1 Introduction

VR is currently utilized in various fields such as education (Predescu et al. 2023), healthcare (Seymour et al. 2002), gaming (Mól et al. 2008), manufacturing (Peruzzini et al. 2021), psychology (Riva 2022), among others. However, uncomfortable symptoms, such as VR sickness, continue to be problematic for many users. These symptoms, also referred to as cybersickness, VR-induced symptoms and effects (VRISE), visually induced motion sickness (VIMS), resemble real-world motion sickness experienced by users during or after virtual reality experiences (Keshavarz et al. 2014). Symptoms may vary among users but typically include dizziness, nausea, disorientation, vomiting, eye fatigue etc. VR sickness is common, especially for first-time users, and these symptoms can hinder user comfort and increase barriers to entry for VR technology. Therefore, addressing these issues is considered a challenge to expand the potential applications of VR technology, necessitating research into various methods to reduce VR sickness.

Research is underway to reduce VR sickness by adding visual elements or effects to virtual environments. One of these methods is reducing Field of View (FOV) (Bos et al. 2010). FOV refers to the range or angle of observable vision.

When users are not moving in reality but are moving virtually, the mismatch between user’s visual and their vestibular signals can cause VR Sickness. Reducing the FOV of the display can alleviate this by decreasing the conflict between these systems. However, it may also diminish presence by obscuring parts of the field of view. Another approach is Depth of Field (DOF) blurring (Carnegie and Rhee 2015), which involves blurring the background surrounding objects focused on, thus reducing sickness. In a stereoscopic scene, the conflict between the user’s perceived distance of an object in the virtual environment (VE) and the proximity of the display can lead to VR sickness. This issue is known as accommodation convergence conflict. Blurring effects can modify the perception of quantitative depth. Additionally, applying DoF blurring can reduce conflicts that result from the monocular regions in stereoscopic images. While this can help alleviate accommodation convergence conflict, the blurred areas might affect content performance and may not always reduce VR sickness, depending on the predefined distances. Rest Frame (RF) (Prothero 1998; Prothero et al. 1997; Prothero and Parker 2003) is a technology aimed at reducing VR sickness by placing virtual objects independent of the VE, serving as fixed visual elements to concentrate the user’s gaze and decrease conflicts in visual information. The virtual nose (Wienrich et al. 2018) is one such example. It is a virtual element fixed to the user’s head, moving with the user’s head movements to focus the user’s gaze and reduce the sensory mismatch between virtual and real environments, thus alleviating sickness. RFs can be implemented in various shapes, such as an orange sphere fixed in the sky (Marengo et al. 2019) or a metal mesh (Cao et al. 2018). While these RFs exist in the user’s field of view, they may obscure or be obscured by objects in the virtual environment, which is a drawback. However, compared to methods like FOV reduction and DOF blurring, RFs are known to achieve a similar reduction in VR sickness with less visible information loss (Shi et al. 2021).

While the RF method is generally known to achieve a reduction in VR sickness with minimal loss of visual information, for practical content applications, research is needed to minimize visible information loss to enhance user presence. In this study, we investigated the effects of varying the opacity and density of RF on reducing VR sickness and enhancing presence by minimizing the proportion of the user’s field of view occupied by RF. Through the analysis of real-time discomfort score (DS) of VR sickness, post-experiment Simulator Sickness Questionnaire (SSQ) responses (Kennedy et al. 1993), and Witmer & Singer (WS) scores (Witmer and Singer 1998) from participants, we identified the most effective RF configuration. By comparing with the results with this configuration, the credibility of our findings was further strengthened. It is hoped that this will lead to the discovery of new RF settings that improve user experience and help alleviate VR sickness.

The study is structured as follows: Chapter 2 introduces previous research on VR sickness and RF. Chapter 3 describes the RF designed for this study. Chapter 4 presents the hypotheses of the study, while chapter 5 outlines the experimental process and results aimed at identifying the most effective RF configuration. Chapter 6 explains the experiment and results comparing the identified RF with the no-RF condition. Chapter 7 discusses the findings, followed by the limitations in Chapter 8 and the conclusion in Chapter 9.

