Virtual reality in lighting research: Comparing physical and virtual lighting environments

Abstract

In the study of lighting, as the construction of a physical test room is costly and time-consuming, researchers have been actively looking for alternative media to present physical environments. Virtual reality, photo and video are the most commonly used approaches in the lighting community, and they have all been used by researchers around the world. Most such studies have been conducted without discussing what gives the subjects a better sense of realism, presence, etc., and which type of media is closer to the ideal, the physical lighting environment. In this paper, we aim to select the optimal alternative media that can present physical lighting environments. We compare a human’s subjective feeling towards a physical lighting environment and three alternative reproduction technologies, namely, virtual reality reproduction, video reproduction and photographic reproduction. We also discuss the feasibility of using virtual reality in representing lighting environments. The selection of the most optimal media is based on the perceptual attributes of lighted space, and the findings are only related to these criteria. The main results of this study are the following:

(a) The order of the overall presentation-ability of the media is physical space > virtual reality reproductions > video reproductions > photo reproductions.

(b) In terms of subjective rating, virtual reality lighting environments are rated closest to the physical lighting environments, and the order of the approximate coefficient of the media is physical space (1) > VR reproductions (0.886) > video reproductions (0.752) > photo reproductions (0.679).

(c) Virtual reality can present lighting attributes of open/close, diffuse/glaring, bright/dim and noisy/quiet consistent with the physical environment.

(d) Human subjects are most satisfied with VR reproductions.

1. Introduction

Light (and lighting) is very important to human beings. Researchers found long ago that light can impact the vision, mood, circadian rhythm, daily behaviour and health of humans.^1,2 While more research is needed to identify the optimal lighting parameters for different applications, massive repeated experiments are also needed. Some authors^3–5 believe that the use of a full-scale, mock-up test room is the most precise and reliable method to evaluate lighted environments. However, in most cases, setting up a mock-up test room is expensive.

For many years, researchers have been actively looking for alternative media to present physical environments. Various studies have suggested that photos^6–9 and videos¹⁰ are promising media for investigating subjective impressions of space and light. Moscoso et al.¹¹ compared photos and physically based renderings on several perceptual lighting attributes and showed that the photos convey statistically the same lighting perception as in a real-world scenario. Schielke¹⁰ compared both photo and video reproductions to a real-size mock-up of the same situation. The results revealed a consistency for the lighting characteristics of brightness and contrast.

Besides discussing the feasibility of applying these media in presenting lighting environment, other studies were focusing on improving the technique of computer-generated images (CGI), aimed at an optimal simulation of the physical environment. Since 2001, Drago et al.¹² and Newsham et al.⁷ have evaluated the lighting simulation accuracy and rendering fidelity of different algorithm approaches and obtained a high-quality, realistic colour image that provides a perceptual match to the real world. Villa et al. used CGI to present lighting configurations and carried out a series of psychovisual tests to select the most suitable tone-mapping operators,¹³ light simulation program and post-production operations¹⁴ for subjective visual comfort.

Although the approaches mentioned above have been frequently used and deeply explored as the subjects cannot get into the space they are unable to accurately immerse themselves and judge the actual experience of the space. Therefore, the presentation-ability of these means is still questioned by the lighting research community.

The goal of this study is to find an alternative media that can present a lighting environment efficiently (and that can replace a physical lighting environment), and test whether it is adequate in presenting physical lighting environments and to measure user’s impressions within various settings.

1.1. Virtual reality and its applications

Virtual reality (VR) is an interactive experience taking place within a virtual environment that incorporates mainly auditory and visual but also other types of sensory feedback like haptic. This immersive environment can be similar to the real world, or it can be fantastical, creating an experience that is not possible in ordinary physical reality. As a new research hotspot, VR is considered to be the most promising media to present the physical world and hopefully may be the best alternative media to replace physical test rooms.

VR can create highly immersive virtual spaces and allow end-users to experience the space from a first-person point of view. VR has been widely applied in the design industry, such as urban and community planning^15,16 and architectural design.¹⁷ VR can result in improved design quality, as it allows end-users to identify design issues within the immersive virtual environment (IVE) that could not be achieved using 2D drawings or building information models; and to compare alternative designs, such as differences in spatial geometries, material finishes and lighting. For example, Majumdar et al.¹⁸ used VR to assist design. During the design process, key professionals and decision-makers were invited to meet in an IVE, and to evaluate and provide feedback during the conceptual design phase of a courtroom. This process reduced the decision-making time significantly and allowed for real-time modifications to the design. Dunston et al.¹⁹ used VR in designing a healthcare building. The design team brought end-users, such as doctors and nurses, into the IVE of a healthcare organization and asked them to evaluate different designs and give their feedback. Based on the feedback, the designers made the necessary changes and adjustments according to end-user advice, resulting in an improved and human-centric design.

The abovementioned cases of VR technology being used in assisting designs suggest that IVEs can offer users a similar impression of a physical environment. This also proves that VR technology can be used in scientific research.

By setting up precisely controlled and readily modulated VR reproductions of physical spaces and immersing participants in them, VR technology provides a better approach for evaluating environmental factors, such as lighting environments. For example, Heydarian et al.^20–22 conducted a benchmarking study on natural light to compare sense of presence and immersion between VR and a physical space, using VR to explore human impressions towards lighting and their behaviours in an IVE, particularly the influence of personal control on the available lighting options in an office environment. In the experiment, they were introduced to various manual and semi-automatic control options to increase the available lighting in an office environment through the use of either natural or artificial light.

