantly decreased when ambient illumination was 0 lux as compared to 500 lux. Refresh Rate For CRT displays, refresh rate is a driving factor of flicker perception. By definition, if the refresh frequency of a display is higher than the CFF of the observer in a giving environment, the observer will not see flicker. Refresh rate determines the frequency of the luminance oscillation from a CRT. Recommendations for minimum refresh rates are usually the focus of VDT flicker standards. The Video Electronics Standards Association (VESA) in 1991 set the acceptable refresh rate at a minimum of 70 Hz for monitors up to 17 inch diagonal (Maxwell, 1992). Under some conditions 60 Hz is entirely flicker free; however. under other conditions even 75 Hz may produce flicker. Bauer, et al. (1983) found that for a 30 x 30 VDU screen viewed in the periphery (30 horizontally from the fixed point of stare to the center of the CRT) at luminance level of 80 and 320 cd/m2, refresh rates must be above 87 Hz in with 80 cd/m2 and 95 Hz for 320 cd/m2 in order that 95% of the population will not perceive flicker. It must be noted that for large CRT monitors 80 cd/m2 is a very bright display. Screen Pattern From the late 1970s until the mid 1980s, nearly all text on a CRT was presented in negative contrast (bright characters on a dark background). This was especially true of text displayed on monochrome CRTs. Currently, there is a trend to use dark characters on a bright background (positive contrast). This form of text presentation is more natural looking (i.e., it appears the way people are accustomed to viewing text on paper). Radl (1980) showed that positive contrast text presentation allows for more accurate reading at greater speeds than to negative contrast. Cushman (1986) demonstrated that the average reading rate from paper is equivalent to the average reading rate from a CRT, when image polarity is the same (i.e. both have positive contrast). However, there are drawbacks to positive contrast presentation. Isensee and Bennett (1983) report that positive contrast can result in increased visual fatigue. Compared with negative contrast presentation, positive contrast results in increased average screen luminance. If the refresh rate is not increased to compensate for the increased luminance, an increase in perceived flicker results. The effects of positive contrast presentation on flicker perception can vary based on the size, font, and spacing of the characters. Typical positive presentation of text results in about 85% of the pixels illuminated. Because a white CRT screen has every pixel element activated, it has the highest possible average screen luminance. Research has demonstrated that white screens have the highest CFF value of any screen pattern (Jones, 1996). Display Size and Angle of View Display size influences flicker perception because larger displays subtend larger visual angles on the retina at a given viewing distance. A large display projects an
Flicker Perception
10
image that covers more of the retina than the image on a smaller display, and more of its image falls outside of the foveal area. The formula for calculating degrees of visual angle from display size and viewing distance as given in Farrell (1987) is: = 360/[tan-1(D/2Ld)] where: = viewing angle in degrees D = screen diagonal in mm Ld = viewing distance in mm (Eq. 1)
Degrees of visual angle and time-averaged luminance can be used to calculate the area of retinal illumination. The effect of a small display viewed up close is the same as that of a large display viewed far away, if both situations produce the same visual angle and time-averaged luminance. Eccentricity The extent to which a display is viewed in the periphery is called eccentricity, and it is known to affect flicker perception. Bauer, et al. (1983) demonstrated that flicker perception increases significantly when a CRT is viewed at 30 horizontally off center from a focal point (see Figure 5). Jones (1996) supported this finding by showing that when a display is viewed in the periphery (fixation point at 20 past the edge of the active area of the CRT), the CFF is significantly higher than if the display is viewed directly (fixation point at the center of the CRT). Eccentricity is important to consider since much work is done with the CRT in the periphery. Also, when viewing a very large CRT directly, much of the screen is in the periphery (horizontally or vertically >10 from point of fixation).
Flicker Perception
11
Figure 5. Eccentricity and visual angle subtended affect CFF values (Bauer, Bonaker, and Cavonius, 1983).