2 Related works

The cause of VR sickness is not entirely clear, but various theories, such as the poison theory (Treisman 1977), postural instability (Riccio and Stoffregen 1991), and sensory conflict theory (Reason and Brand 1975), attempt to explain its origins. The sensory conflict theory suggests that VR sickness occurs due to discrepancies in signals received by the user’s sensory organs. In immersive VR environments using Head Mounted Displays (HMDs), discrepancies mainly arise between visual and vestibular signals. For instance, if a user is walking on a flat surface in reality but visually perceives climbing stairs, the vestibular system doesn’t receive signals for vertical acceleration, unlike the visual cues the user expects. However, the sensory conflict theory does not explain why visual conflict leads to VR sickness symptoms such as nausea. The postural instability theory argues that symptoms arise from postural instability because people naturally strive to maintain posture and balance. Utilizing a treadmill (Jaeger and Mourant 2001), it is possible to mitigate VR sickness by synchronizing visual signals with signals from the vestibular system. However, the use of treadmills presents spatial and cost limitations, as well as the drawback of potentially increasing user fatigue.

According to the Rest Frame Hypothesis (RFH) (Prothero 1998; Prothero and Parker 2003), VR sickness is attributed to the conflict between the implied orientation and motion cues from fixed reference objects, which provide static cues while the user is moving. The hypothesis posits that humans select one of several fixed reference objects as a RF and interpret movement relative to it. Depending on the alignment between the selected RF and the user’s actual movement, the occurrence of sickness can be predicted. If the chosen RF does not allign with the actual movement, users are more likely to experience confusion or sickness. This indicates that the discrepancy between actual physical movement and visually perceived movement can be a factor that induces VR sickness.

To demonstrate this hypothesis, Prothero (1998) utilized the individual visual background (IVB) as a world-fixed (allocentric) RF. However, if users focused their gaze on the background rather than the IVB, the reduction of VR sickness was minimal. Therefore, Lin et al. (2002) designed RFs in the form of stable clouds, which led to a greater reduction in sickness compared to the world-fixed grid IVB. However, RFs in these studies were utilized in projection-based systems, neglecting users' exploration of their surroundings. On the other hand, the use of HMDs has enabled VR experiences, leading to the implementation of various forms of world-fixed RFs such as a desk (Zielasko et al. 2019), an orange sphere (Marengo et al. 2019), a cockpit, and radial (Luks and Liarokapis 2019). However, these RFs have not significantly impacted VR sickness reduction. In particular, the orange sphere may disappear from the user's field of view depending on the observation situation, thus potentially failing to serve its role as an RF.

Research on player-fixed (egocentric) RFs, such as the virtual nose (Wienrich et al. 2018), aimed to overcome the drawbacks of world-fixed RFs by designing RFs to always be present in the user's field of view regardless of their observation situation. The virtual nose, fixed to the user's head, remained in the field of view at all times, allowing it to fulfill its role as an RF and effectively reduce VR sickness without disrupting the VR experience. Somrak et al. (2021) proposed constructing player-fixed RFs in the form of glasses and hats. The experimental results of this study also indicated that player-fixed RFs reduce VR sickness; however, they showed the drawback of increasing VR sickness in slow-motion environments.

The static RF used in the study by Cao et al. (2018) was proposed as a method to address the drawbacks of both world-fixed and player-fixed RFs. In this research, a mesh-like world-fixed RF was implemented to renaub in the user's field of view, enabling observation of the VE without completely obstrucing it. The study showed that the proposed method alleviated VR sickness. However, the study did not analyze the impact of RF on presence, but since RF obstructs the field of view, it is likely to affect user presence. Our study aimed to investigate the effects of RF configurations with varying degrees of visual occlusion on both presence and VR sickness. Through this, the goal was to derive RF configurations that minimize visual occlusion while maximizing VR sickness reduction, providing guidelines for VR content creation.

3 Rest frame

In this study, the design factors of the RF included opacity and density, aiming to achieve various degrees of visual occlusion. In gaming,'opacity' is common graphic settings that affects transparent elements like smoke or flames, allowing users to see more through them as the opacity decreases. Conversely, increasing opacity makes transparent elements less see-through, potentially obscuring more. Drawing inspiration from this concept, the RF's opacity was adjusted to allow users to see the VE by penetrating the RF. Density, on the other hand, adjusts visual occlusion by removing a certain proportion of rendered RF images based on a grid. Thus, lower density and opacity result in minimal visual occlusion, while higher density and opacity lead to maximal occlusion.