Chamilothori et al.²³ used VR as an alternative environment to conduct subjective assessments on daylit spaces, and investigated its adequacy to evaluate five aspects (the perceived pleasantness, interest, excitement, complexity and satisfaction) of subjective perception of daylit spaces. The study compared the subject’s perception of a real daylit environment and its equivalent representation in VR and tested the effect of the display method on the participants’ perceptual evaluations, reported physical symptoms and their perceived presence in the virtual space. The results indicate a high level of perceptual accuracy, showing no significant differences between the real and virtual environments on the studied evaluations.

1.2. Research purpose

Even though studies have found that human performance, perception and behaviour in an immersive virtual environment are not significantly different from that in a physical environment,²⁴ VR still faces some criticisms associated with possible discrepancies due to perceptual alterations. Particularly in lighting research, very few studies have focused on the congruence of VR versus a physical lighting environment and on the congruence of VR versus other media (such as photo and video). In this paper, we compare the subjective feelings of humans when subjected to physical lighting environments and the physical lighting environments presented by display media, namely VR, video and photo. This research discusses in detail and quantifies the congruence of using virtual reproductions in representing a lighting environment versus the actual physical lighting environment.

1.3. Perceptual attributes

A large number of studies^25,26 have noted that different display media differ in presentation characteristics, resulting in different abilities of the reproduced environment to present the actual physical environment. Different presentation characteristics and abilities affect the end-user’s direct impression of the reproduced environments. Previous studies have shown that the following six subjective senses are of great importance to evaluating the presentation abilities of display media: sense of presence,^20,27 sense of realism,^28,29 sense of field depth,^30,31 sense of stereo,³² sense of comfort and sense of clarity.^3–5 For instance, Whisker et al.²⁷ noted that sense of presence decides the end-user’s subjective experience of being in one place or environment, even when one is physically situated in another. In Heydarian et al.’s research,²² a sense of presence was considered one of the key measurements in an IVE through questionnaires. Slater et al.²⁸ studied the impact of visual realism (geometric and illumination realism) on sense of presence assessed through subjective, behavioural and physiological data in an immersive virtual environment, and showed that a better sense of realism can enhance the subjects’ realistic response.

2. Method

In this study, a physical test room with different lighting scenes was constructed at Tongji University, Shanghai, China. Based on it, VR, photo and video reproductions of the physical lighting scenes were created. The subjects were asked, via a questionnaire, to compare the presentation-ability of the VR environment, photo and video to the physical space, and also to assess perceptual and emotional attributes of the lighted environments in the real-world reference scene and for each virtual reproduction displayed in the three presentation modes.

2.1. Test room

Figure 1 shows the layout of the physical test room. It comprises a ‘reference room’ and a ‘display room’. The reference room was 4.3 m long, 3.6 m wide and 2.8 m high and had two windows on the South wall. However, daylight penetration in the experiment was controlled using shades. In the reference room, a bed, desk, cabinet and computer were arranged as shown (Figures 1 and 2). The area of the display room was approximately 15.5 m², in which a TV monitor (12176 mm long, 6849 mm wide) and VR equipment were placed to display the reproductions. The ambient lighting of the display room was kept at a relatively low luminance, approximately 3 cd/m²,^33,34 to ensure that the space did not interfere with the subjects’ evaluation of the reproductions and to ensure that the subjects’ visual focus was always on the screen.

Figure 1

Layout of the reference and display rooms VR: virtual reality

2.2. Luminaires and reference lighting scenes

This experiment used four light-emitting-diode (LED) panel lamps. The LED panel lamps were mounted in the ceiling of the reference room. As shown in Figure 2, all four lamps faced directly down and were dimmable both in correlated colour temperature (CCT) and light intensity via a dimmer. Table 1 shows the parameters of the reference lighting scenes. In this experiment, three CCTs were chosen as reference lighting scenes, including warm white (3000 K), neutral white (4000 K) and cool white (5500 K). In all the reference scenes, the average illuminance on the work plane (75 cm above the floor) of the reference room was set at 800 lux.

Figure 2

Section of the reference room

Table 1

Technical parameters of reference scenes

Scenes	Media	CCT (K)	Illuminance
S1	Physical space	3000	800 lux
S2		4000
S3		5500

Note: Trial scenes: before the test, a reference scene was chosen randomly to be used as a trial scene.

CCT: correlated colour temperature.

Before the test, a reference scene was chosen randomly as a trial scene for the subjects to go through the questionnaire and to ensure their understanding of the questions.

2.3. Reproductions

Tables 1 and 2 show the 12 lighting scenes, which included three reference scenes (in the physical test room) and nine displayed scenes. The nine displayed scenes were reproductions of the reference scenes, three each made using VR, video and photo media separately and presented in the display room.

Table 2

Technical parameters of displayed scenes

Media	Recording and presentation equipment	Recording parameters	Display parameters
VR	360 fly camera and Meizu MX4 Pro	1504 × 1504 30 fps 10,352 kbps	Screen size: 5.5 in. Resolution: 2560 × 1536 Pixel density: 546 ppi
Video	Canon EOS650D camera lens 17.0 mm and Samsung TV monitor (LED, UA46EH5080R)	1920 × 1088 25 fps 43183 kbps	Screen size: 46 in. Resolution: 1920 × 1080
Photo		5184 × 3456 1/51 seconds Iso-100; f/3.5

VR: virtual reality.

