Enhancing legibility in educational settings: Optimizing projection illuminance for varied indoor ambient illuminance

Abstract

Projectors are common display devices in the educational setting. However, dim projector lightbulbs or well-lit classrooms may cause blurriness in the projected image. To determine the optimal projection light under different ambient light conditions, the conjoint effects of projection illuminance and ambient illuminance on the legibility of projection images indoors were investigated. Participants (N = 96) were randomly assigned to one of six indoor ambient light conditions (0, 40, 80, 120, 160, and 200 lx) and performed a visual search task under several projection illuminance conditions (200, 250, 300, 350, and 400 lx). The accuracy and correct response time on the task were collected to evaluate the participants’ visual performance to represent legibility. The optimal projection illuminance (high visual accuracy and fast reaction) was 400 lx (944 ANSI lumen) under all ambient light conditions. To avoid low legibility (accuracy<0.6) and maintain acceptable legibility (accuracy>0.7), the projection illuminance should be increased as the indoor ambient light increases.

1. Introduction

In recent years, projection technology has penetrated modern school education, considerably altering the way people teach and learn [1]. Projectors can display various materials (e.g., texts, images, videos, or animations) efficiently and conveniently, encouraging lively, flexible, and interactive teaching in the classroom [[1], [2], [3], [4], [5], [6], [7], [8]]. However, equipping classrooms with projectors also poses some problems. Projectors use specially designed lightbulbs to project the image from the device onto a screen, and the screen reflects the light. When the bulb is not capable of producing enough light to overcome all the ambient light, blurriness or fuzziness occurs in the image. Over time, poor display quality may increase visual fatigue, decrease task performance, and even result in an upsurge in various eye issues [9,10].

Previous studies have shown that self-luminous screen light and indoor ambient light greatly impact visual performance and visual comfort in electronic visual display (EVD) reading [[11], [12], [13], [14]]. The effect of screen brightness on visibility has been demonstrated for various display devices [[15], [16], [17]]. Brighter display screens are reported to be more legible, leading to better visual performance. However, higher levels of screen light also have negative results, such as causing fatigue and glares [15,18]. Proper indoor ambient lighting also plays a key role in elevating EVD reading efficiency. Early studies found that increasing ambient illuminance could optimize viewers' visual performance [11,[19], [20], [21], [22], [23], [24], [25]]. However, further studies expanded the upper limit of ambient light and found that high illuminance intensity results in more surface reflection [26], visual fatigue [14,27], and worse visual performance [[28], [29], [30]].

In addition, numerous studies have found that the optimal display brightness varies with ambient illuminance. Allan et al. [31] reported that viewers preferred a brighter screen as the ambient light increased when interacting with a high dynamic range (HDR) TV. Satisfaction studies with a liquid-crystal display (LCD) screen yielded similar results [32]. Kim et al. [33] applied alternative forced choices (whether the screen was too bright, too dark, or appropriate) in the subjective evaluation of LCD screens and found that the appropriate screen luminance varied with illuminance conditions. Regarding reading performance on mobile phones, Li et al. [34] proposed optimum screen brightness under different ambient light conditions based on viewers' visual search performances. In an experiment focusing on tablets, Yu and Akita [35] found that viewers’ visual fatigue increased as the ambient-tablet PC luminance ratio increased.

In summary, previous studies have shown that the optimal screen light of various display devices increases with ambient illuminance. However, unlike the images of EVD devices, projected images are not produced by emitted light but by reflected light. The long paths before entering viewers’ eyes may cause more interference from ambient light. However, how to choose the projection illuminance under different indoor ambient illuminances to achieve the optimum legibility is poorly understood. Thus, the main aim of the study is to find the optimal projected illuminance at different ambient illuminances to enhance legibility.

In addition, the scope of the study is limited to educational projects and classroom environment. Legibility, which is measured by a visual search task, is the only metric. The other issues, such as accurate color reproduction or the energy efficiency of the hardware device, are beyond the scope.

2. Method

Existing research [36] involving human vision was limited by the method used to measure the optical parameter of projectors. Specifically, the luminance of the bulb cannot directly represent the luminous flux on the projector screen. The ANSI lumens method is more suitable; this method was proposed by the American National Standards Institute (ANSI) and is a standardized method widely used to measure projector luminous flux [37,38]. In the method, the illuminance of the projectors' screen is measured with the nine-point method, as described below, and the luminous flux was calculated by the following equation:

$${\mathrm{\Phi }}_{\mathrm{v}}={\mathrm{E}}_{\mathrm{v}}\times \mathrm{A}$$

where luminous flux Φ_V in ANSI lumens is equal to the illuminance E_V in lux (lx) multiplied by surface area A in square meters (m²). Notably, in this method, the unit of luminous flux is not the lumen but the ANSI lumen. In terms of brightness, the light of one ANSI lumen is dimmer than the light of one lumen.