The shape of the RF used in the previous study (Cao et al. 2018) has been demonstrated to have a positive effect on reducing VR sickness. Consequently, as shown in Fig. 1, the RF design leaves the center of the VR camera’s field of view (horizontal \(120^\circ \), vertical \(60^\circ \)) free of mesh, rendering the mesh only outside of this area. Users manipulate the rotation of the VR camera through both controllers and actual head movements. However, the RF mesh operates independently of the HMD’s movement, being attached to the virtual user’s position. It only rotates in response to controller input, aligning the forward direction of movement with the VR camera’s orientation. This approach allows users to face forward through the central, mesh-free area of the RF. For the experiment, the opacity and density of the RF were classified into three levels each (0.4: Small—(S), 0.7: Medium—(M), 1.0: Large—(L)). According to previous study (Hombeck et al. 2020), the acceptable range of opacity for overlays in 3D environments lies between 20 and 70%, as this range has been subjectively evaluated as suitable for user perception. In our experimental setup, opacity and density levels below 40% were nearly imperceptible to participants in prototype implementation, leading us to conclude that such low levels would not provide an effective RF effect. Consequently, 40% was selected as the minimum level, and to comprehensively evaluate the impacts on VR sickness reduction and presence, three levels—40%, 70%, and 100% (0.4, 0.7, 1.0)—were used in the experiment.

Fig. 1
figure 1

Rest frame design

Full size image

This classification creates various levels of visual occlusion based on opacity and density. The objective of the experiment is to analyze not only the effects of opacity and density factors on presence and VR sickness reduction but also the interaction between these two factors for various levels of visual occlusion. Figure 2 shows nine types of RF based on opacity and density levels. In the figure, the subscript of the RF represents the opacity level first, and the density level second. For example, RF with opacity 0.4 and density 0.7 is denoted as \({RF}_{SM}\).

Fig. 2
figure 2

View with Rest Frames. D represents fixed density RFs. For example, \({D}_{M}\) refers to \({RF}_{SM}\), \({RF}_{MM},\) and \({RF}_{LM}\). Similarly O represents fixed opacity of RF

Full size image

4 Hypothesis

The use of RF can yield a reduction in VR sickness, but it obstructs the user’s field of view, hindering presence. Therefore, if the visual occlusion of RF is low, the user’s presence is expected to increase. However, it is hypothesized that when the visual occlusion of RF is low, the reduction in VR sickness due to RF diminishes because the proportion of visual signals provided by RF to the user decreases. Thus, this study sets out to investigate the effects of opacity and density, two design factors of RF, on VR sickness and presence, as well as their interaction. No hypothesis was set regarding the interaction effect due to insufficient rationale to evaluate the interaction between the two factors. Because the relationship between opacity and density was not clearly defined, it was deemed appropriate to first examine each factor’s main effects on VR Sickness. Building on prior RF studies, the four hypotheses were formulated as follows:

  • H1: As the opacity of the Rest Frame decreases, the presence of VR users will increase.

  • H2: As the opacity of the Rest Frame decreases, VR sickenss reduction will decrease.

  • H3: As the density of the Rest Frame decreases, the presence of VR users will increase.

  • H4: As the density of the Rest Frame decreases, VR sickness reduction will decrease.

5 Experiment: opacity and densify effects

An experiment was conducted to investigate the impact of RF with various levels of visual occlusion on VR sickness and presence. The experiment was designed based on previous research (Cao et al. 2018) that demonstrated the effectiveness of RF in reducing VR sickness. Similar experimental settings and methods were employed to ensure consistency and comparability.

5.1 Participants

The experiment involved 18 participants (7 males, 11 females) aged between 20 and 26 years (mean \(\pm \) SD 22.8 \(\pm \) 1.9). According to pre-interviews, six participants had no prior VR experience, ten had experienced VR two times or less, and two were frequent VR users. To prevent VR adaptation effects, participants were evenly distributed across three groups based on VR experience level, each with a different opacity levels. All participants were recruited from Jeonbuk National University.

5.2 Apparatus

The application used in this study was developed using the Unity engine (version 2021.3.21f1). The Meta Quest Pro HMD was directly connected to a PC equipped with an Intel i7-9700 k processor (3.6 GHz), 16 GB RAM, and an NVIDIA GeForce RTX 2080 Super graphics card. As depicted in Fig. 3, participants were seated in chairs, allowing them to freely turn their heads. Participants’ virtual movement and rotation were enabled through the Meta Quest Touch Pro controllers.