2.3.1. Virtual reality

In this study, VR scenes were reproductions of the physical test room, which were recorded using a VR camera (360 fly camera) and displayed on a smartphone using an app called 360 fly. When displaying the VR video, the phone was placed in a VR helmet. The procedures to create a VR scene were as follows:

Use the 360 fly camera to record panoramic videos of the reference scenes in the physical test room. During the recording, the camera was mounted on the wall 1.75 m above the floor, as shown in Figures 1 and 2;

Import the panoramic videos into a smartphone and convert them to immersive virtual videos with the 360 fly app;

Play the immersive virtual video on the smartphone and place the smartphone into a VR helmet. The subjects then put on the VR helmet and immersed themselves in the virtual scenes.

In the VR scenes, the subjects could turn their head and walk forward and backward to get a better sense of presence. Figure 3 shows a typical VR scene.

Figure 3

An immersive virtual reality environment (available in colour in online version)

The technical parameters of the 360 fly camera are as follows:

Lens: horizontal 360° and vertical 240° wide-angle

Aperture: f/2.5

Resolution: 1504 × 1504

Camera sensor size: 1/2.3 in.

Frame rate: 30 fps

The display parameters of the smartphone are as follows:

Screen size: 5.5 in.

Resolution: 2560 × 1536

Pixel density: 546 ppi

2.3.2. Video and photo

The video and picture reproductions were both recorded using a Canon EOS650D camera (Camera lens 17.0 mm) and displayed on a Samsung TV monitor (LED, model no. UA46EH5080R). The camera was mounted on the wall (1.75 m above the floor) as shown in Figure 1. The technical parameters are shown in Table 2. Figure 4 shows the typical photo and video reproductions.

Figure 4

Photo and video reproductions (available in colour in online version)

As the photos and VR videos were made by different recording and display equipment, chromatic aberration inevitably occurred between the reproductions. Such differences and deviations were ignored in this study.

2.4. Questionnaire

Table 3 shows the contents of the five-part questionnaire. In the experiment, the participants were asked to provide personal information (e.g., age, gender), evaluate the presentation-ability of the three media, assess perceptual and emotional attributes and, finally, judge their overall satisfaction regarding all scenes.

Table 3

Questionnaire content

Personal information: (age, gender, etc.)
Presentation-ability rating
		0		0.2	0.4	0.6	0.8	1
Sense of presence
Sense of realism
Sense of field depth
Sense of stereo
Sense of comfort
Sense of clarity
Perceptual attributes rating
	3	2	1	0	−1	−2	−3
Open								Close
Warm								Cold
Refreshing								Muggy
Cozy								Desolate
Diffuse								Glaring
Bright								Dim
Emotional attributes rating
	3	2	1	0	−1	−2	−3
Relaxed								Nervous
Pleasant								Unpleasant
Alert								Sleepy
Unrestrained								Restrained
Noisy								Quiet
Overall satisfaction rating
	3	2	1	0	−1	−2	−3
Satisfied								Dissatisfied

Note: This questionnaire was written in Chinese; native language of all the participants was mandarin.

As described in Section 1.1, the overall performance in giving the subjects a sense of presence, sense of realism, sense of field depth, sense of stereo, sense of comfort and sense of clarity is defined as presentation-ability. Moreover, the sense of presence, sense of realism, sense of field depth, sense of stereo, sense of comfort and sense of clarity were defined as presentation factors.

As shown in Table 3, to rate the presentation factors, a six-point scale (0, 0.2, 0.4, 0.6, 0.8 and 1) was used, 1 meaning the performance of a certain presentation factor was perfect and 0 meaning it was the worst.

In lighting research, semantic differential scales are widely used in subjective evaluation.^35,36 In this study, the perceptual attributes, emotional attributes and overall satisfaction ratings all used this evaluation method, and seven-point semantic differential scales (from −3 to 3) were used. Moreover, by using the adjective pairs shown in Table 3, the subjects expressed their impressions of the scenes and lighting environments.

2.5. Recruitment

There were 40 subjects recruited for this study, 21 of which were male and 19 female; 24 of them were undergraduate or graduate students of Tongji University, aged between 18 and 30; eight were aged between 31 and 50; and eight were older than 51 years. Most of the subjects had educational backgrounds in architecture and could understand the questions containing professional technical nomenclature in the questionnaire. Therefore, as there is no need to explain each individual lighting attribute, a single-blind experiment can be easily achieved. All of the subjects had normal visual acuity or wore glasses that gave them qualified visual acuity. As screened at Shanghai Municipal Hospital, all of the subjects were confirmed to be free of ocular defects.

2.6. Procedure

The experimental procedure had the subjects first rate the reference scenes in the reference room and then evaluate the displayed scenes in the display room. Table 4 shows the details of the experimental procedure. Before the test, a trial scene was setup for the subjects to go through the questionnaire and to confirm their understanding of the questions. The trial scene was a randomly chosen reference scene. While the subjects were judging the trial scene, the operator introduced and explained the questionnaire. After evaluating the trial scene, the subjects were asked to evaluate the reference lighting scenes. In order to eliminate the effect of the subjects’ immediately prior lighting experience, the subjects were asked to wear blindfolds both before their first test and between tests. After completing the rating of the reference scenes, the subjects entered the display room and evaluated the displayed scenes. The photos, videos and VR reproductions were displayed in random order, and the subjects wore blindfolds between the tests.

