Feasibility and user experience of virtual reality fashion stores

Types of VR systems

On the other hand, immersive systems provide the user with a realistic experience of virtual environments (Costello 1997) created by highly interactive VR systems (Shahrbanian et al. 2012). Two frequently used output devices that render the virtual environment in immersive VR systems are HMD (Head mounted display) and CAVE (Cave Automatic Virtual Environment) (Ausburn and Ausburn 2004). Such systems require a great amount of resources such as high-performance processors, software, graphic cards, and tracking systems to provide a satisfactory level of realism (Costello 1997). As a result, the user feels a great sense of being in that environment.

VR user experience

A content analysis of VR studies in retail

To understand how VR was studied in retail, a content analysis was conducted to review the method, variables, contexts, and devices used in the previous studies. To select the studies to be examined, two independent researchers collected studies from databases. Combinations of the keywords, “virtual reality”, “consumer”, “shopping”, “retail”, and “store”, were used to search the relevant studies. The search resulted in two hundred articles. These articles were reviewed to evaluate the relevance of the articles to the interest and scope of the current study (i.e., VR use in retail). Many of these studies were not relevant because they either briefly introduced VR concept without truly investigating VR or their focus was technical methods to develop parts of a VR system. After the evaluation, 20 out of 200 studies were determined as relevant and used for further analyses.

Out of 20, the majority (85%, n = 17) of articles was empirical studies while only three (15%, n = 3) were conceptual studies. Among the empirical studies, the majority of the studies (82.4%, n = 14) used a non-immersive system, having the users view the virtual environment on a computer screen. Only three studies (17.6%) used an immersive VR system. Participants interacted with the VR environment using input devices in many studies (e.g., zooming in and out, rotating, clicking, etc.). Regardless of the VR system type, all studies had their participants seated in a chair (either in front of a computer monitor in non-immersive systems or in a room wearing a headset in immersive systems). Therefore, the participants did not change their location physically when they interacted with the VR environments.

While the sample size ranged greatly from 5 to 1313, the studies using an immersive VR system had a relative small sample size (10–41). The study contexts can be largely grouped into two types: online shopping mall (30%, n = 6) and brick-and-mortar store (40%, n = 8). Online shopping mall included e-commerce site which display products using VR techniques and virtual shopping mall where participants can navigate the environment. Several kinds of brick-and-mortar stores were used in the studies such as grocery stores, tobacco store, convenience stores, and restaurant.

The presence and absence of VR were used as independent variables to compare the differences caused by VR application in the previous studies. The perception of the environment (e.g., perceived crowding) or the display cues (e.g. the visibility of cigarette and health signs) were also investigated as independent variables. Dependent variables included emotion, attitude, and intention for the retail stores as well as the evaluation of the virtual reality experience (e.g., perceived discomfort, feeling of presence). The product or environment created by VR system caused higher level of pleasure and arousal (Oh et al. 2008), more positive attitude, and higher purchase intention (Lee and Chung 2005; Suh and Lee 2005) than VR absent condition.

The content analysis revealed the shortage of immersive device usage compared to non-immersive device in VR literature. With the emergence of immersive VR systems in the consumer market, it will be important to investigate the immersive VR systems to provide insights to the retailers. Moreover, no previous study created or investigated VR apparel stores. Compared to supermarkets or tobacco stores where the products are often packaged and displayed in boxes and the store layouts are simple grids, apparel stores tend to emphasize artistic and free layouts and have soft surface products. Therefore, the findings of previous VR studies may not be applicable to the apparel stores (Fig. 1).

Developing VR apparel stores

As the content analysis revealed, little previous research used a truly immersive system to understand the costs and benefits of using such a system. Thus, the focal interest of this study was to develop a VR environment that simulates a real shopping experience that allows users to move freely without using an input device.