This study used the ANSI lumens method to obtain the luminous flux provided by the projector. The luminous flux in ANSI lumens can be calculated using the above formula (in this study, A = 2.36 m²). However, considering that the actual measurement was illuminance, we deemed it more appropriate to utilize the illuminance value as a representation of the light reflected by the screen.

2.1. Pilot survey

A pilot survey was performed before the formal experiment. The projection illuminance and the indoor ambient illuminance in the actual education environment were surveyed. Note that all illuminance is vertical. The main aim of the survey was to confirm the lighting conditions in a real educational setting and select the experimental light conditions based on the results.

2.1.1. Projection illuminance

20 education projectors were randomly selected from the small and mid-sized classrooms of our campus and their working illuminance was measured according to the ANSI lumens measurement method. A 100% white image were first displayed on the entire projection field and then the projection illuminance at 9 fixed points in the field were measured with a lux meter (see Fig. 1). The lux meter is also used in subsequent studies, and its parameters will be reported specifically in the Apparatus section. The average of all 9 points represents the mean illuminance of the projection system. Fig. 2 shows the projection illuminance of the 20 education projectors we measured. The projection illuminance ranged from 120 to 400 lx (M = 260.35 lx). It was found that when the projection illuminance was lower than 200 lx, the images projected were too blurry to allow the content of the images to be recognized. Although the blurring of the images may be related to the type of projector or bulb aging, only the light from the projector output was considered in this survey. Based on the survey results, five light levels of 200, 250, 300, 350, and 400 lx were selected as experimental levels.

Fig. 1.

Schematic of ANSI nine-point measurement.

Fig. 2.

Projection illuminance of 20 education projectors.

2.1.2. Indoor ambient illuminance

Ambient light in classrooms, including both daylight and artificial light, can vary with time, weather, and classroom layout. Therefore, the ambient light levels were measured in various daylight and artificial light conditions. Specifically, the measurement was taken under different weather conditions (10 sunny days and 10 cloudy or rainy days) and at different times (8:00 a.m., 10:00 a.m., 12:00 p.m., 2:00 p.m., 4:00 p.m., and 8:00 p.m.) in June.

For the grid of the illuminance measurement, a method similar to the nine-point measurement mentioned above and the same lux meter were used. Due to our concern about the effect of ambient illuminance on the projection field, we used the nine fixed points to measure the vertical illuminance of the projection field. The average of all 9 points represents the mean illuminance at one measurement time.

At night, without artificial light, the ambient illuminance of the projection area was 0 lx. Once the light was turned on, the ambient illuminance increased to 60–70 lx. In the daytime, without artificial light, the ambient illuminance was between 15 and 50 lx on cloudy and rainy days and 50–200 lx on sunny days. Therefore, six indoor ambient light levels of 0, 40, 80, 120, 160, and 200 lx were selected as experimental levels.

2.2. Participants

The sample size was determined using G*Power 3.1.9.7 [39]. A moderate effect size of f = 0.25 and a significance level of α = 0.05 were employed in the analysis. The result of the analysis showed a power of 1-β = 0.95, and the sample size of participants was 54. To equalize each experimental condition, 96 college students with normal or corrected-to-normal vision (48 females; mean age 21.2 ± 1.4; range 18–24) were recruited by online posters. Before the experiment, they all read and signed the informed consent form. They were all paid 15￥ after the experiment. They were naive to the aim of the study and had not taken part in a similar experiment before.

2.3. Design

A 5 × 6 mixed design was used. Projection Illuminance was the within-subjects variable with five levels (200, 250, 300, 350 and 400 lx). The indoor ambient illuminance was the between-subjects variable with six levels (0, 40, 80, 120, 160, and 200 lx).

Specifically, the projection illuminance was determined by measuring the light from the projector output on the projection area, and the indoor ambient illuminance was determined by measuring the light from two LED lamps, which was used to simulate indoor lighting on the projection area. In order to keep consist to our pilot study, the illuminance of projection and indoor ambient was vertical and we also used ANSI nine-point measurement to measure the illuminance. The measurement diagram in was shown in Fig. 3. In addition, the Spectral Power Distribution (SPD) was measured for each light condition with the spectroradiometer (PJG3 produced by Torchbearer) and the specific figures of SPD were in the supplementary material.