Table 4

Details of the experimental procedure

Subjects	Operator	Scenes
a) Enter the reference room, sit on the bed.	a) Enter the reference room; give the subjects the questionnaire.	/
b) Wear blindfold (1 minute).	b) Switch the light to trial scenes.	Trial scenes
c) Take off blindfold; evaluate the space and lighting environment; complete the questionnaire under the guidance of the operator (5 minutes, free gaze).	c) Help the subjects understand the questions and guide them to complete the questionnaires.	Trial scenes
d) Wear blindfold (1 minute).	d) Switch to reference scenes.	Reference scenes
e) Take off blindfold; evaluate the space and lighting environment, complete the questionnaire (5 minutes, free gaze).	e) Order of reference scenes was random.
f) Complete all reference scenes as for steps d) and e).	e) Order of reference scenes was random.
g) Enter the display room; sit on the chair; wear blindfold (1 min).	f) Enter the display room; switch to displayed scenes.	Experimental scenes
h) Take off blindfold; evaluate the space and lighting environment, complete the questionnaire (5 minutes, free gaze).	g) Order of displayed scenes was random.
i) Complete all displayed scenes as for steps g) and h).	g) Order of displayed scenes was random.

3. Data analysis and results

By adding all of the scores of the subjects and calculating the mean value of each lighting attribute separately, the descriptive statistics were attained. They are summarized in Tables 5 to 7.

Table 5

Presentation factors: Mean ratings and standard deviations

SCENE	MEDIUM	CCT	presence		realism		field depth		stereo		comfort		clarity
SCENE	MEDIUM	CCT	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV
S1	real space	3000K	0.84	0.177	0.86	0.158	0.82	0.168	0.82	0.191	0.74	0.182	0.80	0.222
S2		4000K	0.88	0.134	0.9	0.136	0.84	0.152	0.8	0.203	0.8	0.240	0.88	0.162
S3		5500K	0.88	0.134	0.88	0.162	0.86	0.130	0.86	0.158	0.76	0.177	0.8	0.157
S4	video	3000K	0.52	0.186	0.58	0.211	0.54	0.204	0.58	0.142	0.54	0.182	0.58	0.202
S5		4000K	0.52	0.134	0.52	0.162	0.48	0.226	0.52	0.186	0.52	0.186	0.56	0.152
S6		5500K	0.56	0.177	0.6	0.157	0.52	0.186	0.58	0.168	0.5	0.163	0.6	0.128
S7	photo	3000K	0.46	0.204	0.54	0.158	0.46	0.130	0.56	0.152	0.58	0.168	0.6	0.157
S8		4000K	0.44	0.152	0.5	0.136	0.4	0.091	0.42	0.142	0.4	0.222	0.58	0.211
S9		5500K	0.52	0.162	0.54	0.182	0.48	0.162	0.52	0.207	0.46	0.158	0.58	0.211
S10	VR	3000K	0.68	0.134	0.66	0.093	0.68	0.134	0.64	0.122	0.54	0.182	0.48	0.162
S11		4000K	0.8	0.091	0.8	0.091	0.76	0.122	0.74	0.130	0.68	0.099	0.66	0.130
S12		5500K	0.68	0.134	0.7	0.136	0.66	0.093	0.58	0.142	0.54	0.158	0.58	0.109

CCT: correlated colour temperature; STDEV: standard deviation; VR: virtual reality.

Table 6

Perceptual attributes: Mean ratings and standard deviations

SCENE	MEDIUM	CCT	open-close		warm-cold		refreshing -muggy		cozy-desolate		diffuse -glaring		bright-dim
SCENE	MEDIUM	CCT	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV
S1	real space	3000K	0.50	1.377	2.40	0.672	0.60	1.446	2.40	0.672	1.50	1.377	1.20	1.418
S2		4000K	1.40	1.446	0.40	1.374	1.80	1.265	0.40	1.128	1.00	1.432	2.10	1.150
S3		5500K	1.50	1.519	-0.20	1.344	2.30	0.791	-0.20	1.682	0.40	1.446	2.30	0.911
S4	video	3000K	0.70	1.814	1.30	1.018	0.70	1.114	0.80	0.883	1.00	1.109	1.30	1.285
S5		4000K	0.50	1.881	-0.60	1.446	1.60	1.033	0.00	1.109	0.20	1.344	1.50	1.132
S6		5500K	0.60	1.646	-0.30	1.506	1.20	1.418	0.00	1.281	0.60	1.033	1.60	1.033
S7	photo	3000K	-0.10	1.892	0.80	1.418	1.10	1.392	1.00	1.013	1.30	1.114	1.30	1.114
S8		4000K	-0.10	1.892	-1.10	1.722	1.10	1.392	-1.00	1.867	0.00	1.432	1.60	0.810
S9		5500K	1.20	1.181	0.20	1.344	1.80	0.992	0.60	1.128	0.90	1.057	1.90	0.841
S10	VR	3000K	1.10	1.150	1.40	0.928	0.60	1.215	1.30	1.203	1.60	1.033	0.90	0.709
S11		4000K	1.70	1.114	0.10	1.464	1.60	0.928	0.50	1.519	0.70	1.018	1.60	1.215
S12		5500K	1.10	1.317	-0.90	1.598	1.40	0.810	-0.50	1.649	0.20	1.091	1.80	0.758

CCT: correlated colour temperature; STDEV: standard deviation; VR: virtual reality.