Illuminance Illuminance on the CRT screen can affect flicker perception to a small degree. The amount by which room illumination increases the average luminance (i.e. both reflected and emitted) of the CRT is usually quite small. In an office environment, illumination is usually between 300 and 500 lux (Sanders and McCormick, 1993). The ANSI/HFES 100 standard recommends between 200 and 500 lux of ambient illumination. Very high levels of illumination can make a CRT appear low in contrast and difficult to view. Very low levels of illumination can make it difficult to read source documents or other printed materials. Jones (1996) found that flicker sensitivity in a dark room (0 lux illumination) is significantly lower than for a room with adequate illumination (500 lux). Phosphor Persistence Phosphor persistence influences flicker perception. Long persistence phosphors are less likely to cause flicker; however, they can cause dynamic images to smear. Short persistence phosphors are more prone to flicker and require higher refresh rates but allow for smooth image motion. Turnage (1966) showed that CFF is related to the persistence characteristics of the phosphor used. In an experiment with seven different CRT phosphors and three luminance levels, Turnage found that on average P20 phosphor, a medium short persistence phosphor showed the most flicker. In a study
Flicker Perception
12
comparing eight common types of CRT phosphors, Bryden (1966) demonstrated that over a range of luminance and refresh rates P4 phosphor produces the most flicker (see Figure 6). Since 1966, many additional types of phosphor have been developed. Pearson (1991) points out that: modern phosphors are often a physical blend of two or more component phosphors, each phosphor having its own distinct persistence characteristics, and many phosphors have non-exponential persistence characteristics. This complicates the matter of how phosphor persistence affects flicker perception.
Display Luminance (candelas/meter2)
Figure 6. CFF plotted as a function of refresh rate and average display luminance for different CRT phosphors (Bryden, 1966). Typically, phosphor persistence has been described in terms of microseconds until a 10% decay in luminance occurs. Because of the complexity of phosphor blends, a better way of describing persistence is the modulation index of the fundamental frequency (also called the "ripple ratio", Turnage, 1966). This ratio is calculated by dividing the amplitude of the Fourier series term containing the fundamental frequency by the average value of the waveform (Pearson, 1991). The first Fourier fundamental represents the primary sinusoidal component of the complex temporal waveform. Modulation is the ratio of varying luminance to time-averaged luminance (Kelly, 1972). For refresh rates between 50 and 80 Hz, the medium short P4 phosphor has a ripple ratio of 1.97, one of the highest of any phosphor. By comparison, a long persistence phosphor such as PC214 has a ripple ratio of 0.44 (Pearson, 1991). Theoretically, the highest ripple ratio is 2, which indicates a 200% modulation (Farrell, 1987). Ripple ratios are very useful in calculating the absolute energy that comes from
Flicker Perception
13
a CRT. While ripple ratios do give some insight into predicting flicker perception, other factors must also be considered. Farrell's Equation The luminance emitted by a CRT follows the pattern of a complex waveform. DeLange (1958, 1961) and Kelly (1961, 1969, 1971) demonstrated that by using Fourier analysis this complex temporal wave form could be analyzed and the CFF could be predicted. It was found that CFF could be predicted using the frequency and absolute amplitude of the lowest or fundamental Fourier series term. Because the frequency of the fundamental in CRTs is equal to the refresh frequency, all that is needed to predict flicker perception at frequencies greater than approximately 20 Hz is the absolute amplitude. Farrell (1987) developed an equation for predicting flicker based on the absolute amplitude of the fundamental frequency. The equation developed by Farrell for the prediction of flicker perception is: CFF = m + n[ln{Eobs}] (Eq. 2) where m = coefficient derived from linear regression n = coefficient derived from linear regression Both m and n values come from linear regression and fit the energy observed curve to the actual CFF curve The frequency at which flicker fusion occurs is dependent on the observed energy of the fundamental frequency given by: Eobs = DC x c1/c0 (Eq. 3) in which DC = (Lt - Lr)A A = Area of the retina illuminated Lt = total luminance emitted from the CRT in cd/m2 Lr = light reflected from the CRT from illumination (cd/m2) c1/c0 = temporal modulation index of the fundamental frequency based on the phosphor persistence (ripple ratio) The luminance reflected from the CRT, Lr, is subtracted from the total luminance emitted, Lt, because as Kelly (1959) indicated, at high frequencies absolute amplitude modulation is what determines whether flicker detection occurs. Calculating the area of retinal illumination is the most accurate way of measuring the how much light illuminates the retina of the eye. The formula for retinal illumination used by Farrell is modified from Crawford (1936) and is given by: A = b0 x Ltb1 in which b0 = 12.45284 b1 = -0.16032 Lt = total light from CRT (in cd/m2) (Eq. 4)
Flicker Perception
14
The CFF equation developed by Farrell (1987) and plotted for P4 phosphor, with angular viewing dimensions of 10, 30, 50, and 70 has been proposed for use in the ANSI/HFES 100-199x standard as a graph based method for quickly estimating if flicker will be seen under a certain conditions. The m and n parameters used for the ANSI/HFES 100-199x standard are given in Table 2. These values are from the linear regression equation of the natural logarithm of observed energy values that Farrell found to be flicker free for 90% of the observers (see Figure 8). Twenty "young rested observers" between the ages of 23 and 39 were used as subjects. The CFF curves for a range of different angular viewing dimensions is based on this data (see Figure 9). Using linear regression (see Figure 7), Farrell determined values for parameters called m and n that allow the CFF equation to fit observed CFF values (see Table 1). These values were calculated for a display that subtended 30 degrees of visual arc from data of previous studies. Later, more studies were performed and m and n values were obtained for a variety of angles of visual arc (see Table 2) and a linear regression equation was developed for CFF values for the flicker thresholds of 90% of observers.