Fig. 3.

The example of illuminance measurement in formal experiment. Left is about indoor ambient illuminance and right is about projection illuminance.

Ninety-six participants were randomly assigned to each ambient illuminance condition and completed a visual search task under different projection lighting conditions. Dependent variables were the correct response time (CRT) and task accuracy (ACC) of the target search task, which represents legibility.

Notably, the raw ACC data consisted of only 0 and 1 (0 indicates an incorrect reaction, and 1 indicates a correct reaction), and the CRT is only the response time of accurately completed trials. The mean CRT and ACC values for each participant in each experimental condition were calculated separately, so each participant had a CRT value and an ACC value in every experimental condition.

2.4. Apparatus

A Lenovo laptop (1024 × 768-pixel resolution) and a connected Sharp projector (XG-MX460a) were used to display the experimental materials programmed with E-prime software (1024 × 768-pixel resolution). The images were all projected onto a project screen (projection area 1.76 m × 1.34 m) that was installed on the wall. The project screen was all white, and its reflectance was approximately 1. The contrast of the projection is 4500:1.

A keyboard to the laptop (Lenovo KB1021) was connected to the computer to record the reaction data.

For each level of ambient illuminance, the indoor lighting was simulated by two LED lamps (LFV-Q30w produced by Lifei Photographic Equipment). The position and illumination intensity of the lamps could be adjusted. To ensure the consistency of the illumination level, illuminance measurements were taken while each participant performing the experiment.

The illuminance was measured with a lux meter. The lux meter (YF2006) was manufactured by EVERFINE. The illuminance measuring range of the device could be adjusted freely, and there were four ranges in total (0.2 lx–200 lx/2 klx/20 klx/200 klx). The relative error of measurement was ±4%, which was sufficient for our needs.

2.5. Stimuli

Each image in the visual search task consisted of a 5 × 5 array of two standard optotypes, Landolt Cs and Tumbling Es (see Fig. 4). The target stimuli were black Landolt Cs, the gap of which was placed in different positions: up, down, left, right, and 45° positions in between.

Fig. 4.

An example of the image used in the search task.

The distractor stimuli were dark gray Tumbling Es, which can face in four directions: up, down, left, and right. The Landolt Cs were superimposed on the Tumbling Es to increase the identification difficulty and prevent a ceiling effect. Each symbol was oriented in a random direction and designed a total of 17 such images, two for the practice and 15 for the formal experiment.

The size of the projection image was consistent with that of the projection area, 19.96 × 15.26° (1.76 × 1.34 m). The size of a single stimulus was 1.49° (0.13 m), and the width of the gap in Landolt Cs was 0.07°. The visual angle of the objects was approximately the size of its image on the retina, and the value depended on the distance between the object and the participant, which was 5 m.

2.6. Procedure

The experiment was conducted from 9am to 6pm in a dark laboratory. Two LED lamps were set out in the lab to reach a specific ambient illuminance level (40, 80, 120, 160, or 200 lx) on the projection area by adjusting their positions. The condition of no ambient light (0 lx) was achieved by drawing the curtains and turning off the light. Then, the brightness of the projector bulb was adjusted to reach a specific projected illuminance level (200, 250, 300, 350, or 400 lx) on the screen.

The participants signed a consent form before the experiment. We then explained the instructions to the participants and ensured that they fully understood them. The chair the participants sat on was 1.2 m high and located on the normal line extending back from the center of the middle screen. The chair was fixed 5 m from the screen, simulating the observation point of the audience sitting at the back of the classroom (see Fig. 5).

Fig. 5.

Diagram of the experimental setup.

A revised precueing paradigm [40] similar to that used in our previous study [41] was applied. This task was suitable for measuring the legibility of the visual interface (see Fig. 6). Each trial started with a central fixation point for 500 ms. Then, an arrow cue was displayed for 1000 ms. The arrow pointed randomly in one of eight directions (left, right, bottom, top, and the 45° positions in between). Afterward, the image with a 5 × 5 array of Landolt Cs superimposed on Tumbling Es was displayed. The participants were asked to search for and count the number of Landolt Cs that faced the same direction as the previous arrow as soon as possible within 10 s. Once they completed the search, the participants were asked to press the space bar immediately and entered their answer by pressing the appropriate number key while a blank screen was displayed.

Fig. 6.

Procedure schematic of each trial.

Each light condition included 24 trials, consisting of 3 randomly selected images displayed 8 times paired with eight direction cues in random order. After completing each condition, the participants rested for about 1 min. Then, the task continued in another projection light condition. The order of the conditions followed a Latin-square design to counterbalance the sequential effects. It took each participant approximately 30 min to complete all five projection illuminance conditions. Before the formal task, the participants completed 16 practice trials with the 2 additional images in unmanipulated light conditions.