Table 7

Emotional attributes and satisfaction: Mean ratings and standard deviations

SCENE	MEDIUM	CCT	pleasant unpleasant		alert-sleepy		unrestrained- restrained		noisy-quiet		relaxed-nervous		satisfied- dissatisfied
SCENE	MEDIUM	CCT	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV	mean value	STDEV
S1	real space	3000K	1.40	1.128	0.90	1.057	1.30	1.363	-0.30	1.636	1.90	0.955	1.00	1.281
S2		4000K	1.40	1.446	2.00	1.109	1.60	1.215	-0.60	1.128	1.50	1.132	1.70	1.285
S3		5500K	1.00	1.432	2.50	0.816	1.20	1.181	-1.40	1.033	1.20	1.556	1.60	0.928
S4	video	3000K	0.90	1.598	0.90	1.392	0.40	1.516	0.10	1.236	0.80	1.556	0.90	1.150
S5		4000K	0.10	1.464	1.10	1.317	0.40	1.582	-0.80	0.992	0.30	1.572	0.70	1.363
S6		5500K	0.30	1.363	1.20	1.488	0.80	1.682	-0.70	1.114	0.90	1.464	0.90	0.841
S7	photo	3000K	0.80	1.181	1.20	1.344	0.70	1.285	-0.90	1.236	1.20	1.181	0.80	0.758
S8		4000K	0.00	1.867	0.70	1.572	0.00	1.754	-1.50	0.934	0.00	1.695	0.70	0.911
S9		5500K	1.00	0.906	1.80	0.758	0.60	1.033	-0.70	0.911	1.30	1.018	1.40	0.672
S10	VR	3000K	1.00	1.109	0.30	1.436	0.90	0.955	-0.20	1.620	1.10	1.236	1.10	0.545
S11		4000K	1.40	0.810	1.80	0.883	1.10	0.955	-0.60	1.516	1.20	1.091	1.60	0.928
S12		5500K	0.10	1.598	1.60	1.215	0.70	1.285	-0.90	1.236	0.10	1.722	1.00	1.013

CCT: correlated colour temperature; STDEV: standard deviation; VR: virtual reality.

3.1. Presentation factors analysis

3.1.1. Presentation-ability rating

Figure 5 shows a radar chart that contains the mean ratings of the six presentation factors of the subjective senses, namely, the sense of presence, sense of realism, sense of field depth, sense of stereo, sense of comfort and sense of clarity. The area surrounded by these mean subjective rating scores suggested the presentation-ability of the media, with a larger area meaning higher scores on the rating scale, and hence better presentation-ability. The descending order of the media’s presentation-ability was VR > video > photo.

Figure 5

Mean ratings of presentation-ability of the different media VR: virtual reality

Based on the mean ratings, a Pearson correlation analysis indicated that the rating of the physical space showed a significant correlation with those on VR (p ≤ 0.05), while the correlations with video and photo were not significant (p > 0.5).

The subject-rated VR environment performed the best in five of the six presentation factors, showing high presentation-ability for the factors of presence, realism, field depth and stereo, with average scores of 0.72, 0.72, 0.70 and 0.65, respectively. For the rating of comfort, as the subjects wore a helmet, which was not the normal way they would otherwise experience a space and environment, the score for comfort was relatively low (0.59). In terms of clarity evaluation, VR was rated worse than video and photo (0.57). The reason for this was attributed to the use of less-advanced VR equipment. Nevertheless, the other results suggest that VR would have shown even better presentation-ability if the clarity of the VR videos could have been increased. It should be noted that even though the VR reproductions in this experiment were recorded and displayed in video format instead of a 3D immersive VR model, they still performed better than normal video and photo. This suggests that the display form of VR could improve the subjects’ feeling on presence, realism, field depth and stereo. If more movements and interactions (such as walking and moving objects) were introduced, VR would show higher presentation-ability. In conclusion, using VR to present lighting environments was more effective and reliable than using videos and photos. Furthermore, with the development of VR technology, the feasibility and reliability of applying VR in lighting research are expected to grow in the future.

3.1.2. Correlation among presentation factors

To understand the correlation among the presentation factors, Pearson correlation analyses were used to analyse the scores of presentation factors. The results show that correlations between presence, realism, field-depth, stereo, comfort and clarity were all significant (p < 0.05, see Table 8). This means that, if one factor varied, it could affect the subjects’ impression of the other factors. This could explain why VR ranks very high in most of the presentation factors. Since the subjects could interact (such as turning around and walking) with the IVE, it offered them a better sense of presence and realism; consequently, it increased the subjects’ impressions of other senses, such as stereo and field depth. In other words, to optimize the presentation-ability of VR equipment, all the presentation factors must be evenly improved; otherwise, the weaker factor would weaken the other factors.

Table 8

Correlations among presentation factors

	Presence	Realism	Field depth	Stereo	Comfort	Clarity
Presence	1	0.896**	0.827**	0.751**	0.638**	0.484**
Realism		1	0.798**	0.745**	0.623**	0.537**
Field depth			1	0.836**	0.648**	0.564**
Stereo				1	0.706**	0.602**
Comfort					1	0.615**
Clarity						1

p < 0.01 (bilateral), meaning significantly correlated.