Table 1. Parameters m and n were first obtained from regression equations generated by several studies (Farrell, 1987).
CFF = m + n [ln(E(f)obs)] Parameter Display Size 10 30 50 70 m 11.48 13.87 8.32 6.79 n 7.50 8.31 9.73 10.03 0.994 0.998 0.977 0.991 R2
Flicker Perception
15
Figure 7. Flicker thresholds were plotted a function of the logarithm of observed luminance energy and frequency (Farrell, 1987).
Table 2. Farrell (1987) found parameter values of m and n for different visual angle display sizes. Note: Farrell (1987a) is referenced in this study, but Farrell (1987b) is from an unpublished paper referenced in Farell (1987a).
CFF = m + n [ln(E(f)obs)] Parameter Data m Eriksson & Backstrom (1987) Farrell (1987b) Farrell (1987a) Farrell (1986) 13.87 15.76 24.27 17.06 n 8.31 7.45 7.33 7.72 0.998 0.933 0.984 0.933 20 12 10 6 R2 N Subjects
Flicker Perception
16
Figure 8. Farrell (1987) plotted a regression equation based on the logarithm of energy from a CRT observed. According to Farrell (1986), room illumination should not affect flicker threshold, " . . . unless the display is very dim (where, presumably, only the rods are responding) and the room illumination is very high (where cones are saturated), . . .. " Farrell makes this assumption based on research by Kelly (1964) which demonstrated that the visual system acts as a linear system in analyzing absolute amplitude of temporal frequencies at higher frequencies. In calculation of the energy observed (Eobs), illuminance (ambient light reflected from the screen of the CRT) is subtracted from the total light coming from the CRT. In this way, illuminance is not considered by Farrell to be a factor that significantly affects flicker perception. The Farrell equation makes no mention of eccentricity. These methods assume the display is viewed straight ahead. Previous research indicates that CFF values are considerably higher when a display is viewed in the periphery. Farrell (1987) does not make mention of how subjects were instructed to look for flicker. As Bauer, et al. (1986)
Flicker Perception
17
demonstrated, sensitivity to flicker is increased when subjects are instructed to view the edge of the CRT rather than its center. The objective this study was to determine if the Farrell method accurately predicts flicker perception under conditions representative of actual working environments. This study examined how the factors of luminance, illumination, display size, and viewing eccentricity affected the psychophysical modeling of flicker. Research from psychophysics has indicated that adding a steady light to a high frequency flickering stimulus does change not the flicker threshold (Kelly, 1961), yet research has also indicated that a dark room results in significantly lower sensitivity to flicker than a well lighted room (Jones, 1996). Research has indicated that viewing eccentricity has a significant effect on flicker perception (Bauer, et al. 1983), yet this is not accounted for in the Farrell equation. The well-established effects of display size and illuminance on flicker perception will be tested to determine if Farrell's equation adequately accounts for these factors.