Fig. 3
figure 3

A seated participant with Meta Quest Pro HMD and Touch Pro controllers

Full size image

In the VE, all participants followed a predefined path, as illustrated in Fig. 4, to ensure that their movements were as consistent as possible. To prevent deviation from the path, a warning message was displayed if participants strayed off the designated route.

Fig. 4
figure 4

The experimental map with white arrows. When passing through the waypoint (green arch), the screen transitions to evaluate the discomfort score (DS) (bottom right)

Full size image

5.3 Procedure

As shown in Fig. 5, the experiment consisted of 3 sessions. To minimize potential cross-session effects among participants experiencing the same VE, sessions were spaced 2 to 7 days apart, with RF density being a within-subject factor. Unlike RFs that display the same pattern of images regardless of opacity adjustments, varying RF images based on density may help further reduce potential cross-session effects. Each participant experienced all three RFs with varying densities. Within each group, participants were exposed to these RF densities in a completely counterbalanced order, ensuring that each unique sequence of densities was experienced by only one participant per group.

Fig. 5
figure 5

Design of a three-session experiment

Full size image

Each session consisted of 15 min of VR experience followed by 15 min of questionnaire completion. Before the first session, participants completed a consent form and received detailed instructions about the experimental procedure. After each session, participants filled out the SSQ and WS to evaluate VR sickness, subjective discomfort, and presence. In the next session, participants experienced the RFs with different densities and repeated the same procedure.

Considering the maximum threshold of a 15 min for immersive VR experience (DiZio and Lackner 1992; Stanney et al. 2003), participants completed four laps in the VE which included a total of six waypoints per lap. Following the real-time VR discomfort assessment method of the previous study (Fernandes and Feiner 2016), participants rated their current level of DS at each green arch structure-shaped waypoint along the path. At each waypoint, the display transitioned to the screen shown in Fig. 6, allowing them to rate their DS on a scale from 0 to 10. 0 indicates no discomfort, while 10 indicates severe discomfort, to the point of nausea. Participants were able to rate their DS by clicking the + or– buttons using the controller index trigger. After completing the rating, the screen transitioned back to the original VE scene. If a participant rated their discomfort as 10 or complained of symptoms during the experiment, the session would immediately terminate. In this case, a DS of 10 was assigned to the rest of the waypoints.

Fig. 6
figure 6

Screen for evaluating DS

Full size image

5.4 Result

In this study, each participant experienced three different densities of RF. Accordingly, each participant in the three groups experienced nine RFs with different degrees of visual occlusion, resulting in six data pairs for each participant and a total of 54 data pairs. Each data pair included three types of data: real-time DS, symptoms of VR sickness obtained through SSQ, and presence obtained through WS. To analyze these data, mixed-design ANOVA was conducted, examining the effects of RF opacity (a between-subject factor) and RF density (a within-subject factor). This analysis aimed to identify any statistically significant interactions between RF opacity and density, as well as differences between groups. After significant differences were found through ANOVA, Tukey’s HSD test was used for post hoc analysis.

5.4.1 Discomfort score

When the data is divided based on opacity as shown in Fig. 7a, the group \({O}_{S}\) with opacity 0.4 had 11 out of 18 participants completed all 24 waypoints, while \({O}_{M}\) had 11 participants, and \({O}_{L}\) had 12 participants complete them. Groups \({O}_{S}\) and \({O}_{L}\) showed a tendency for more participants to complete the waypoints with repetitions of the sessions, but in group \({O}_{M}\), 2 out of 6 participants did not exhibit such a tendency. The horizontal axis represents the waypoints passed by participants.

Fig. 7
figure 7

DSs for individual participants (thin line) and average (bold line) over time: a DS for Opacity, b DS for Density. The vertical axis represents DSs, and the horizontal axis represents waypoints. Diamonds indicate the last completed waypoint for each participant

Full size image

Table 1 presents the statistical results. According to the analysis in Table 1, there was a significant difference in DS concerning opacity. The group \({O}_{M}\) with an opacity 0.7 exhibited a higher DS (\({M}_{{O}_{M}}=5.6, \)\({SD}_{{O}_{M}}= 2.1,\)\({M}_{{O}_{S}}=3.9, \)\({SD}_{{O}_{S}}= 1.5,\)\({M}_{{O}_{L}}=3.5,\)\( {SD}_{{O}_{L}}= 1.9\), F(2,15) = 3.208, \({\upeta }^{2}\) = 0.12, p < 0.05) compared to \({O}_{S}\) and \({O}_{L}\). Here, M refers to the mean value, and SD to the standard deviation. However, there was no statistically significant difference in density or interaction effect at \(\alpha =.05\). Moreover, the significant difference in opacity mainly appeared between \({O}_{M}\) and \({O}_{L}\) and between \({O}_{S}\) and \({O}_{M}\) (p < 0.05). No significant difference was found between \({O}_{L}\) and \({O}_{S}\).