3.1.3. Weight analysis of the presentation factors

In order to evaluate the weight of each presentation factor, data were analysed in SPSS, using dimension reduction analysis (DRA) and factor analysis. Table 9 shows the total variance explained, the variance (%) of each component means its contribution rate and indicates its weight to the presentation-ability. As shown in Table 9, the descending order weight of each component is the following: presence (0.740) > reality (0.109) > field depth (0.061) > stereo (0.048) >comfort (0.025) > clarity (0.016). The result indicates that, for presentation-ability, sense of presence is very important, as its weight is the highest among the factors and several times that of the other factors.

Table 9

Variance explained by different presentation factors

Components	Initial eigenvalue
Components	Total	% variance explained	Cumulative % variance explained
Presence	4.441	74.02	74.02
Reality	0.657	10.95	84.97
Field depth	0.366	6.10	91.07
Stereo	0.290	4.83	95.91
Comfort	0.151	2.51	98.42
Clarity	0.095	1.58	100.00

3.2. Presentation modes and lighting attributes

Data were analysed in SPSS. By conducting Friedman non-parametric tests and non-parametric post hoc analyses, the impact of presentation modes on the subjective rating on the lighting scenes was examined. As the p-values for the differences between presentation modes are all less than 0.05, it means that the difference in subjective ratings among the presentation modes is statistically significant. In order to further check the results, non-parametric tests of Friedman analyses were used to compare the paired samples; namely, real space and VR reproduction, real space and video reproduction and real space and photo reproduction. The p-values of these are 0.01, 0.01 and 0.05, respectively. These results indicate that the presentation modes can influence the subjects’ ratings of the lighting attributes.

To further evaluate the influence of presentation modes on the rating of each lighting attribute, according to the rating results of each of the lighting attributes separately, a Friedman test is carried out to compare the difference of the three presentation modes with the real space. Table 10 shows the results of each mode on each attribute. Most of the p-values are less than 0.05, which is consistent with the earlier tests, suggesting that presentation modes can influence the subjects’ ratings towards the overall lighting environment. However, for each lighting attribute, for photo reproductions, the rating difference on warm/cold, diffuse/glaring, bright/dim and noisy/quiet with the physical lighting environment is not statistically significant. This result indicates that photo reproductions can present these attributes with a non-significant difference from the physical lighting environment. Also, for video reproductions, the rating difference on bright/dim is not statistically significant. For VR reproductions, the results suggest that it can present open/close, diffuse/glaring, bright/dim and noisy/quiet attributes with a non-significant difference with the physical lighting environment.

Table 10

Statistical significance (p value) of the Friedman test between the ratings of the real space and those given under the different presentation modes

	VR	Video	Photo
Open/close	0.346**	0.007	0.004
Warm/cold	0.000	0.000	0.084**
Refreshing/muggy	0.000	0.000	0.003
Cozy/desolate	0.001	0.000	0.000
Diffuse/glaring	0.317**	0.000	0.190**
Relaxed/nervous	0.000	0.000	0.000
Pleasant/unpleasant	0.003	0.000	0.001
Alert/sleepy	0.000	0.000	0.000
Unrestrained/restrained	0.004	0.000	0.000
Noisy/quiet	0.646**	0.001	1.000**
Bright/dim	0.005	0.0736**	0.011**
Satisfied/dissatisfied	0.371**	0.000	0.000

VR: virtual reality.

p-value > 0.05, meaning difference is not statistically significant.

3.3. Congruence analysis

Data were analysed using SPSS for Distances Similarity Measures (from SPSS menus, choose among Analyse/Correlate/Distances) to evaluate the congruence of the reproduced environment and its corresponding physical environment. As shown in Table 11, the approximate coefficient of VR reproduction was 0.886, that of video reproduction was 0.800 and that of photo reproduction was 0.694. The approximate coefficients of the media had the following descending order: VR reproduction > video reproduction > photo reproduction. This result shows that, compared with video and photo, VR presented the lighting scenes closer to physical space, with fewer alterations.

Table 11

Approximate congruence coefficients for each presentation mode

	Physical space	VR	Video	Photo
Physical space	1.000	0.886	0.800	0.694

VR: virtual reality.

3.4. Overall satisfaction rating

Figure 6 shows the values of mean overall satisfaction rating for the physical lighting scenes and the three reproductions. Among the three reproductions, the subjects were most satisfied with the lighting scenes presented by VR. The mean satisfaction score of physical lighting scene was 1.43, and the mean satisfaction scores of VR reproduction, video reproduction and photo reproduction were 1.23, 0.83 and 0.97, respectively. The mean satisfaction score of VR was very close to those of the reference scenes and far higher than those of videos and photos. To double check the results and analysis above and evaluate the significance of the presentation modes’ impact on satisfaction rating, data were analysed in SPSS, running an analysis of variance (ANOVA). As shown in Table 12, at 4000 K and 5500 K, p-values are both less than 0.05, whereas at 3000 K, the p-value is 0.061, slightly higher than 0.05, meaning that the satisfaction rating difference among presentation modes is significant. These results confirm the previous conclusions on overall satisfaction rating.

Figure 6

Mean overall satisfaction ratings

Table 12

Impact of presentation modes on satisfaction rating

CCTs Results	3000 K	4000 K	5500 K
p	0.061	0.000	0.001
Eta	0.215	0.390	0.315

Furthermore, as shown in Figure 6, except for 3000 K, the satisfaction rating trends at 4000 K and 5500 K physical lighting scenes and their counterpart reproductions are similar to the trend of the mean value. Also, the ANOVA was conducted to evaluate the impact of CCT on satisfaction rating. As shown in Table 13, except for the video, the satisfaction rating difference among CCTs is significant, with p-values of 0.018 for real space, 0.000 for photo and 0.004 for VR, respectively. For video reproduction, the result indicates that the impact of CCT on satisfaction rating is not significant because the video is dynamic, and the movement is unpredictable to the subjects, leading to the losing sight of some information, such as CCT difference.