Flicker Perception
18
METHOD Participants Ten paid volunteers (five males and five females) from the Virginia Polytechnic Institute and State University took part in this study. All participants were between 18 and 35 years of age (average age: 26.3 yr.) and passed optometric tests for vertical phoria, lateral phoria, color normalcy (Dvorine Pseudoisochromatic Plates), and near point acuity (i.e. 20/30 or better with correction). Participants were free of any visual pathology or any other health problems that might have affected the results of the experiment. Due to the possibility of flicker induced seizures, individuals with a history of epilepsy were not be permitted to participate in this study. Equipment A color CRT monitor (Model: NEC Multisync 6FGp, 19 in. diagonal) driven by a high-bandwidth programmable video generator (Fox Quantum Data 8701, 400 Mhz) was used present the test screens at refresh rates between 55.5 and 94.5 Hz. The video generator was programmed in FORTH to ensure consistent image luminance. A chin rest was used to maintain a viewing distance of approximately 550 mm. A mixture of incandescent and florescent ceiling mounted light sources provided illumination. The light sources were fully adjustable and evenly spaced to provide uniform illumination to the work area. The active area of the CRT was masked by foam core bezels to reveal viewable areas of 14, 16, and 18 diagonal. Peripheral targets were located on a bar extending from the CRT at 15, 30, and 45 degrees measured from the center of the screen. Experimental Conditions The walls and ceiling of the testing room were painted 18% reflectance matte gray to reduce glare. The experimenter was separated from the participant by 6 partitions to prevent the participant from obtaining any performance feedback that could bias responses. An adjustable padded metal chin rest was used to ensure participants maintained a viewing distance of approximately 550 mm. The 19 diagonal display used in this experiment was masked by white foam core bezels to reveal 14, 16, or 18 diagonal viewable screen areas with width to height aspect ratios of 4:3. The 14 viewable area measured 289 mm by 214 mm, thus subtending approximately 36 of visual angle measured diagonally when viewed from 550 mm. The 16 viewable area measured 322 mm by 242 mm, thus subtending 40 of visual angle measured diagonally at 550 mm. The 18 display measured 364 mm by 273 mm, thus subtending 45 of visual angle measured diagonally at a distance of 550 mm. A white wooden frame held each foam core bezel in place. Ambient illumination was provided at 0, 250, and 500 lux, measured from the plane tangent to the center of the display measured using an illuminometer (Model: Minolta). Ceiling mounted light fixtures consisting of a bank of adjustable fluorescent lights and three adjustable incandescent lights evenly illuminated the workspace. In the
Flicker Perception
19
0 lux condition, all light sources were turned off. In the 250 lux condition, fluorescent lighting provided about 70% of the illumination with incandescent lighting providing the rest. In the 500 lux condition, florescent lighting provided about 66% of the illumination with incandescent lighting providing the rest. Two levels of display luminance were examined, a 20 cd/m2 white screen and a 60 cd/m2 white screen. Time-averaged luminance was measured at the center of the screen with a spot photometer (Model: Minolta CS-100). The video generator was programmed to ensure that luminance remained constant as refresh rate changed. Participants viewed test screens from four different angles: 0 (center of the screen), 15, 30, and 45 to the left of the center of the screen. The chair that participants sat in had a swiveling base. Participants were asked keep facing straight ahead and swivel in their chairs when required to look at peripheral fixation points. Fixation points consisted of black dots measuring approximately 8 mm in diameter located on a 3 cm wide white bar attached to the left side of the CRT. The peripheral fixation points were located at approximately the same distance from the participant as the center of the surface of the CRT screen. Procedure Before beginning the experiment, participants received a brief explanation of the data collection protocol and were given an informed consent form to read and sign. Participants completed a questionnaire regarding epilepsy and visual pathology. All individuals with a history of epilepsy were screened from this study due to the danger of flicker induced seizures. Participants then were given tests for vertical phoria, lateral phoria, visual acuity, and normal color vision. A flickering screen (55 Hz at 60 cd/m2) and a non-flickering screen (94.5 Hz at 60 cd/m2) were shown to the participants and any questions were answered. In the 0 level of eccentricity, participants were instructed to look at the center of the screen. In each of the remaining conditions of eccentricity, participants were instructed to look at one of three small fixation points (each peripheral target dot was numbered 1, 2, or 3) while focusing their attention on the display. Participants were instructed to respond "yes" if they detected flicker and "no" if they did not detect flicker. Subjects were instructed to respond yes if any flicker appeared anywhere on the screen. The Method of Limits with random starting points was used to measure CFFs in all conditions. This method consists of alternating ascending and descending trials. Refresh rate was changed in 1.5 Hz steps. In an ascending trial, refresh rate was set low and increased until the participant indicated he or she could no longer perceive flicker. In descending trials refresh rate was set high and decreased until the participant indicated he or she could perceive flicker.