The 3D modeling computer program “SketchUp” was used to create the virtual apparel stores. SketchUp is widely used for architectural and interior design in addition to mechanical and product design. Four models were developed in SketchUp to represent various kinds of store layouts and design themes that reflect typical fashion stores. Specifically, two store models were designed based on natural design themes (e.g., hardwood floors, wood walls) while the other two store models reflected modern design themes (e.g., metal wall panels). After the SketchUp models are created they were exported, converted to CAD data, and then converted again to VR data. Because CAD data are usually in a parametric form whereas VR data are in the form of tessellated polygons, a mesh of triangles (Purschke et al. 1998), CAD data need to be converted to polygons which define the surfaces of visual objects in VR. For example, if the surface of an object is curved in CAD data, arrangements of polygons that closely match the curved surface replace the surface in VR. Only this polygonal representation of a model allows for simultaneous rendering for VR (Beier 2000).

One common issue in creating a VR model is concerned with a number of polygons in the model. Because software renders objects by drawing polygons that approximate the target objects, modeling is simpler when the object is created with flat and angular faces (e.g., a box, pyramid). As the complexity of objects in a model increases, the number of polygons also increases (Nilsson et al. 2004). A large number of polygons is problematic as it requires a large amount of processing which slows down the VR system significantly (Marks et al. 2014). A VR model with more polygons reduces the rendering speed of the system, causing delayed display of the scene following the user’s head movement (Kim 2016). This affects a VR system’s ability to provide immersive experiences, and is known to induce simulator sickness (Kim 2016). To prevent this problem, model optimization tools [e.g., Level of Detail (LOD)] that reduce the number of polygons rendered for an object (Yilmaz et al. 2004) are used. However, such optimization processes may decrease the details and visual realism of models which interrupts user immersion (Kim 2016).

This polygon issue was particularly problematic for fashion VR retail stores due to the complexity of the products and various objects that are unique to fashion stores. A fashion retail store carries full product lines including over a thousand products and has a number of mannequins that display such products. These products and objects are made with a lot of curved surfaces and details (i.e., textures, trims, patterns) which poses a major challenge for developing a VR model store since it increases the number of polygons in the model. Possible solutions to resolve this issue may include using the most up-to-date computational resources (Berg and Vance 2017a, b) and optimization software tools (Martinez 2015).

Empirical testing with VR model stores

Apparatus

The testing of the VR stores was conducted at the Virtual Reality Design Lab, located in the public courtyard (25-by-25-foot) at the University of Minnesota. This VR lab is equipped with the necessary technologies to create an immersive VR experience. Four major technologies are needed to develop a truly immersive VR store: (1) the virtual headsets that provide the visual displays to the user and block out the real world sensory information, (2) the graphics rendering system that creates the visual sensations that are output to the user, (3) the tracking system that detects and reports the user’s position and orientation, and (4) the database construction and maintenance system that manages the models (Burdea and Coiffet 1994; Brooks 1999) (Fig. 2).

The lab is set up based on coordination between two systems: the tracking system and the graphics system [VRDL (n.d.)]. The tracking system traces a user’s real-time location, movement, and head positioning within the courtyard. The tracking system consists of the PhaseSpace Impulse camera system, LED markers, and camera server. The coordination and communication between these three elements allow the VR system to track the user positions immediately and accurately.

In the center of the courtyard, a truss is installed which serves as a framework for mounting the camera system. The camera system consists of 36 PhaseSpace camera modules placed in groups of three at 12 nodes of the truss. Each module contains a binocular pair of cameras and a processor. The cameras capture images of movement at the rate of 960 frames per second. The processor analyzes the images to find LED markers and determines their location in the courtyard. The LED markers are attached to head-mounted displays (HMDs). Each user wears a HMD with LED markers that emit unique light signals visible to the camera system. The camera server, the last part of the tracking system, process the data collected from each camera module to locate the actual positions of the participants within a 1-mm accuracy.