Table 13

Impact of correlated colour temperature on satisfaction rating

Modes Results	Real space	Video	Photo	VR
p	0.018	0.664	0.000	0.004
Eta	0.257	0.084	0.370	0.297

VR: virtual reality.

4. Summary and conclusions

In this study, we compared human subjective feelings towards a physical lighting environment, an IVE lighting environment and a set of video and photo reproductions. Through a variety of data analyses, the following conclusions can be drawn:

The order of presentation-ability of the media is physical space > VR > video >photo;

Compared to video reproductions and photo reproductions, an IVE shows higher congruence (approximate coefficient of 0.886) with physical scenes;

VR can present the lighting attributes of open/close, diffuse/glaring, bright/dim and noisy/quiet consistently to its counterpart physical environment;

the subjects are more satisfied with the VR reproductions.

Since VR can provide a more immersive environment that allows users to experience the environment from the first-person point of view, the novel display form of IVE was able to produce feelings that were very close to the true feelings of humans experiencing the actual physical scenes. The experimental results indicate that it is feasible to apply VR to represent lighting environments when evaluating lighting attributes such as open/close, diffuse/glaring, bright/dim and noisy/quiet. Moreover, to a certain extent, IVE can be used in some scientific investigations. With further development of VR technology, IVE will be able to accurately display a specific lighting scene, and so can be used in studies requiring increased precision.

Even though this study presented the results of a trial and preliminary experiments, the findings are quite reasonable. However, there still are many deficiencies that need to be solved in future studies, including the following.

In this experiment, the VR environment was recorded by a camera. Therefore, instead of a 3D model, the VR was presented in the form of a panoramic video. This surely compromised the rating of stereo and clarity. In the future, experiments can be performed to compare the congruence of different VR reproductions (recorded VR video and 3D immersive VR models) with the physical space.

Since the VR equipment was quite simple, the subjects could do only very small movements (such as turning their heads) to feel their presence in the IVE. Moreover, the playback of the VR videos experienced some lagging. All of these technical deficiencies influenced the subjects’ evaluation. Therefore, in the next stage of the study, a 3D immersive VR model and advanced VR equipment can be introduced. Such improvements will allow the subjects to walk in the IVE, and perform other activities that can be conducted in a maturely developed IVE.

As the recording and display equipment were not the same among the reproductions, there were differences in the resolution and colour inclinations among them. In the future, the photo, video and VR environments should be precisely designed to reduce these differences.

In conclusion, using VR to present lighting environments is more effective and reliable than using videos and photos. With the development of VR technology, the feasibility and reliability of applying VR in lighting research are expected to evolve.

Author Note

Y Chen is also affiliated to College of Architecture and Urban Planning, Tongji University, Shanghai, China.

Footnotes

Acknowledgements

We wish to thank all of the people who supported this study, including all of the members of the Luminous Environment Lab, Tongji University, Shanghai, China.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: the National Natural Science Foundation of China (Grant No. 51508396).

References

Czeisler

Allan

Strogatz

Ronda

Sanchez

Rios

Freitag

Richardson

Kronauer

. Bright light resets the human circadian pacemaker independent of the timing of the sleep-wake cycle. Science 1986; 233: 667–671.

Lam

. Seasonal Affective Disorder and Beyond: Light Treatment for SAD and Non-SAD Conditions, Washington DC: American Psychiatric Press, 1998.

Cui

Hao

. Lighting of a cardiac intensive care unit: emotional and visual effects on patients and nurses. Lighting Research and Technology 2018; 50: 701–715.

Chen

Hao

Cui

. Study on the human factors of lighting in neutral color living room. China Illuminating Engineering Journal 2014; 4: 29–32.

Chen

Hao

Cui

. Study on the human factors of lighting in different color-tone living rooms. China Illuminating Engineering Journal 2015; 4: 40–43.

Mahdavi

Eissa

. Subjective evaluation of architectural lighting via computationally rendered images. Journal of the Illuminating Engineering Society 2002; 31: 11–20.

Newsham

Richardson

Blanchet

Veitch

. Lighting quality research using rendered images of offices. Lighting Research and Technology 2005; 37: 93–115.

Cauwerts C. Influence of presentation modes on visual perceptions of daylit spaces. PhD thesis. Louvain, Belgium: Université Catholique de Louvain, 2013.

Murdoch

Stokkermans

MJM

Lambooij

. Towards perceptual accuracy in 3D visualizations of illuminated indoor environments. Journal of Solid State Lighting 2015; 2: 12–12.

10.

Schielke

. Validity of simulations for lighting and brand image evaluation. Lighting Research and Technology 2016; 48: 473–490.

11.

Moscoso

Matusiak

Svensson

Orleanski

. Analysis of stereoscopic images as a new method for daylighting studies. ACM Transactions on Applied Perception (TAP) 2015; 11: 21–21.

12.

Drago

Myszkowski

. Validation proposal for global illumination and rendering techniques. Computers and Graphics 2001; 25: 511–518.

13.