Flicker Perception
20
Experimental Design A four factor, within-subjects design was used to collect data under the various viewing conditions. The dependent variable was display refresh rate (measured in Hertz) that produced a no-flicker response from the observers. The four independent variables were: Display Size: 14, 16, or 18 diagonal screen areas. Ambient Illumination: 0, 250, and 500 lux Display Luminance: a 20 cd/m2 white screen and a 60 cd/m2 white screen. Eccentricity: 0 (center of the screen), 15, 30, and 45 to the left of the center of the screen. In this study there were a total of 72 treatment conditions (3 screen size levels x 3 illumination levels x 2 luminance levels x 4 eccentricity levels). Data collection was blocked by display size. Within each display size block, data collection also was blocked by ambient illumination. All blocks were presented to each participant in a unique random order sequence. Participants received all levels of eccentricity and luminance in random order before illumination or display size was changed. Participants were allowed to adapt to each level of illumination before beginning each illumination block.
Flicker Perception
21
RESULTS The CFF values for all participants were subjected to a four-factor, withinsubjects Analysis of Variance (ANOVA) procedure using the Greenhouse-Giesser correction for sphericity. Least Significant Difference and Newman-Keuls tests were conducted on the significant interactions. The results are listed in Table 3. The main effects of display size (F[2,18] = 12.259, p = 0.0012), illuminance (F[2,18] = 11.401, p = 0.0028), luminance (F[1,9] = 164.897, p = 0.0001), and eccentricity (F[3,27] = 25.145, p = 0.0001) were significant. The two-way interactions of size x illumination (F[4,36] = 4.080, p = 0.0161), size x luminance (F[2,18] = 4.004, p = 0.0454), and luminance by eccentricity (F[3,27] = 6.748, p = 0.0056) also were significant. Overall average CFF was 65.146 Hz, with = 4.962. An Anderson-Darling Normality test indicated CFF values were normally distributed.
Flicker Perception
22
Table 3. ANOVA Summary Table. Source subj size size * subj Illum illum * subj lum lum * subj ecc ecc * subj size * illum size * illum * subj size * lum size * lum * subj illum * lum illum * lum * subj size * ecc size * ecc * subj illum * ecc illum * ecc * subj lum * ecc lum * ecc * subj size * illum * lum size * illum * lum * subj size * illum * ecc size * illum * ecc * subj size * lum * ecc size * lum * ecc * subj illum * lum * ecc illum * lum * ecc * subj size * illum * lum * ecc size * illum * lum * ecc * subj df 9 2 18 2 18 1 9 3 27 4 36 2 18 2 18 6 54 6 54 3 27 4 36 12 108 6 54 6 54 12 108 Sum of Squares 5149.416 1301.018 955.154 707.372 558.395 3060.937 167.064 1885.694 674.945 297.287 655.808 22.259 50.033 .699 70.029 21.254 208.596 19.845 232.707 62.444 83.285 9.613 109.092 34.853 546.284 39.837 206.910 10.329 188.136 68.099 315.782 Mean Square 572.157 650.509 53.064 353.686 31.022 3060.937 18.563 628.565 24.998 74.322 18.217 11.130 2.780 .350 3.891 3.542 3.863 3.307 4.309 20.815 3.085 2.403 3.030 2.904 5.058 6.639 3.832 1.722 3.484 5.675 2.924 F-Value PG-G 12.259 11.401 164.897 25.145 4.080 4.004 .090 .917 .767 6.748 .793 .574 1.733 .494 1.941 .0012 .0028 .0001 .0001 .0161 .0454 .8844 .4489 .5293 .0056 .5062 .6714 .1592 .7252 .1176
Flicker Perception
23
Size Display size had a significant effect on CFF. Average CFF increased 3.08% from 14 to 16 displays and 5.16% from 14 to 16 displays (14 - 63.407; 16 - 65.357; and 18 - 66.679). The main effect of size on CFF is shown in Figure 9. The average CFF at 18 was significantly higher than the CFF at 14. Pairwise comparisons failed to show a significant difference between 14 and 16 or between 16 and 18 conditions.
