Perception of Light Environment in University Classrooms Based on Parametric Optical Simulation and Virtual Reality Technology

Abstract

University classrooms, core to higher education, have indoor light environments that directly affect students’ learning efficiency, visual health, and psychological states. This study integrates parametric optical simulation and virtual reality (VR) to explore light environment perception in ordinary university classrooms. Forty college students (18–25 years, ~1:1 gender ratio) participated in real virtual comparative experiments. VR scenarios were optimized via real-time rendering and physical calibration. The results showed no significant differences in subjects’ perception evaluations between environments (p > 0.05), verifying virtual environments as effective experimental carriers. The analysis of eight virtual conditions (varying window-to-wall ratios and lighting methods) revealed that mixed lighting performed best in light perception, spatial perception, and overall evaluation. Light perception had the greatest influence on overall evaluation (0.905), with glare as the core factor (0.68); closure sense contributed most to spatial perception (0.45). Structural equation modeling showed that window-to-wall ratio and lighting power density positively correlated with subjective evaluations. Window-to-wall ratio had a 0.412 direct effect on spatial perception and a 0.84 total mediating effect (67.1% of total effect), exceeding the lighting power density’s 0.57 mediating effect sum. This study confirms mixed lighting and window-to-wall ratio optimization as keys to improving classroom light quality, providing an experimental paradigm and parameter basis for user-perception-oriented design.

1. Introduction

As the primary setting for higher education, the quality of the light environment in university classrooms directly influences students’ learning efficiency, visual health, and psychological well-being [1,2,3,4]. With the continuous development of higher education infrastructure in China, students’ expectations for classroom lighting quality have steadily increased. Research has shown that a high-quality lighting environment can enhance students’ concentration by 15–20% and improve knowledge retention [1]. Moreover, it can significantly reduce visual fatigue caused by prolonged periods of study [2]. However, numerous issues persist in the lighting environment design of university classrooms. For example, the insufficient integration of natural daylight in some classrooms leads to consistently high energy consumption for lighting. In addition, excessive or uneven light distribution is prone to causing glare, which in turn reduces learning efficiency and impairs vision [5]. Moreover, classroom orientation significantly affects the quality of natural lighting. Studies have shown that south-facing classrooms generally receive more adequate daylight than those facing north or west, thereby enhancing students’ visual comfort and learning efficiency [6,7]. With the advancement of the “double carbon” initiative and the widespread adoption of green design principles, balancing comfort and energy efficiency in classroom lighting and daylighting design has become increasingly urgent [8]. Therefore, a thorough investigation of the characteristics of university classroom lighting environments is essential for enhancing the quality of teaching spaces and advancing the sustainable development of educational buildings.

Existing research on lighting environment evaluation tends to be predominantly single-dimensional. An analysis of data from the Web of Science Core Collection over the past five years reveals that most related studies are limited to either objective parameters, such as illuminance and color temperature, or subjective perceptions, such as comfort and emotional responses. For example, Kompier et al. (2017) primarily examined the effects of illuminance and color temperature on visual comfort but did not establish a quantitative correlation with subjective user experience [5]. Similarly, Linares Vargas et al. (2018) investigated emotional responses to spatial environments in virtual settings, yet they did not concurrently analyze objective environmental parameters [9]. A recent review of 200 studies on lighting and ecological psychology found that approximately 75% of the research focuses on single-dimensional variables, primarily due to the challenges associated with integrating multiple factors. Such fragmented research approaches have hindered the systematic understanding of the complex reciprocal mechanisms between environmental metrics and human sensory responses, thereby limiting the advancement of refined lighting design. Nevertheless, ecological psychology has increasingly shifted toward a multi-factor integrated research paradigm [10]. However, the integration of subjective and objective evaluations in the field of architectural lighting environments remains in its exploratory stage.

The innovation of virtual reality (VR) technology has marked a breakthrough in architectural environment simulation. With features such as high-resolution rendering and real-time lighting calculation, VR has greatly enhanced the realism of virtual environments. Carrozzino et al. (2010) first demonstrated that there are no significant differences in the presence between VR and real environments [11]. Strand (2020) further showed that AI-enhanced immersive VR improves color accuracy by approximately 41% compared to conventional media formats [12]. More recent studies have continued to advance this field. For example, Gómez-Tone et al. (2022) validated the reliability of VR in simulating indoor daylighting through luminance matching experiments [13], while Mousavi et al. (2024) reported that real-time shadow rendering in VR can reduce spatial perception errors by 27% [14]. These advancements have helped overcome many of the limitations associated with real-world experiments.