Table 1 Statistical result of DS and finished waypoint
Full size table

5.4.2 Waypoints

Figure 8 illustrates the average number of completed waypoints and individual data for participants who experienced the respective RFs. The numbers at the top indicate the count of participants who completed all waypoints when experiencing the respective RF. Similar to previous research, the number of completed waypoints was measured instead of the time participants spent in the experiment to avoid disregarding moments when participants remained stationary in the VE. According to the experimental results, the \({RF}_{ML}\) had the fewest completed waypoints, while \({RF}_{LM}\) had the most (\({M}_{SS}=18.83\), \({SD}_{SS}=8.72\),\({M}_{SM}=21.33\), \({SD}_{SM}=3.67\), \({M}_{SL}=22.83\), \({SD}_{SL}=2.40\), \({M}_{MS}=20.00\), \({SD}_{MS}=6.96\), \({M}_{MM}=21.33\), \({SD}_{MM}=6.53\), \({M}_{ML}=17.17\), \({SD}_{ML}=8.13\), \({M}_{LS}=21.83\), \({SD}_{LS}=4.02\), \({M}_{LM}=23.17\), \({SD}_{LS}=2.04\), \({M}_{LL}=21.00, {SD}_{LL}=4.00\)). However, according to Table 1, there was no statistically significant difference observed for opacity, density, and the interaction effect at \(\alpha =.05\).

Fig. 8
figure 8

The average and individual data of finished waypoints

Full size image

      

5.4.3 SSQ

After analyzing the results of the SSQ completed by participants post-experiment, we focused on 16 items to analyze three subscales (nausea, oculomotor, disorientation) and the total score. While the symptom of nausea was highest for \({O}_{S}\) (\({M}_{{O}_{S}}=85.33, {M}_{{O}_{M}}=75.26, {M}_{{O}_{L}}= 60.95\)), other symptoms were generally highest for \({O}_{M}\) and \({D}_{M}\), where opacity and density were both set to 0.7. However, according to Table 2, statistically significant differences were observed only in the nausea and total score for opacity (N: F(2,15) = 13.09, \({\upeta }^{2}\) = 0.05; TS: F(2,15) = 10.20, \({\upeta }^{2}\) = 0.01, p < 0.05). No statistically significant differences were found for other items related to opacity (O: F(2,15) = 5.23; D: F(2,15) = 0.34, p > 0.05) and density (N: F(2,15) = 0.48; O: F(2,15) = 0.58; D: F(2,15) = 0.08; TS: F(2,15) = 0.37, p > 0.05). Similarly, no statistically significant interaction effect was observed either (N: F(2,15) = 0.09; O: F(2,15) = 0.16; D: F(2,15) = 0.15; TS: F(2,15) = 0.02, p > 0.05). This indicates that contrary to our hypothesis, reducing the density of RF did not significantly affect the reduction in VR sickness; the effect remained consistent even with decreased density. In contrast, opacity impacted the SSQ measures, particularly in the nausea and total score; however, no statistically significant differences were observed between the individual opacity levels (Table 3).

Table 2 Statistical Results of the main effect and interaction effect SSQ
Full size table
Table 3 Statistical Results of the main effect and interaction effect of WS
Full size table

5.4.4 WS

Analyzing presence across five subcategories-realism, possibility to act, quality of interface, possibility to examine, self-evaluation of performance-through 19 WS items, we found no statistically significant differences in opacity except realism and self-evaluation of performance (Realism: F(2, 15) = 6.98, \({\upeta }^{2}\) = 0.03; Self-evaluation of Performance: F(2, 15) = 11.78, \({\upeta }^{2}\) = 0.07, p < 0.05; Possibility to Act: F(2, 15) = 6.55; Quality of Interface: F(2, 15) = 3.51; Possibility to Examine: F(2,15) = 0.02, p > 0.05). Additionally, there were no significant differences observed for density (Realism: F(2, 15) = 0.35; Possibility to Act: F(2, 15) = 0.16; Quality of Interface: F(2, 15) = 0.06; Possibility to Examine: F(2,15) = 0.10; Self-evaluation of Performance: F(2, 15) = 0.12 p > 0.05) or interaction effects (Realism: F(2, 15) = 0.10; Possibility to Act: F(2, 15) = 0.22; Quality of Interface: F(2, 15) = 0.36; Possibility to Examine: F(2,15) = 0.52; Self-evaluation of Performance: F(2, 15) = 0.16, p > 0.05). This suggests that decreasing the density of RF does not significantly impact presence. Although opacity influenced realism and self-evaluation of performance, no statistically significant differences were observed between the individual levels of this factor.

6 Experiment: comparing conditions with and without RF

In the first experiment, opacity was found to affect both VR sickness and presence. Specifically, when analyzing the DS results, opacity levels of 0.4 and 1.0 were both effective in reducing VR sickness compared to 0.7. Although no significant difference was observed between the two, their similar effectiveness, combined with the goal of reducing the field of view as least as possible, suggests that the opacity of 0.4 may be the most effective. Similarly, changes in RF density did not significantly affect VR sickness or presence, but the RF with a density of 0.4, which reduces the field of view the least, was considered to be the most effective. Therefore, we concluded that an \({RF}_{SS}\), with both opacity and density set to 0.4, is the most effective. Although \({RF}_{SS}\) showed the most significant effect among the nine RFs with varying levels of visibility reduction, it had not yet been compared to a no-RF condition to determine whether it further reduced VR sickness. Therefore, an additional experiment was conducted to compare the VR sickness reduction effects and sense of presence between the \({RF}_{SS}\)(Minimal RF, MRF) and a no-RF(No RF, NRF) condition.

6.1 Participants

The experiment involved 12 participants (8 males, 4 females) aged between 20 and 26 years (mean \(\pm \) SD: 24 \(\pm \) 1.6) who had not participated in the first experiment. Based on pre-interview assessments, Five participants had no prior VR experience, five had used VR two times or less, and two had extensive VR experience. Participants were evenly divided into two groups (MRF1NRF2, NRF1MRF2) based on their VR experience levels, with six participants in each group.

6.2 Apparatus & procedure

The apparatus and experimental procedure were the same as in the first experiment, with the only difference being that this experiment consisted of two sessions, as shown in Fig. 9. In this experiment, the condition (MRF, NRF) was treated as a within-subjects factor, while the order of conditions served as a between-subjects factor. Participants in the MRF1NRF2 group experienced the MRF condition first, whereas those in the NRF1MRF2 group experienced the NRF condition first.

Fig. 9
figure 9

Design of a two-session experiment

Full size image

6.3 Result

Similar to the first experiment, a mixed-design ANOVA was conducted with condition as the within-subjects factor and order of condition as the between-subjects factor. Tukey’s HSD test was also used for post hoc analysis.

6.3.1 Discomfort score

Figure 10 shows the real-time DS for participants in the NRF1MRF2 and MRF1NRF2 groups. It follows the same method as Fig. 7, with different colors representing each condition. For both groups, VR sickness decreased when experiencing the second RF.

Fig. 10
figure 10

DSs for individual participants (thin line) and average (bold line) over time: a DS for NRF1MRF2, b DS for MRF1NRF2

Full size image

Table 4 presents the statistical results, showing a significant difference between NRF1 and MRF2. The DS was higher for NRF1 than MRF2 (\({M}_{NRF1}=5.1, \)\( {M}_{MRF2}=2.1, \)\( F\left(\text{1,10}\right)=6.107, \)\({\upeta }^{2}=0.38, p<0.05\)). However, no statistically significant main effects were found at \(a=.05\), and there was no significant difference between MRF1 and NRF2.

Table 4 Statistical result of DS
Full size table

6.3.2 Waypoints

Figure 11 shows the number of waypoints completed by each group. According to the results, the fewest waypoints were completed when participants experienced NRF1. Both groups were able to complete all waypoints after experiencing the second RF condition (\({M}_{NRF1}=19.83,\)\({M}_{MRF2}=24.00, \)\( {M}_{MRF1}=23.50,\)\( {M}_{NRF2}=24.00\)). However, no statistically significant differences were observed at \(a=.05\).

Fig. 11
figure 11

The average and individual data of finished waypoints

Full size image

6.3.3 SSQ

As in the first experiment, SSQ scores were used as a secondary measure. Both groups showed lower SSQ scores after experiencing the second RF condition, with the highest scores recorded when participants experienced NRF1. However, as shown in Table 5, the main effects of RF conditions on SSQ (N: F(1,10) = 0.13, O: F(1,10) = 0.07, D: F(1,10) = 0.43, TS: F(1,10) = 0.05, p > 0.05) were not statistically significant.

Table 5 Statistical results of the main effect and interaction effect SSQ
Full size table

6.3.4 WS

Table 6 presents the statistical analysis of participants’ WS scores. According to Table 7, the main effects of NRF and MRF conditions (Realism: F(1, 10) = 0.01, Possibility to Act: F(1, 10) = 0.25, Quality of Interface: F(1, 10) = 0.07, Possibility to Examine: F(1,10) = 0.11, Self-evaluation of Performance: F(1, 10) = 0.49, p > 0.05) were not statistically significant at \(a=.05\). Similarly, interaction effects were also not statistically significant (Realism: F(1, 10) = 0.87, Possibility to Act: F(1, 10) = 0.82, Quality of Interface: F(1, 10) = 0.18, Possibility to Examine: F(1,10) = 4.12, Self-evaluation of Performance: F(1, 10) = 0.72, p > 0.05).

Table 6 Statistical results of the main effect and interaction effect of WS
Full size table

7 Discussion

This study examined the influence of RF opacity and density on VR sickness and presence through two experiments. The first experiment focused on evaluating the effects of various RF opacity and density levels on these measures, while the second experiment compared the most effective RF configuration identified from the first experiment to a no-RF condition. These findings offer insights into designing RF configurations that effectively balance VR sickness reduction and presence enhancement.

According to the DS analysis, difference in VR sickness symptoms were observed between \({O}_{S}\) and \({O}_{M}\), and \({O}_{M}\) and \({O}_{L}\). The symptoms were highest when the opacity was 0.7 in \({O}_{M}\), showing a significant difference in opacity according to the analysis. Consequently, participants who experienced \({O}_{S}\) and \({O}_{L}\) were able to maintain a more comfortable state over the long term compared to those who experienced \({O}_{M}\). Similarly, VR sickness symptoms due to RF density variations were highest for the medium-sized \({D}_{M}\), but no significant difference in density was observed. These results differed from the expected outcomes (H2, H4). The significant difference in VR sickness symptoms with opacity variations may be attributed to each experimental group experiencing different opacity conditions. However, within a single group, experiencing three densities of RF did not result in significant differences in VR sickness symptoms. This suggests that participants in that group had similar reactions to VR sickness despite experiencing various conditions. Such outcomes indicate that density variations in RF may be less critical than opacity in evaluating their impact on VR sickness. Therefore, to confirm this, additional experiments should divide groups based on density and expose them to various opacity conditions, contrary to our study design.

When comparing SSQ results across RF densities, \({D}_{M}\) showed the highest values. However, when comparing across opacities, nausea symptoms were highest for \({O}_{S}\) and lowest for \({O}_{L}\), while other symptoms were highest for \({O}_{M}\). Lower opacity levels corresponded to more pronounced nausea symptoms, possibly due to a decrease in the visual signals provided by the RF to the users. Other symptoms peaked at the medium opacity and density level of 0.7 in \({RF}_{MM}\), indicating that this level may impose the greatest restriction on the users’ field of view or movement. No statistically significant effect was observed for density; however, a significant effect was found for opacity, though no significant differences were found between individual opacity levels. The significant effect observed for opacity, similar to the DS results, suggests that the opacity of the RF may play an important role in VR sickness. However, the absence of significant differences between opacity levels and for density may indicate the possibility of VR adaptation among the experiment participants. Throughout repeated sessions, the number of completed waypoints increased, indicating that participants generally experienced increased comfort during the experiment over time. Similar findings were noted in previous studies. In Fernandes's (2016) study, despite using FOV restrictors in the first session and not in subsequent sessions, the results of the two sessions were similar. Similarly, in the study of Cao et al. (2018), there were no significant differences between the results of sessions with static RF experiences and those without RF experiences.