71 70 69 68 67 66 65 64 63 62 61 60 14" 16" S ize 18"
Figure 9. Main Effect of size on CFF. Error bars show +/- 1 standard error of the mean. Illumination Illuminance had a significant effect on CFF. Average CFF increased 3.14% from 0 lux to 250 lux and 3.44% from 0 lux to 500 lux (0 lux - 63.750; 250 lux - 65.749; and 500 lux - 65.944). The main effect of illumination is shown in Figure 10. The average CFF at 0 lux illuminance was significantly lower than the average CFFs at 250 lux and 500 lux. Pairwise comparisons failed to show a significant difference between the average CFF at 250 lux and the average CFF at 500 lux illuminance.
CFF
Flicker Perception
24
71 70 69 68 67 66 65 64 63 62 61 60 0 lux 250 lux
Illu m ination
CFF
500 lux
Figure 10. Main effect of illumination on CFF. Error bars show +/- 1 standard error of the mean.
Luminance Luminance had a significant effect on CFF. Average CFF increased 6.54% from 20 cd/m2 to 60 cd/m2 (20 cd/m2 - 63.086; 60 cd/m2 - 67.209). The main effect of luminance is shown in Figure 11. The average CFF at 20 cd/m2 was significantly higher than the average CFF at 60 cd/m2.
Flicker Perception
25
71 70 69 68 67 66 65 64 63 62 61 60 59 58 20 cd/m2 60 cd/m2
CFF
L u m inance
Figure 11. Main effect of luminance on CFF. Error bars indicate +/- 1 standard error of the mean.
Eccentricity Eccentricity had a significant effect on CFF. Average CFF increased 0.86% from 0 to 15, 6.1% from 0 to 30, and 4.73% from 0 to 45 (0 - 63.299, 15 - 63.841, 30 - 67.159, and 45 - 66.291). The main effect of eccentricity is shown in Figure 8. Pairwise comparisons indicate that CFF at 0 is not different from CFF at 15, nor is CFF at 30 different from CFF at 45. However, both CFF at 0 and CFF at 15 are different from both CFF at 30 and CFF at 45.
Flicker Perception
26
71 70 69 68 67 66 65 64 63 62 61 60 0 deg 15 deg 30 deg 45 deg
CFF
Eccentricity
Figure 12. Main Effect of eccentricity on CFF. Error bars indicate +/- 1 standard error of the mean.
Interactions The significant ANOVA interactions follow main effect data trends. Three twoway interactions were significant: size x illumination, size x luminance, and luminance x eccentricity. Size x illumination: Figure 13 shows the two-factor interaction of size and Illuminance on CFF values. Overall, average CFF increases as illumination increases. A Least Significant Difference (LSD) test indicated that the condition of a 250 lux, 14 screen is different from a 250 lux, 18 screen; a 0 lux, 14 screen; and a 0 lux, 16 screen. The condition of a 500 lux, 18 screen is different from a 14 screen at either 0 or 250 lux. No other conditions were significantly different. These results demonstrate the importance of testing for flicker under adequate conditions of illumination.
Flicker Perception
27
70 69 68 67 66 65 64 63 62 61 60 0 lux 250 lux
Illum ination
18" 16" 14"
CFF
500 lux
Figure 13. The size x illumination interaction. Error bars indicate +/- 1 standard error of the mean.
Size x luminance Figure 14 shows the two-factor interaction of size x luminance on CFF results. The main effect trends for size and luminance are reflected in this interaction. A Newman-Keuls test indicated that CFFs in all 20 cd/m2 conditions are different from CFFs in all 60 cd/m2 conditions, except an 18 screen at 20 cd/m2 is no different from a 14 screen at 60 cd/m2. In the 60 cd/m2 conditions, the 14 display size is different from the 18 display size; however, 14 displays are not different from 16 displays, nor are 16 displays different from 18 displays. Likewise, a 20 cd/m2 screen is only significantly different at the 14 and 18 sizes. These findings clearly demonstrate the linear relationship between screen size and luminance.
Flicker Perception
28
70 69 68 67 66 65 64 63 62 61 60 14" 16"
Size
60 cd/m2 20 cd/m2
CFF
18"
Figure 14. The size x luminance interaction.
Luminance x eccentricity Figure 15 shows the two-way interaction between luminance and eccentricity. A Newman-Keuls test indicated that: in the 20 cd/m2 conditions an eccentricity of 0 is not different from 15, nor is a 30 different from 45. Similarly, in the 60 cd/m2 conditions an eccentricity of 0 is not different from 15, nor is 30 different from 45. The tests also found that conditions of 20 cd/m2 at 30 and 45 are not different from conditions of 60 cd/m2 at 0 and 15. All remaining conditions are different from each other. These findings are consistent with previous studies that demonstrated that peripheral viewing dramatically increases flicker perception (Jones, 1996; Bauer, et al. 1983).
Flicker Perception
29
71 70 69 68 67 66 65 64 63 62 61 60 0 deg 15 deg 30 deg 45 deg
CFF
60 cd/m2 20 cd/m2
E c c e n tric ity
Figure 15. The luminance x eccentricity interaction.
Flicker Perception
30
REFINING THE FARRELL MODEL Average CFF values obtained under conditions of 250 and 500 lux and 30 eccentricity were much higher on average than the CFF values obtained under any of the other conditions. Because CFF values at 30 eccentricity and 250 and 500 lux illumination conditions were not significantly different within display size, these values were averaged within display size. By taking these CFF values and adjusting them by the sample standard deviation ( = 4.962), new values that conservatively reflect 90th percentile observers were estimated for each of the display sizes considered. Based on these 90th percentile CFF values at 20 cd/m2 and 60 cd/m2, a new linear regression equation was developed for each display size. New m and n parameters then were calculated that more accurately accounted for observed flicker perception under conditions of 250 and 500 lux and 30 eccentricity. While the Farrell graph-based method was intended to be a conservative method for predicting flicker perception, under well-lighted, peripheral viewing conditions it fails to adequately estimate flicker among observers. Table 5 lists the new parameters, and Figure 13 shows the new curves compared to the Farrell graphmethod curves. The new m and n parameters are like the original parameters in the Farrell model; they are additive and multiplicative coefficients derived from linear regression of observed CFF values. Furthermore, the new m and n values are adjusted by standard deviation to give values that are intended to be truly flicker-free for 90% of the user population. Farrells energy observed values are retained in the calculation of the new m and n parameters because the accuracy obtained from using these values (under 0 eccentricity) has been demonstrated. Also, in keeping with the worst-case scenario devised in the Farrell graph-based method, a phosphor persistence value of 2 (representing P4 phosphor) was used in these calculations. The majority of phosphors used in VDTs are of the medium-short variety because longer persistence phosphors tend to cause image smear.
Table 4. New m and n values for 30 eccentricity and between 250 and 500 lux illumination CFF = m + n(ln(Eobs)) Display Size 14 diagonal at 550 mm 36 visual angle 16 diagonal at 550 mm 40 visual angle 18 diagonal at 550 mm 45 visual angle Parameters m 33.393 37.097 36.831 n 5.145 4.988 5.353
Flicker Perception
31
80 78 76 74 72
CFF
70 68 30 deg 66 64 62 60 36 deg 40 deg 45 deg 50 deg
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
Luminance
Figure 16. New CFF curves (for 36, 40, and 45 visual angle based on CFFs at 30 eccentricity and between 250 and 500 lux illumination) compared with Farrells curves for 30 and 50 visual angle. Note: observations were only taken 2 2 at 20 cd/m and 60 cd/m . All data between these values are interpolated.
Because only two luminance levels were tested in this study, precise statements of the amount of variance accounted in regression across luminance can not be made. However, the CFFs for the two luminance levels tested in this study are based on a large number of observations, and the findings of this study are consistent with previous research.