However, current VR-based research on lighting environments still presents three key limitations. First, more than half of the studies have not calibrated photometric parameters—such as luminance and illuminance consistency—between virtual and real settings [15,16]. Second, most experiments focus on a single evaluation metric and lack integrated analyses that combine subjective and objective data [17]. Third, no studies have yet quantified the perceptual pathways through which environmental parameters influence subjective evaluations, representing a critical application gap in translating environmental psychology’s multi-factor theory to the architectural domain [18].

To address these gaps, this study makes the following three key contributions: (1) it proposes a dual-calibration method for photometric parameters (luminance and illuminance) to ensure consistency between virtual and real environments; (2) it integrates subjective perception scales with objective physiological indicators (e.g., pupil dilation) for comprehensive evaluation; and (3) it quantifies the mediating pathways through which environmental parameters influence subjective evaluations using structural equation modeling, thereby bridging the application gap of multi-factor theory in architectural research.

2. Materials and Methods

2.1. Experimental Design

This research focuses on typical university classrooms and student participants, employing a mixed-method approach that combines subjective questionnaire assessments, VR-based virtual simulation, and parametric modeling. Two sets of controlled experiments were conducted in both real and virtual environments. The experimental design is illustrated in Figure 1.

• Experiment (A): Validation of Virtual Environment Effectiveness

A high-fidelity virtual scene was constructed based on the lighting environment parameters (e.g., illuminance levels) of real classrooms. Its accuracy was validated through calibration by comparing measured data with simulation outputs. Participants performed perceptual evaluations in both real and virtual settings under identical lighting conditions and completed a presence questionnaire to assess the realism and immersive quality of the VR environment. This experiment aims to verify whether the virtual environment can effectively substitute for the real environment in lighting perception research, a concept that has been explored in previous studies. For example, Nikookar et al. (2024) found that virtual environments could replicate real-world lighting conditions with high accuracy, offering a viable alternative for studying lighting effects in controlled settings. Similarly, studies by Bellazzi et al. (2022) demonstrated that immersive VR could replace physical environments in assessing visual comfort, offering consistent results comparable to those observed in actual spaces [19,20].

• Experiment (B): Perceptual Evaluation of Lighting Environment Conditions

Building on the validation from Experiment A, this experiment investigates participants’ subjective perceptual differences under eight distinct lighting conditions, including natural daylighting and mixed lighting. By examining the correlations between objective parameters (e.g., window-to-wall ratio, lighting power density) and subjective evaluations, this study aims to inform the optimization of classroom lighting design parameters.

2.2. Research Platform and Related Tools

The construction of virtual lighting environments is crucial for VR-based perceptual research on general classroom lighting. Its key processes cover the following four core steps: spatial modeling, physical lighting simulation, color tone adjustment, and presentation model [21], with the commonly used construction tools for each step detailed in

Table 1. Virtual scene construction path.

Process	Space Construction	Light Environment Simulation	Tone Adjustment	Presentation Method
Function	Construct a space model	Simulate the light environment of real scenes	Adjust image information to enable the virtual scene to convey the information of the real environment better	Technical means such as relevant equipment
Common Software and Equipment	Sketchup	Radiance	Reinhard	2D image static display
	3dmax	Light Tools	iCAM	3D image, video, dynamic display
	Rhino	V-ray	Linear	Immersive VR
	Unity	Real-scene	Enscape

.

This research employs Rhino + Grasshopper as the parametric design platform [22,23,24]. This platform possesses powerful parametric modeling capabilities, enabling the precise construction of the spatial structure of general classrooms to provide an accurate geometric foundation for subsequent lighting environment simulation. Meanwhile, the platform integrates the Radiance engine (LB) and Enscape real-time rendering engine [25,26,27].

The Radiance engine (LB), based on advanced physical optics principles, can accurately simulate complex processes such as light propagation, reflection, and refraction in classroom spaces. It takes into account factors such as the optical properties of different materials and the spectral distribution of light sources, thereby achieving a highly realistic physical lighting environment simulation. Through this engine, lighting environment parameters such as illuminance, luminance, and color temperature at various positions in the classroom can be accurately calculated, providing reliable data support for subsequent research.

The Enscape real-time rendering engine provides an efficient and realistic solution for presenting virtual lighting environments. With its real-time rendering feature, it enables immediate feedback on rendering effects during the design process, facilitating operations such as color tone adjustment for researchers. Moreover, Enscape has a built-in VR interface, allowing for seamless integration with virtual reality devices. In terms of device selection, this study employs the HTC Vive P110 head-mounted virtual reality system as the scene presentation tool. This system boasts high resolution, a wide field of view, and precise tracking capabilities, delivering an immersive virtual lighting experience that enables participants to perceive classroom lighting environments in a lifelike manner [28].