The simple experiment scene is shown in Fig. 7.

Fig. 7.

Sample diagram of the experiment scene.

3. Results

3.1. Accuracy of the visual search task

Fig. 8 shows the average accuracy (ACC) of the participants across light conditions. A 5 (projection illuminance) × 6 (ambient illuminance) repeated-measures ANOVA showed that the main effect of projection illuminance was significant, F_{(4, 360)} = 141.475, p < 0.001, η_p² = 0.611, but the main effect of ambient illuminance was not, F_{(5, 90)} = 1.786, p = 0.124, η_p² = 0.090. Critically, the interaction effect of projection illuminance and ambient illuminance was significant, F_{(20, 360)} = 2.185, p = 0.003, η_p² = 0.108. Further simple effect analysis was conducted with Bonferroni and results were shown in Fig. 9. When the ambient illuminance ranged from 0 to 120 lx, the participants' task accuracy was highest under 350 lx projection illuminance. When the ambient illuminance was 160 or 200 lx, the participants’ task accuracy was highest under 400 lx projection illuminance.

Fig. 8.

Mean ACC under different projection illuminance and ambient illuminance conditions. Error bars represent ±1 SE.

Fig. 9.

Mean ACC for different projection illuminances depending on ambient illuminances of (a) 0, (b) 40, (c) 80, (d) 120, (e) 160, and (f) 200 lx. Error bars represent ±1 SE. Different letters A, B, C, and D indicate significant differences between different projection illuminance conditions at p < 0.05. The values with the same letter are statistically identical.

3.2. Correct response time

The extreme data (0.41% of the total) that exceeded three standard deviations from the mean correct response time (CRT) was eliminated before conducting further analysis. Fig. 10 shows the average CRT of the participants over different light conditions. A 5 (projection illuminance) × 6 (ambient illuminance) repeated-measures ANOVA showed that the main effect of projection illuminance was significant, F_{(4, 180)} = 4.106, p = 0.003, η_p² = 0.084, but the main effect of ambient illuminance (F_{(5, 90)} = 2.953, p = 0.062, η_p² = 0.116) and the interaction between the two factors (F_{(20, 360)} = 0.490, p = 0.970, η_p² = 0.021) were not significant. To further explore the difference among projection conditions, Bonferroni's post hoc analysis was employed. The results revealed that the 400lx projection illuminance had a significantly shorter CRT than the 300 lx (p < 0.05) and 350 lx (p < 0.01) projection illuminances.

Fig. 10.

Mean CRT under different projection illuminance and ambient illuminance conditions.

3.3. Linear interpolation for a rough recommendation

To provide a rough recommendation for educators, the linear interpolation method was used to detect the minimum projection illuminance under varied indoor illuminance. The abovementioned results reveal that ACC was more sensitive in detecting changes in light conditions than CRT. Thus, ACC was selected as a representative indicator. For a rough recommendation, 0.6 was selected as the lower threshold. On the one hand, 0.6 represented the 25th percentile of all 30 experimental conditions, indicating relatively low legibility. On the other hand, 0.6 is often identified as a passing standard in most of tests in China. As can been seen in Fig. 7, the maximum ACC just reached to 0.8, indicating the certain difficulty level of our experiment task. Thus, 0.7 was selected as the criterion of acceptable legibility for it was the Median ACC of all the 30 experimental conditions.

Fig. 11 illustrates our rough recommendation. The area below the blue line (ACC = 0.6) indicates low legibility that needs to be avoided, and the area above the green line (ACC = 0.7) indicates high legibility that needs to be maintained. In addition, the minimum projection illuminance required to reach a certain ACC increases with increasing indoor ambient illuminance. Notably, the standard (blue line and green line) is only a rough estimate of legibility for practical educational application, not a precise prediction model. As a result, our recommendation can be a reference instead of a certain criterion.

Fig. 11.

Minimum projection illuminance required when ACC = 0.6 or 0.7 under varied indoor illuminance.

4. Discussion

The present study aimed to investigate the conjoint impact of ambient light and projection light on visual performance in projector displays. The results showed that participants’ performance varied across projection light conditions. High projection illuminance led to higher search accuracy and shorter response times than low projection illuminance. When the projection illuminance reached 400 lx, the legibility of the projection image was the highest (high accuracy and short reaction time) regardless of the indoor ambient illuminance.