64
Flicker Perception
32
DISCUSSION The purpose of this research was to examine the utility of the Farrell method in the prediction of flicker perception. In particular, this study examined how well Farrells graph-based method predicted flicker perception under a variety of viewing conditions. As expected, when tested in the conditions for which it was developed (i.e. 0 eccentricity, 0 lux illumination), Farrells equation correctly predicted flicker perception. For example, Farrells equation predicted that for a 20 cd/m2 screen subtending a visual angle between 30 and 50, 90th percentile average CFF would be between 61.49 Hz and 64.08 Hz. This study found that 50th percentile average CFF values (0 eccentricity, 0 lux illumination) for a 20 cd/m2 screen were as follows: at 35, 60.033 Hz; at 40, 61.94 Hz; and at 46, 62.791 Hz. It must be noted that Farrells predicted refresh frequencies are intended to be flicker free for 90% of the population, while the values reported in this study are average observed CFFs. Likewise, at luminance levels of 60 cd/m2 the Farrell equation correctly predicted flicker perception. The Farrell equation predicted that CFF values would not exceed 73.05 Hz for a display that subtended less than 50 visual angle. At 60 cd/m2 (0 eccentricity and 0 lux illumination) displays of 36, 40, and 45 visual angle produced CFF values of 63.674 Hz, 65.019 Hz and 66.338 respectively. When tested under conditions more representative of working environments, Farrells equation did not always correctly predict flicker perception. For example, the Farrell equation predicted that a 20 cd/m2 display subtending less than 50 visual angle, would be flicker-free for 90% of the population if refreshed at any rate over 64.08 Hz. However, this study found that average CFFs for the three screen sizes tested in this experiment exceeded 64.08 Hz when illumination was 250 lux or higher and when viewing angle was 30 or higher. Eccentricity and illumination were found to affect CFF in the same manner in the 60 cd/m2 conditions. Farrells equation predicted that a 60 cd/m2 display subtending less that 50 visual angle would be flicker-free for 90% of the population if refreshed at any rate over 73.05 Hz. However, this study found that the average (50th percentile) CFF for a 60 cd/m2 display subtending 45 visual angle and viewed under 250 lux illumination and from 30 eccentricity was 73.464 Hz. Because the viewing angle of 30 eccentricity and the illumination levels of 250 and 500 lux produced the most flicker, the CFF values under these conditions were adjusted using standard deviation to reflect 90th percentile observers and used to calculate new m and n parameters. As shown, the new prediction curves reflect an increased sensitivity to flicker due to the effects of eccentricity and illumination. While the new curves are established on observations from only two luminance levels, they are based on a large number of observations and readings and are consistent with previous studies. For 90% of the observers in this study, the new curves show refresh rates that are flicker-free. Since the ANSI/HFES 1988-100 standard was adopted, display technology has made many advances. In light of the current state of the art, it is recommended that the new standard recommend that displays shall be flicker free for 99% of the intended
Flicker Perception
33
user population. Such a standard is well within the capabilities of todays video display technologies. Based on the findings of this research, it is recommended that the pending ANSI/HFES 100-199x standard employ the Farrell equation, with new parameters based on 99th percentile observers to account for the effects of illumination and eccentricity on CFF values. Implementation of a versatile and predictive method of flicker perception such as the Farrell equation would be useful for consumers and manufactures alike. Since the ANSI/HFES 100-1988 standard was developed, technology has allowed for very large, high resolution CRTs to become commonplace. The current growth of multimedia applications in business and education is only likely to increase the need for a new, reliable, and easy to use flicker standard. ANSI/HFES 100-1988s flicker testing procedure is no longer suitable for todays displays.
Flicker Perception
34
REFERENCES American National Standard for Human Factors Engineering of Visual Display Terminal Workstations. (1988). Santa Monica, CA: The Human Factors Society, Inc. American National Standard for Human Factors Engineering of Visual Display Terminal Workstations. Unpublished (199x). Santa Monica, CA: The Human Factors Society, Inc. Amir, T., and Ali, M. R. (1989). Critical flicker frequency, personality, and sex of subjects. Perceptual and Motor Skills, 69, 1019-1026. Bauer, D., Bonacker, M., & Cavonius, C. (1983). Frame repetition rate for flicker-free viewing of bright VDU screens. Displays, Jan, 31-33. Brown, John Lott (1965). Flicker and intermittent stimulation. In C. H. Graham (Ed.), Vision and visual perception. (pp. 251-320). New York: Wile...