The graphic system translates the user location and head position data and creates 3D visual scenes that correspond to the user’s point of view. Unlike previous trackable HMDs that required heavy equipment on the users, this HMD uses tablets and cellphones, which makes it cost-effective and lightweight. The devices were loaded with a custom app that retrieves the location data from the camera server via Wifi and renders 3D visual scenes in stereo. In the testing, four Apple iPad mini tablets were used.

Using VR technology in creating virtual apparel stores

One of the purposes of this study is to evaluate the feasibility of using VR apparel stores. In the following section, challenges and advantages of using VR technology for virtual apparel stores are discussed to examine practical benefits and costs of VR apparel store creation.

Experience of VR users

Another goal of the study was to understand user experience in the VR apparel stores. Examining experience of the VR users is important for several reasons. First of all, the users must perceive the VR stores as realistic and engaging for VR stores to be successfully implemented. Secondly, it was of interest of the researchers to see if the users feel positive (or negative) emotions and form attitude and intention in response to the VR stores. Lastly, the severity of the negative user experience factor, simulator sickness, needed to be examined and considered. Therefore, the users’ perception of the environment, emotional states, their attitudinal and behavioral responses to the environment, and felt simulator sickness after the session were evaluated. These experiences were measured through the self-administered post-session survey. Prior to the analysis, it was confirmed that there was no significant mean difference in the measured variables among the four VR stores. Therefore, the findings are consistent in all four models.

The participants generally reported high telepresence (M = 5.10, SD = 1.14) and perceived realism (M = 4.76, SD = 1.09), suggesting the overall positive evaluation of these two measures. Participants’ exploratory tendency did not affect their telepresence or perceived realism scores. The results imply that the virtual stores provided participants with an immersive experience that can be comparable to real store experiences.

In order to examine how participants’ immersive experience with the virtual store influenced shopping outcomes (i.e., pleasure, arousal, attitude toward virtual stores, purchase intention, time spent), bivariate correlation analyses were conducted. The results showed a significant positive correlation between telepresence and perceived realism (r = .49, p < .01), pleasure (r = .383, p < .05), attitude (r = .356, p < .05), and purchase intention (r = .47, p < .01). These findings are consistent with previous research that telepresence of virtual apparel stores positively influences consumers’ shopping outcomes such as enjoyment and purchase intention (Song et al. 2007).

Inconsistent with previous research that reported positive relationships between telepresence and arousal (Baumgartner et al. 2008; Lombard et al. 2000), telepresence was not significantly correlated with arousal (r = .21, p = .21) or time spent in the store (r = − .02, p = .90). This conflicting finding may be due to the different experiment designs and how arousal was measured. Baumgartner et al. (2008) used continuous EEG and psychophysiological measures (i.e., skin conductance response, heart rate) to detect arousal while the current study relied on a self-reported questionnaire to measure arousal. The reported arousal level at the end of VR store experience could be different from the real-time fluctuation of arousal levels during their experience. Alternatively, the scenes created by Baumgartner et al. might not have induced as high arousal. Considering how arousal has an inverted-U shaped relationship with many outcome variables, a very high level of arousal induced by VR experience may not have been detected through bivariate correlation tests. Additional analysis revealed that there was a significant quadratic relationship between pleasure and arousal (R² = .47, F(2,35) = 15.70, p < .001).

The participants experienced notable symptoms of simulator sickness. While only one participant out of forty felt sick enough to terminate the session, simulator sickness affected a good portion of participants. Almost 40% of the participants felt some level of discomfort, 35% of them experienced dizziness, 33% of them experienced eyestrain, and 30% experienced nausea. This is consistent with other studies that the major simulator sickness includes general discomfort (Jinjakam and Hamamoto 2011) and eyestrain and nausea (Ruddle 2004). This implies researchers and retailers must use caution when implementing VR in their business or studies and plan for necessary breaks especially when the users are exposed to the environment for an elongated period.