Villa C, Labayrade R. Psychovisual assessment of tone-mapping operators for global appearance and colour reproduction: Proceedings of Conference on Computer Graphics, Imaging and Visualization, Sydney, Australia, 7–10 August, 2010: pp. 189–196.

14.

Villa C, Parent E, Labayrade R. Calibrating a display device for subjective visual comfort tests: selection of light simulation programs and post-production operations: Proceedings of the CIE Lighting Quality and Energy Efficiency Conference, Vienna, 14–17 March, 2010.

15.

Maffei

Masullo

Pascale

Ruggiero

Puyana Romero

. On the validity of immersive virtual reality as tool for multisensory evaluation of urban spaces. Energy Procedia 2015; 78C: 471–476.

16.

Maffei

Masullo

Pascale

Ruggiero

Romero

. Immersive virtual reality in community planning: acoustic and visual congruence of simulated vs real world. Sustainable Cities and Society 2016; 27: 338–345.

17.

Niu

Pan

Zhao

. A virtual reality integrated design approach to improving occupancy information integrity for closing the building energy performance. Sustainable Cities and Society 2016; 27: 275–286.

18.

Majumdar T, Fischer MA, Schwegler BR. Conceptual design review with a virtual reality mock-up model: building on IT: Joint International Conference on Computing and Decision Making in Civil and Building Engineering. Rotterdam: CIB, 2006: 2902–2911.

19.

Dunston PS, Arns LL, McGlothlin JD Lasker JC, Kushner AG. An immersive virtual reality mock-up for design review of hospital patient rooms. In X Wang and JJ-H Tsai (eds) Collaborative Design in Virtual Environments. Dordrecht, Netherlands: Springer, 2011: pp.167–176.

20.

Heydarian

Carneiro

Gerber

Becerik-Gerber

. Immersive virtual environments, understanding the impact of design features and occupant choice upon lighting for building performance. Building and Environment 2015; 89: 217–228.

21.

Heydarian A, Carneiro JP, Gerber D, Becerik-Gerber B, Hayes T, Wood W. Immersive virtual environments: experiments on impacting design and human building interaction: Proceedings of the 19th International Conference on Computer-Aided Architectural Design Research in Asia (CAADRIA), Kyoto, Japan, 14–17 May, 2014: pp. 729–738.

22.

Heydarian

Carneiro

Gerber

Becerik-Gerber

Hayes

Wood

. Immersive virtual environments versus physical built environments: a benchmarking study for building design and user-built environment explorations. Automation in Construction 2015; 54: 116–126.

23.

Chamilothori

Wienold

Andersen

. Adequacy of immersive virtual reality for the perception of daylit spaces: comparison of real and virtual environments. Leukos 2018; 15: 1–24.

24.

Iachini

Coello

Frassinetti

Ruggiero

. Body space in social interactions: a comparison of reaching and comfort distance in immersive virtual reality. PLoS One 2014; 9: e111511–e111511.

25.

Kim

Wang

Love

PED

Kang

S-C

. Virtual reality for the built environment: a critical review of recent advances. Electronic Journal of Information Technology in Construction 2013; 18: 279–305.

26.

Alves

Bártolo

. Integrated computational tools for virtual and physical automatic construction. Automation in Construction 2005; 15: 257–271.

27.

Whisker

Baratta

Yerrapathruni

Messner

Shaw

Warren

Rotthoff

Winters

Clelland

Johnson

. Using immersive virtual environments to develop and visualize construction schedules for advanced nuclear power plants. Proceedings of ICAPP 2003; 3: 4–7.

28.

Slater

Khanna

Mortensen

. Visual realism enhances realistic response in an immersive virtual environment. IEEE Computer Graphics and Applications 2009; 32: 76–84.

29.

Diemer

Mühlberger

Pauli

Zwanzger

. Virtual reality exposure in anxiety disorders: impact on psychophysiological reactivity. The World Journal of Biological Psychiatry 2014; 15: 427–442.

30.

Rolland

Gibson

Ariely

. Towards quantifying depth and size perception in virtual environments. Presence 1995; 4: 24–49.

31.

IJsselsteijn

Oosting

Vogels

IMLC

de Kort

YAW

van Loenen

. A room with a cue: the efficacy of movement parallax, occlusion, and blur in creating a virtual window. Presence 2008; 17: 269–282.

32.

Parsons TD. Neuropsychological assessment using virtual environments: enhanced assessment technology for improved ecological validity. In Brahnam S, Jain LC, editors. Advanced Computational Intelligence Paradigms in Healthcare: Virtual Reality in Psychotherapy, Rehabilitation and Assessment. Berlin: Springer-Verlag, 2011: pp. 271–289.

33.

Commission Internationale de l’Eclairage. CIE Publication No. 41. Light as a True Visual Quantity: Principles of Measurement. Vienna: CIE, 1978.

34.

Stockman

Sharpe

. Into the twilight zone: the complexities of mesopic vision and luminous efficiency. Ophthalmic and Physiological Optics 2006; 26: 225–239.

35.

Flynn

Hendrick

Spencer

Martyniuk

. A guide to methodology procedures for measuring subjective impressions in lighting. Illuminating Engineering 1979; 8: 95–110.

36.

Hao

Cui

. Experimental study of lighting in CICU based on visual comfort – the effect of illuminance and lighting modes on patients and staff’s satisfaction at color temperature of 4000 K. China Illuminating Engineering Journal 2015; 26: 1–6.