This platform organically integrates the four key steps of spatial modeling, physical lighting simulation, color tone adjustment, and presentation mode, ensuring the accuracy and operability of the entire virtual lighting environment construction process. To further enhance the consistency between the virtual lighting environment and the real scene, researchers used professional illuminance meters to conduct detailed measurements of illuminance and luminance in the actual classroom. By comparing the measurement data of the real scene and the virtual lighting environment, the virtual environment was calibrated to maintain basic consistency in illuminance and luminance with the real scene, with the specific calibration process shown in Figure 2. This calibration method effectively improves the authenticity and reliability of the virtual lighting environment, laying a solid foundation for subsequent experimental research.

2.3. Experimental Questionnaire Overview

2.3.1. Participant Selection

Forty healthy college students (mean age = 20 years; 20 males and 20 females; 20 from liberal arts majors and 20 from science and engineering majors) were recruited as participants. Over 70% had an architectural education background, enabling them to accurately interpret professional terms (e.g., illuminance, lighting environment) in the questionnaire and ensuring the validity and professionalism of subjective evaluation data. All participants signed informed consent forms, had normal or corrected-to-normal vision, and no color blindness or color weakness, thus meeting the visual perception requirements of the experiment.

2.3.2. Presence Questionnaire Design

To assess the simulation accuracy of VR environments, this study adopted the presence evaluation framework proposed by Schubert et al. [29]. This framework was selected because it has been widely validated in VR and human–computer interaction studies and offers a comprehensive approach to measuring presence [30]. The framework deconstructs presence into three core factors, spatial presence, engagement, and realism, which are highly relevant to the immersive experience in VR classroom environments. This makes it an ideal tool for capturing participants’ subjective experience of the VR environment. Additionally, the Immersive Presence Questionnaire (IPQ), developed from this framework, has been validated in numerous studies [31] and provides a reliable means of assessing the quality of immersion. In this experiment, the IPQ included one overall evaluation item and three sub-factor scales, all using 7-point Likert scales (1 = “strongly disagree”, 7 = “strongly agree”). Items ranged from “I felt completely immersed in the virtual environment” to “Objects in the virtual environment looked realistic,” quantifying participants’ immersive experience of the VR classroom scenes(

Table 2. Presence questionnaire design.

Content		Evaluation Indicator	Very	Fairly	Somewhat	Moderate	Somewhat	Fairly	Very	Evaluation Indicator
			1	2	3	4	5	6	7
Spatial Presence	Is the spatial information in the virtual environment consistent with the real space?	Inconsistent								Consistent
Sense of Participation	Are you immersed in the virtual environment?	No participation								With participation
Sense of Realism	Are the relevant components and materials in the virtual light environment realistic?	Unrealistic								Realistic
Overall Evaluation		Dissatisfied								Satisfied

). This choice was made because the framework’s multidimensional approach allows for a more nuanced understanding of how different factors contribute to the immersive experience in educational VR environments [32].

2.3.3. Subjective Perception Questionnaire Design

Guided by environmental psychology theories, which posit that ecological stimuli can elicit emotional responses and reshape spatial cognition, this study constructed a three-dimensional subjective perception evaluation system [33,34,35,36,37,38]. The development of this evaluation model was grounded in established frameworks from environmental psychology and informed by an understanding of how individuals psychologically perceive and respond to spatial environments. Given the complexity of human–environment interactions, the model was designed to assess both objective environmental features (e.g., spatial configuration and lighting) and subjective psychological responses (e.g., emotional feedback, comfort, and spatial cognition) [39].

Spatial Perception Evaluation: To better understand participants’ cognitive responses to classroom spaces, including privacy needs and psychological states, the following five sets of semantic differential scale items were designed: “narrow to spacious,” “enclosed to open,” “boring to interesting,” “public to private,” and “tense to relaxed.” These dimensions were selected based on established models in environmental psychology, which emphasize that spatial layout and environmental characteristics significantly influence individual psychological experiences. For example, previous studies have shown that perceptions of openness and privacy directly affect emotional comfort and stress levels. This evaluation system aims to quantify how classroom morphology, specifically its physical structure and layout, influences spatial perception and psychological well-being [40,41].

Light Perception Evaluation: Two core indicators, brightness perception and glare sensation, were selected to quantify participants’ visual experience of the lighting environment. Questions such as “How bright is the light in this environment?” and “Do the lights cause you discomfort or glare?” were used to assess these perceptions. This section of the evaluation aimed to examine how lighting design influences the overall classroom atmosphere and students’ ability to concentrate, drawing on prior research linking lighting conditions to cognitive performance [42].

Overall Lighting Environment Evaluation: From a comprehensive perspective, incorporating both comfort and satisfaction, items such as “How satisfied are you with the current classroom lighting environment?” were used to assess the overall lighting quality. This section aimed to evaluate not only the technical performance of the lighting system but also the subjective experience of comfort and usability [40]. All evaluation indicators were measured using 7-point Likert scales to ensure data quantifiability and comparability (

Table 3. Subjective perception questionnaire design.

	Evaluation Indicator	Very	Fairly	Somewhat	Moderate	Somewhat	Fairly	Very	Evaluation Indicator
		1	2	3	4	5	6	7
Spatial Perception Evaluation	Narrow								Spacious
	Closed								Open
	Boring								Interesting
	Public								Private
	Nervous								Relaxed
Light Perception Evaluation	Dark (Brightness Sensation)								Bright
	Dazzling (Glare Sensation)								Soothing
Overall Light Environment Evaluation	Dissatisfied								Satisfied

).

Through the above questionnaire design, this study established a comprehensive evaluation system covering both the simulation effect of the VR environment and subjective perceptions of the lighting environment, providing data support for the subsequent analysis of lighting perception differences between virtual and real scenarios.

2.4. Experimental Conditions

2.4.1. Real-World Scenario

For the real-world scenario, a typical classroom at a Beijing university was selected as the test room. Real-scene experiments were conducted under overcast conditions from 9:00–11:00 and 14:00–16:00, during which natural illuminance fluctuations in the north-facing classroom were minimal. Virtual scenarios synchronously simulated the lighting conditions of these time periods to ensure light environment stability. As shown in Figure 3, the room dimensions are 8200 mm (bay) × 8200 mm (depth) × 4500 mm (clear height). The walls and ceiling are finished with gray paint, while the floor is covered with gray diffuse-reflective tiles. The windows measure 2700 mm × 2900 mm with a windowsill height of 900 mm, resulting in a window-to-wall ratio of approximately 60%. The desk height is around 750 mm.

2.4.2. Virtual Scenario

For the virtual scenario, the spatial model references the test room, with identical spatial dimensions and material properties, and the indoor working plane is set at 0.75 m. The reflectance of each surface in the model is set according to the following Architectural Daylighting Design Standard and Munsell values: the walls of the room are made of white diffuse reflection materials with a reflectance ρ of 0.75; the floor is made of gray diffuse reflection materials with a reflectance ρ of 0.30; and the window glass is a transparent material with a transmittance τ of 0.65 [43,44]. According to China’s Architectural Lighting Design Standard, the indoor lighting environment can be divided into two different states, natural daylighting and hybrid lighting [45]. This experiment selected the window-to-wall ratio and lighting power density as variables. The values of artificial lighting power density were calculated by referring to the number of lighting fixtures in actual classrooms, which were 0 W/m² and 9 W/m², respectively, resulting in a total of eight different experimental conditions (

Table 4. 8 Different Light Environment Working Conditions.

Number	Working Condition 1	Working Condition 2	Working Condition 3	Working Condition 4	Working Condition 5	Working Condition 6	Working Condition 7	Working Condition 8
Window-to-Wall Ratio	20%	20%	40%	40%	60%	60%	80%	80%
Lighting Power Density (W/m2)	9	0	9	0	9	0	9	0
Lighting Method	Mixed Lighting	Natural Daylighting	Mixed Lighting	Natural Daylighting	Mixed Lighting	Natural Daylighting	Mixed Lighting	Natural Daylighting

). The virtual condition scenarios are shown in Figure 4.

2.5. Experimental Procedures

First, the consistency between the real and virtual scenarios needs to be verified. The study selected a north-facing classroom under overcast conditions as the experimental subject to avoid interference from direct sunlight. For Experiment (A), since the test classroom was fully consistent with the parameters of Condition 6 (both with a 60% window-to-wall ratio and 0 W/m² lighting power density, i.e., no artificial lighting), Condition 6 was chosen as the virtual scenario. The effect comparison is shown in Figure 5, indicating that the simulated illuminance of the virtual scenario is consistent with that of the real scenario. Before the experiment, an illuminance meter was used to measure the illuminance of the real scenario, and relevant illuminance data of the virtual scenario (Condition 6) were obtained through simulation with Ladybug Tools. As shown in Figure 6, the overall illuminance of the virtual light environment is consistent with that of the real scenario.

The workflow of the experimental design is shown in Figure 7. The experiment recruited 40 college students aged around 20, with 20 males and 20 females, including 20 from liberal arts majors and 20 from science and engineering majors. Participants were randomly assigned to the real scenario and virtual scenario (both under overcast conditions). They were asked to perceive the environment for 2 min in each scenario and complete a perception questionnaire. After four rounds of perception, they filled out a presence questionnaire. To reduce the impact of the first perception, a 1 min rest interval was set between the two rounds of perception.

The testing environment for Experiment (B) had constant temperature and humidity and was free of noise. Forty participants were divided into 20 groups, with two individuals per group experimenting simultaneously, and each participant performed environmental perception for four working conditions. To prevent interference from similar working conditions, the experimental conditions were selected by lottery. After drawing lots, participants filled in their personal information on paper questionnaires, wore VR headsets, and sat quietly at the experimental station. The corresponding working condition was activated on the computer, and the participants conducted 2 min environmental perception in the selected condition. After the perception, the staff asked questions, and the participants scored each evaluation index in sequence according to the questionnaire, they transmitted the score information by gestures, and the staff recorded it on the paper questionnaire. After completion, they rested with their eyes closed for 30 s and switched to the next working condition. The experimental process is shown in Figure 8.

3. Results

3.1. Subsection

Regarding the presence questionnaire, the mean scores of all indicators in the presence questionnaire were not only higher than the value corresponding to a moderate level (i.e., 4 points) but were also higher than the relevant evaluation scores in previous studies. The overall presence evaluation score was 5.31. Among them, the spatial presence score was the highest, reaching 5.39 points, indicating no difference in spatial information between the virtual and real scenarios. Engagement (interactivity, immersiveness) is recognized by the academic community as a crucial influencing factor in creating a virtual environment consistent with the real world. It scored 5.28, and the realism score was 5.19. Overall, the spatial presence score > engagement score > realism score, which is similar to the scoring trend in previous studies. In terms of realism scores, there has been a significant improvement compared to prior studies.

To determine if participants’ perception evaluations were similar in real and virtual environments under the same light conditions, an independent sample t-test was conducted on subjective evaluations from Experiment (A). Results (

Table 5. Results of the t-test for subjective evaluation in different scenarios.

	Scene	Mean Value	Significance (P)	95% Confidence Interval of Difference
				Lower Limit	Upper Limit
Sense of Spaciousness	RE	3.56	0.513	−0.721	0.388
	VE	3.67
Sense of Openness	RE	3.52	0.456	−0.675	0.286
	VE	3.62
Sense of Interest	RE	3.26	0.355	−0.233	0.733
	VE	3.15
Sense of Privacy	RE	3.57	0.227	−0.245	0.856
	VE	3.25
Sense of Relaxation	RE	4.81	0.221	−0.288	0.839
	VE	4.62
Overall Evaluation of Spatial Sensation	RE	3.75	0.405	−0.318	0.762
	VE	3.63
Sense of Brightness	RE	3.75	0.425	−0.722	0.334
	VE	3.76
Sense of Glare	RE	4.42	0.612	−0.705	0.428
	VE	4.45
Light Perception Evaluation	RE	4.25	0.312	−0.752	0.252
	VE	4.32
Overall Light Environment Evaluation	RE	4.39	0.261	−0.769	0.214
	VE	4.41

) showed that all p-values were >0.05, indicating no significant differences in subjective assessment between the two environments. The analysis of presence evaluations and perception differences confirmed that the virtual environment can effectively replace the real one.

3.2. Analysis of Subjective Perception Evaluation Under Different Working Conditions

3.2.1. Overall Lighting Environment Evaluation

The user’s overall evaluation of the indoor environment is the result of comprehensive factors. Establishing a comprehensive evaluation model based on user feedback can provide references for the subsequent regulation of indoor visual comfort. According to the questionnaire results of satisfaction with nine working conditions, multiple regression analysis was conducted on overall lighting environment evaluation, spatial perception, and light perception evaluation (R = 0.808, R² = 0.751), yielding regression coefficients and significance of various factors (

Table 6. Results of multiple regression analysis.

Regression Parameter		Light Perception Evaluation	Spatial Perception Evaluation
Overall Light Environment Evaluation	Regression Coefficient	0.905	0.115
	Significance	0.000	0.068

). According to the regression results, light perception evaluation had the greatest impact on overall lighting environment assessment, with a regression coefficient of 0.905 (p < 0.01). Subjective evaluation results (Figure 9) showed that Condition 8—featuring an 80% window-to-wall ratio (WWR) under natural daylighting—achieved the highest scores in both light perception and overall lighting evaluation. Notably, Conditions 8 and 7 shared an 80% WWR, but Condition 8 (no artificial lighting) outperformed Condition 7 (full lighting), indicating that higher lighting power density does not necessarily correlate with higher scores. This highlights the need to optimize window design and lighting layout based on user subjective feedback.

3.2.2. Spatial Perception Evaluation

Multiple regression analysis was performed on overall spatial perception and its sub-item evaluations (R = 0.920, R² = 0.901), yielding regression coefficients and the significance of each sub-item (

Table 7. Results of multiple regression analysis between spatial perception evaluation and other spatial sensation factors.

Regression Parameter		Sense of Spaciousness	Sense of Closure	Sense of Interest	Sense of Privacy	Sense of Tension
Spatial Perception Evaluation	Regression Coefficient	0.262	0.452	0.181	0.251	0.256
	Significance	0.000	0.000	0.001	0.000	0.000

). The results show that enclosure sensation had the greatest impact on spatial perception evaluation (regression coefficient = 0.452), followed by tension and spaciousness. According to the subjective evaluation results (Figure 10), the comprehensive scores of spatial perceptions for Condition 2 and Condition 4 were poor, with average values both below 4. In terms of sub-item scores for spaciousness, enclosure, and tension, Condition 2 was perceived as narrower, enclosed, and tense compared to other conditions, yielding lower scores. In contrast, Condition 8 achieved the highest scores. In terms of spatial interest ratings, scores were relatively low across all conditions, with Conditions 7 and 8 (80% window-to-wall ratio) achieving the highest scores. For privacy ratings, Conditions 1, 3, and 5 were perceived as more private, yielding higher scores.

3.2.3. Light Perception Evaluation

Multiple regression analysis was conducted on overall light perception and its sub-item evaluations (R = 0.886, R² = 0.802), yielding regression coefficients and the significance of each sub-item (

Table 8. Results of multiple regression analysis between light environment evaluation and other light environment factors.

Regression Parameter		Sense of Brightness	Sense of Glare
Light Perception Evaluation	Regression Coefficient	0.455	0.681
	Significance	0.000	0.000

). According to the regression analysis results, glare sensation had the greatest impact on light perception evaluation, with a regression coefficient of 0.681. This indicates that a uniform and soft indoor light environment is conducive to improving light comfort evaluation. Combined with the subjective evaluation results (Figure 11), all working conditions showed good comfort, with comprehensive scores all exceeding 4. In terms of glare perception evaluation, Condition 1 received the highest score, featuring a softer indoor light environment and the highest light environment comfort. For brightness perception, Condition 2—with only natural lighting and a 20% window-to-wall ratio—was relatively dark indoors, resulting in a lower evaluation score.

3.3. Parametric Simulation Analysis of Different Working Conditions

The simulation results are statistically shown in Figure 12. In terms of average illuminance, all working conditions met the requirements of the national standard [26] and were higher than the specified values. Meanwhile, the indoor average illuminance under mixed lighting was 40% higher than that under natural lighting as a whole. In terms of illuminance uniformity, Condition 1 met the requirements of the national standard, while the other working conditions were relatively low. Furthermore, the indoor illuminance uniformity was highest (0.78) under the smallest window-to-wall ratio (WWR) with artificial lighting, followed by mixed lighting, while natural lighting had the lowest uniformity (mean = 0.44). As shown in

Table 9. Results of Pearson correlation analysis between working condition parameters and simulation results.

Variable	Lighting Power Density	Window-to-Wall Ratio
Average Illuminance	0.515 **	0.699 **
Illuminance Uniformity	0.813 **	0.302 *

Note: ** Significantly correlated at the 0.01 level (two-tailed). * Significantly correlated at the 0.05 level (two-tailed).

, Pearson’s correlations revealed the following significant associations (all p < 0.001): average illuminance with LPD (r = 0.515) and WWR (r = 0.699); illuminance uniformity with artificial lighting (r = 0.813) and WWR (r = −0.302). This highlights the trade-off between WWR-driven illuminance and uniformity, which is that optimal design requires balancing natural daylight.

3.4. The Relationship Between Objective Conditions and Subjective Evaluations

Light perception varied significantly across different working conditions. Condition parameters (window-to-wall ratio, lighting power density) influenced indoor illuminance distribution, thereby affecting participants’ subjective perceptions (Figure 13). The subjective evaluations by users were determined by multiple factors, with influencing factors including not only working condition parameters such as window-to-wall ratio, lighting mode, and lighting power density, but also objective physical indices like average illuminance and illuminance uniformity. Since objective physical indexes are affected by working conditions—serving as mediating variables—and there is an interactive effect between illuminance uniformity and average illuminance within these mediating variables, it is difficult to analyze the relationship between working conditions (window-to-wall ratio, lighting power density) and subjective evaluations using traditional multiple regression analysis or factor analysis methods, yielding unconvincing conclusions. This study established a structural equation model (SEM) to analyze the relationships among multiple independent variables, mediating variables, and dependent variables. Path coefficient analysis was used to identify key influencing factors, while the model also revealed the explanatory power of each latent variable and the magnitude of interactions between variables.

Figure 14 presents the path analysis results. In terms of direct effects, window-to-wall ratio (WWR) showed a significant positive direct correlation with spatial perception evaluation (path coefficient = 0.412, p < 0.01). The mediational analysis (

Table 10. Result of mediating effect.

	Mediating Model	Mediating Effect Value
1	Window-to-Wall Ratio—Sense of Spaciousness—Spatial Perception Evaluation	0.123 ×0.255 = 0.03
2	Window-to-Wall Ratio—Sense of Closure—Spatial Perception Evaluation	0.415 × 0.385 = 0.16
3	Window-to-Wall Ratio—Sense of Interest—Spatial Perception Evaluation	0.321 × 0.181 = 0.06
4	Window-to-Wall Ratio—Average Illuminance—Spatial Perception Evaluation	0.745 × 0.455 = 0.34
5	Window-to-Wall Ratio—Average Illuminance—Sense of Spaciousness—Spatial Perception Evaluation	0.745 × 0.461 × 0.255 = 0.09
6	Window-to-Wall Ratio—Average Illuminance—Sense of Closure—Spatial Perception Evaluation	0.745 × 0.488 × 0.385 = 0.14
7	Window-to-Wall Ratio—Average Illuminance—Sense of Interest—Spatial Perception Evaluation	0.745 × 0.382 ×0.181 = 0.05
8	Window-to-Wall Ratio—Illuminance Uniformity—Sense of Privacy—Spatial Perception Evaluation	−0.752 × 0.355 × 0.281 = −0.08
9	Window-to-Wall Ratio—Average Illuminance—Illuminance Uniformity—Sense of Privacy—Spatial Perception Evaluation	0.745 × 0.655 × 0.355 × 0.281 = 0.05
10	Window-to-Wall Ratio—Sense of Glare—Light Perception Evaluation	0.310 × 0.64 = 0.20
11	Window-to-Wall Ratio—Average Illuminance—Sense of Brightness—Light Perception Evaluation	0.745 × 0.323 × 0.511 = 0.12
12	Lighting Power Density—Average Illuminance—Spatial Perception Evaluation	0.581 × 0.455 = 0.26
13	Lighting Power Density—Average Illuminance—Sense of Spaciousness—Spatial Perception Evaluation	0.581 × 0.461 × 0.255 = 0.07
14	Lighting Power Density—Average Illuminance—Sense of Closure—Spatial Perception Evaluation	0.581 × 0.488 × 0.385 = 0.11
15	Lighting Power Density—Average Illuminance—Sense of Interest—Spatial Perception Evaluation	0.581 × 0.382 × 0.181 = 0.04
16	Lighting Power Density—Illuminance Uniformity—Sense of Privacy—Spatial Perception Evaluation	0.461 × 0.355 × 0.281 = 0.05
17	Lighting Power Density—Average Illuminance—Illuminance Uniformity—Sense of Privacy—Spatial Perception Evaluation	0.581 × 0.655 × 0.355 × 0.281 = 0.04
18	Lighting Power Density—Sense of Brightness—Light Perception Evaluation	0.315 × 0.511 = 0.16
19	Lighting Power Density—Average Illuminance—Sense of Brightness—Light Perception Evaluation	0.581 × 0.323 × 0.511 = 0.10

Note: The gray-marked parts are mediating models with mediation effect values greater than 0.1.

) revealed nine mediational models (Models 1–9) for spatial perception influenced by WWR, with a total mediational effect of 0.84. Among these, Model 4 (“WWR → average illuminance → spatial perception evaluation”) exhibited the strongest effect, with a value of 0.34. A total of six mediational models (Models 12–17) were identified for the influence of lighting power density (LPD) on spatial perception, with a cumulative mediational effect of 0.57. Among these, Model 12 (“LPD → average illuminance → spatial perception evaluation”) exhibited the strongest effect, with a mediational value of 0.26. In terms of light perception, two mediational models (Models 10–11) were identified for the influence of window-to-wall ratio (WWR), with a cumulative mediational effect of 0.32. Model 10 (“WWR → glare sensation → light perception evaluation”) had the strongest effect (0.20). For lighting power density (LPD), two mediational models (Models 18–19) yielded a total effect of 0.26, with Model 12 (“LPD → brightness perception → light perception evaluation”) demonstrating the highest effect (0.16).

4. Discussion

The integration of parametric optical simulation and virtual reality (VR) technology in this study provides a novel methodological framework for exploring the perception of classroom light environments. The validation of the VR experimental paradigm, with no significant differences in perceptual evaluations between virtual and real scenarios (p > 0.05), aligns with the findings of previous studies that highlight the potential of VR in replicating physical environments for perceptual research. This consistency not only confirms the reliability of our experimental design but also addresses the long-standing challenges of high costs and poor repeatability in traditional physical experiments, offering a cost-effective and controllable alternative for future investigations.

The superior performance of mixed lighting in comprehensive evaluations echoes the existing understanding that natural daylight positively impacts learning efficiency, while further emphasizing the synergistic effect of combining natural and artificial light. The ability of mixed lighting to mitigate glare issues from natural lighting and compensate for the deficiencies of artificial lighting in spatial atmosphere creation provides empirical support for its application in university classrooms. This finding extends beyond previous research by quantifying the advantages of mixed lighting across multiple perceptual dimensions, including light perception, spatial perception, and overall evaluation.

The results of the perception weight analysis shed light on the key factors influencing classroom light environment satisfaction. The dominant role of light perception (β = 0.905, p < 0.01) underscores the importance of precise control over optical parameters such as illuminance, color temperature, and glare in design practices. Additionally, the high weights of glare sensation (β = 0.681, p < 0.01) and closure sense (β = 0.452, p < 0.01) highlight the need for the careful consideration of both visual comfort and spatial layout in creating optimal learning environments. These findings provide a clear hierarchy of priorities for designers, guiding them to focus on the most impactful aspects of light environment design.

The structural equation model analysis reveals the complex pathways through which environmental parameters influence subjective evaluations. The stronger indirect effect of the window-to-wall ratio (WWR) compared to lighting power density (LPD) challenges the conventional emphasis on artificial lighting parameters in design. Instead, it emphasizes the significance of architectural form and daylighting strategies in the early design stage. The total mediating effect of WWR (0.84) through nine pathways, accounting for 67.1% of the total effect, indicates that optimizing WWR can have far-reaching impacts on spatial perception through multiple intermediate variables. This finding provides a theoretical basis for prioritizing daylighting design in classroom planning, moving beyond the sole focus on increasing lighting power density.

However, this study is not without limitations. The reliance on a single north-facing classroom with a standard layout may restrict the generalizability of our conclusions to classrooms with different orientations, non-rectangular layouts, or specialized functions. The exclusive use of overcast conditions for real-world experiments and the limited coverage of extreme weather and diurnal variations in virtual simulations also limit the applicability of our results to dynamic natural light conditions. Furthermore, the sample consisting primarily of 18–25-year-old college students may not represent the perceptual preferences of other age groups, highlighting the need for more diverse participant pools in future studies.

5. Conclusions

This study investigates the perception of light environments in ordinary university classrooms using parametric optical simulation and VR technology, yielding the following key conclusions:

The validated VR experimental paradigm demonstrates advantages of high repeatability, precise variable control, and low cost, providing an efficient and reliable technical approach for classroom light environment research. Its ability to closely replicate real scenarios establishes it as a viable alternative to traditional physical experiments.

Mixed lighting emerges as the optimal lighting strategy for university classrooms, outperforming natural and artificial lighting in light perception, spatial perception, and overall evaluations. It effectively balances glare control and spatial atmosphere creation, validating its practical value in classroom design.

Perception weight analysis identifies light perception as the core determinant of overall light environment satisfaction, with glare sensation being the key factor within this dimension. In spatial perception, closure sense contributes the most, emphasizing the need for targeted design interventions.

Structural equation modeling reveals that WWR and LPD have significant positive correlations with subjective evaluations, with WWR exerting a stronger total influence through mediating pathways. This highlights the priority of optimizing daylighting strategies and architectural form in early design stages.

This research provides an experimental paradigm and parameter optimization basis for user-perception-oriented classroom light environment design. Future work should explore multi-sensor fusion to enhance VR realism, expand experimental scenarios and sample diversity, and incorporate the long-term monitoring of learning behavior to deepen the understanding of light environment impacts.

This study has the following limitations that warrant acknowledgment:

(1)

Spatial and environmental constraints: The real-world experiments were conducted in a single fixed north-facing classroom with a standard layout, which may limit the generalizability of conclusions to classrooms with diverse orientations (e.g., south-facing), non-rectangular layouts, or specific functional attributes (e.g., lecture halls vs. seminar rooms).

(2)

Weather condition limitations: The real scenarios were tested under overcast conditions only, and virtual simulations did not fully cover extreme weather or diurnal light variations, which may affect the robustness of findings across dynamic natural light conditions.

(3)

Sample scope restrictions: Participants were primarily aged 18–25 (college students), and the virtual working conditions focused on typical lighting parameters rather than edge cases. This limits the applicability of results to broader age groups or specialized classroom designs.

Future research will address these limitations by expanding experimental sites to include multi-oriented and multi-layout classrooms, integrating dynamic simulations of extreme weather and diurnal light changes and recruiting participants across broader age ranges to enhance the generalizability of conclusions. Additionally, the long-term monitoring of learning behavior and physiological indicators will be combined to deepen the understanding of light environment impacts on learning efficiency and well-being.