According to the WS analysis results, the presence experienced with \({O}_{M}\) and \({D}_{M}\) was lower compared to other RFs. Similar to SSQ, a significant difference was observed only in opacity, and it was found exclusively in the ‘realism’ and ‘self-evaluation of performance’ items n WS. However, no significant differences were found between the individual opacity levels. These results were contrary to the hypotheses H1 and H3 established before the experiment. Although there were no significant differences based on changes in RF density, the lower presence experienced with \({D}_{M}\) could be attributed to the somewhat heterogeneous nature of the RF shape with the applied density. However, the reason for not finding significant differences could be attributed to the lack of interactive elements during the VR experience, as participants mainly followed a predefined path without much opportunity for interaction. This may explain why no significant differences in RF opacity were observed for the WS items ‘possibility to act’, ‘quality of interface’, and ‘possibility to examine’. Ultimately, while the opacity of RF had an effect on presence, no significant differences were observed between the levels, and density did not affect presence. Nonetheless, \({RF}_{SS}\), which has the minimal visual occlusion, is considered the most effective RF.

The use of SSQ allows the detailed analysis of nausea into three sublevels of symptoms through various questions, presenting an advantage. However, post-experiment SSQ completion may be influenced by several factors such as prolonged VR experiences. While the effect size observed in the SSQ analysis was relatively small, this study embraced the DS analysis results for their advantage in providing real-time symptom feedback. In conclusion, excluding \({O}_{M}\), the VR sickness reduction effects among RFs were similar. Therefore, \({RF}_{SS}\), which has the minimal visual occlusion, is considered the most effective RF for VR sickness reduction in the first experiment.

When comparing the most effective RF (\({RF}_{SS}\)) identified in the first experiment with the no-RF condition, results showed that VR sickness was significantly reduced in the RF condition. However, these results may have been influenced by an order effect, as statistically significant differences were only observed when participants experienced the no-RF condition first, followed by the RF condition afterward. Furthermore, the greater number of waypoints completed in the RF condition suggests that the use of RF may be more effective than no-RF condition in reducing VR sickness, possibly because participants felt more comfortable in the VE for a longer duration.

8 Limitation

The study has several limitations. Firstly, the experimental design in the first experiment may have been influenced by order effects, as participants experienced the same VE three times with different RF conditions. This repetition could have made it difficult to fully isolate the impact of individual conditions on VR sickness. Additionally, the sample size of 18 participants in this study may be insufficient to fully validate the effects of the nine RF conditions.

While we considered using electroencephalograms (EEG) as an objective method for evaluating VR sickness, we chose a self-report approach following the method proposed by Prothero (1998), Fernandes et al. (2016), and Cao et al. (2018). Although this approach may lead to interruptions during VR experiences, it offers a significant advantage by providing real-time symptom feedback. For instance, even if two participants score the same on the SSQ, one participant may have completed all waypoints and spent more time in the VE. Despite the lack of objective measurement, such real-time feedback plays a crucial role in enhancing the usefulness and accuracy of the study.

9 Conclusion

The present study proposed an experiment to measure user presence and VR sickness reduction effects by adjusting opacity and density to create a total of nine different levels of visual occlusion in RF. The aim was to investigate the most effective form of RF that minimizes visual occlusions while maximizing user presence and mitigating VR sickness.

Although we did not verify whether the comparison with no-RF affects presence, the results suggest that using the reduced visual occlusion of RF, which is \({RF}_{SS}\), provides a similar presence to no-RF. Furthermore, since there was no significant difference between \({O}_{L}\) and \({O}_{S}\), it is anticipated that these two RFs provide similar VR sickness reduction effects. Therefore, with similar VR sickness reduction effects and the least visual occlusion, \({RF}_{SS}\) emerged as the most effective RF.

In this study, experiments were conducted using opacity and density to regulate the visual occlusion, similar to the metal mesh RF proposed by the previous study, which has been proven to reduce VR sickness. Further research is needed to validate if similar results can be obtained when experimenting with different forms of RF. Moreover, to mitigate the order effect observed in this study, it is necessary to recruit a larger number of participants and divide them into experimental groups based on density rather than opacity for validation. Through such additional experiments, it is expected that the limitations of this study can be overcome, and a deeper understanding of the effects of RF characteristics on VR sickness reduction and user presence can be achieved.