The result was in line with many studies on other screen display devices [31,33,34,[42], [43], [44]] but was not consistent with Liu's [17] finding that projection illuminance was inversely proportional to screen visibility. Notably, Liu manipulated the luminous flux on the screen surface simply by adjusting the ambient light. Thus, the deterioration of the projection's visibility may be due to the increasing ambient illuminance. Furthermore, other EVD display studies have found that exposure to particularly strong screen light results in decreased performance and increased discomfort [35,[45], [46], [47]].

In contrast to the CRT results, the ACC results indicated that the projection light that provided the highest accuracy depended on the ambient light level. As the ambient light increased, higher projection illuminance was required to reach equivalent task accuracy. The result was consistent with that of Li [34], who found a similar tendency in mobile phone screens. However, the suggested projection illuminance in the present study was higher than those in Liu [48], who suggested projected illuminances of 52.88, 186.24, 223.20, or 283.57 lx when the ambient illuminance was 0, 120, 160, or 200 lx, respectively. This incongruence may result from the difference in experimental design. First, Liu [17] had the participants sit 3 m away from the projection screen, while in the present study, the reading distance was 5 m, which might require a higher projection illuminance. Second, the conclusion of Liu [17] was based on subjective evaluation results, different from our objective behavioral performance. Future studies should further examine and compare viewers’ subjective satisfaction, physiological comfort, and objective behaviors at higher levels of projected light.

5. Limitations

The formal experiment utilized only a single projector in an indoor teaching environment, while different types of projectors may vary in their projection size, distance, position, and lumen. For example, previous studies have shown that screen size affects viewer visual performance for different EVD devices [[49], [50], [51], [52], [53]], which may result in variations in image legibility. Therefore, the single projector used in our study may have limited the generalizability of our results.

In addition, bulb aging and the light reflected in the experimental environment may also have affected our results. Projector bulbs have a limited service life [7], and aging can cause instability in the light source [54]. Thus, projector aging could affect the legibility of the projection image. However, since the formal experiment used only a single projector, it did not allow us to verify this effect. The light reflectance of the experimental environment may also have an effect. Ideally, the light from the projector output is determined by the light itself and the reflectivity of the projection area. Nevertheless, in the real world, the problem is more complex, and reflected light from the surrounding environment could affect the projection quality [55].

Even if our experimental setting was of great ecologically valid, stimulating the indoor ambient light condition with two LED lamps was different from the real scenario. This may limit the generalizability of our results.

Moreover, our study focused on the legibility of projection in educational setting, however, other issues closely associated with eye symptoms and subjective feelings, such as visual fatigue [14,27], visual comfortable [56] and permissible viewing times [8] during projection viewing, were not discussed. Our future studies intend to investigate these issues to improve the users experience of projection.

In addition to the above shortcomings, from the perspective of energy conservation, there may be a relationship between light control and energy consumption [57], which seems to imply that the results of our study have the potential to benefit energy conservation. However, energy conservation is beyond the scope of this study, and the possibility needs to be addressed in future studies.

6. Conclusions

In summary, our results obtained the optimal optical parameters of a projector under indoor ambient light conditions based on the data of a legibility task. Given that the ANSI lumen is prevailing in the industry, the optical projection illuminance (400 lx) was transformed to the ANSI unit (944 ANSI lumen) to guide selection. A linear interpolation method was used to obtain the projection illuminance standard, which could help users avoid low legibility and maintain high legibility.

We expected the results to benefit the educators and make education more effective.

Ethics declarations

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Zhejiang Sci-Tech University (202103E001).

Funding

This work was supported by the Foundation of Key Laboratory of Human Factors Engineering [Grant No. 6142222210103, 6142222210301]; Space Medical Experiment Project of China Manned Space Program [Grant No. HYZHXM3001]; Natural Science Foundation of Zhejiang Province [Grant No. LQ19C090008]; Foundation strengthening plan technical field fund [Grant No. 2021-JCJQ-JJ-1308].

Data availability statement

Data will be made available on request.

CRediT authorship contribution statement

Zhengyin Gu: Writing – review & editing, Visualization, Validation, Methodology, Investigation, Formal analysis, Data curation. Xinle Bao: Writing – review & editing, Visualization, Software. Ying Zhu: Writing – original draft, Visualization. Saiwei Song: Investigation, Formal analysis, Data curation. Wei Gao: Supervision, Formal analysis. Chunyan Kang: Validation, Supervision. Qijun Wang: Software, Project administration, Funding acquisition, Data curation. Duming Wang: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Conceptualization. Yu Tian: Supervision, Resources, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: