Sketching in-vehicle ambient lighting in virtual reality with the Wizard-of-Oz method

ABSTRACT

Designers desire to sketch their concepts and produce prototypes in a real-time setting. This study developed a virtual reality (VR) supported lighting sketch tool for vehicle interior designers. The tool incorporated lighting design with drag-and-drop functionality and high-definition vehicle rendering. The tool also used the Wizard-of-Oz method to help design participants easily immerse themselves into the design tasks. Six designers participated in two consecutive hour-long design workshops to evaluate the design tool. Each designer sketched eight scenarios, and the workshop gathered 48 design outcomes. In addition, a thirty-minute post-interview evaluated usability, the immersive experience of the virtual environment, and the expected role of the wizard. Finally, thematic coding was used on both workshop and post-interview verbal records to further evaluate the tool's compatibility. As a result, three vital roles of the tool were identified. First, the tool encouraged the iteration process. The tool also gave realistic design-practice experiences and would be suitable for sketching lighting in the workplace. Finally, the tool allowed designers to immerse themselves in the design space. This research anticipates a well-illustrated virtual context and situations that invite designers into an inspirational lighting design workplace.

KEYWORDS:

• Car interior design
• lighting design
• virtual reality
• virtual sketching

1. Introduction

Virtual reality (VR) has advanced, expanding its application in creative environments, especially active investigations in the development process (Ottosson 2002; Wilson and D'Cruz 2006; Choi, Jung, and Noh 2015). It serves various purposes with compatible applications and contents by providing a realistic appearance of its simulated environment. As a result, designers and architects have increasingly adopted VR as a working environment for prototyping and ideation processes (Ye et al. 2006; Bordegoni and Ferrise 2013). VR tools have reduced the time and costs spent on prototype models (Chen, Cui, and Hao 2019). The automobile car industry has actively adopted VR to encourage designers' creativity by presenting full-scale prototype models with realistic VR simulations (Berg and Vance 2017; Hyundai-Motor-Group 2021; Audi 2021).

VR allows designers to situate vehicles outside the studio boundaries in various user environments (Boffi et al. 2019; Ann et al. 2017). It allows designers to efficiently apply various colours, materials, and product finishing. It also supports complex lighting components alongside the display-oriented human-machine interfaces inside the vehicle (Reuding and Meil 2004). It provides a more immersive experience with the support of head-mounted displays (HMDs) that overextend the field-of-view (FOV) of the photo-realistic vehicle renderings (Zimmermann 2008). This expansion of FOV provides room for creativity in searching for ideas for in-car lighting in its roles of indicating information, decorating styles, or supporting an atmosphere rather than simply illuminating the space (Löcken, Heuten, and Boll 2016).

Professionals in architecture and design have actively investigated and developed VR as a lighting sketch tool (Rockcastle et al. 2021; Krupiński 2020; Jelvard and Mullins 2019; Wong et al. 2019; Natephra et al. 2017). A similar approach in tool development has been tried in the automotive sector. For example, Ekstromer and his research colleagues investigated a VR-based lighting sketch tool for car exteriors to allow designers to sketch while simultaneously prototyping exterior vehicle lighting (Ekströmer et al. 2019). However, that tool was only suitable for designing exterior lighting and required better interior lighting implementation. The implementation of an interior lighting VR system requires more optical knowledge as the brightness range of the passenger car varies in its luminosity from 3 lux in the dark to more than 100 lux on the brightest sunny day (Phipps-Nelson et al. 2009).

The study examines and analyzes ways to implement VR for automobile interior ambient lighting design. First, the VR light design tool created in the study is designed after a thorough review of literature that utilized VR lighting design. Then, a design workshop with the carefully crafted design sketch tool is conducted where experienced designers are invited to perform interior lighting design activities. Finally, the design sketch tool and areas of improvement are evaluated in a post-interview after the design workshop.

2. Related works

2.1. Designing ambient lighting in the car

Today's drivers are assisted by intelligent driving systems such as advanced driver-assistance systems (ADAS). As a result, there is a growing expectation of lighting to support people's moods while in the vehicle (Caberletti et al. 2010; Taesu et al. 2021). Kim et al. studied the emotional responses to in-car light provided in the driver's cockpit (Taesu et al. 2021). The research investigates how light attributes include static properties like colours and distributions and behavioural characteristics such as lighting signatures. Hence, car manufacturers introduce stylish ambient interior lighting signatures with diverse patterns and colours (Stylidis, Wickman, and Söderberg 2020). Furthermore, in-car lighting offers new opportunities by portraying a futuristic image of the car for advertising. For example, media coverage of the recent Mercedes Benz vehicles, such as the new S Class presented in 2021, highlighted their futuristic interior lighting design (Mercedes-Benz 2021).

With the increased importance of in-vehicle lighting, the research explores design tools to aid designers in their practices. However, existing lighting design tools require professional knowledge in programming and electrical engineering to properly control and manage (Adafruit 2021; Nooelec 2021). Therefore, researchers investigate lighting prototypes and designing designer-friendly tools to control the light and manage the light sources. For example, Jeong et al. introduced a designer-friendly tool called ‘c. light:’ a Bluetooth-connected set of manipulatable LEDs and individual light source manipulation with a designer-friendly lighting attribution control interface (Jeong et al. 2018). As an alternative to custom-made solutions, designers also adopt professional solutions such as ‘Hue’ from Philips or ‘Yeelight’ from Xiaomi that require technical competence for usability (Philips 2021; Xiaomi 2021). The abovementioned tools and related research pointed out that a lighting sketch tool requires easy-to-use usability and thorough instruction about manipulating the tool to allow designers to concentrate on sketching lighting in the given environment while designing.

2.2. Lighting design supported by VR

As VR becomes a working space for lighting design, research has explored methods and design capability. For example, Jelvard and Neaephra suggested a VR-based lighting design method to install lighting components quickly. Users can articulate the attributes using simple graphical user interfaces (GUIs) to virtually prototype the lighting environment of a building (Jelvard and Mullins 2019; Natephra et al. 2017). Some researchers have focussed more on the interactive aspect with the moderator while designing with VR. For example, Wong adopted the Wizard-of-Oz method to make a system that conducts real-time verbal communication between a VR user and its operator to control lighting attributes (Wong et al. 2019). Current studies on lighting control in VR are limited and simple, focussing on intuitive interaction. Hyedarian and his colleagues pointed out that a more intuitive and straightforward system can provide a more immersive VR simulation environment (Heydarian et al. 2015).

Many researchers have developed indoor lighting VR design tools that imitate a realistic environment and then evaluated the tools and methods utilizing them (Rockcastle et al. 2021; Chamilothori, Wienold, and Andersen 2019). Previous studies referenced actual illuminant and object luminance measurements using a colour metre and a spectrophotometer. As a result, they achieved a more realistic environment in VR, and most spectators reported that the perceived presence of the virtual environment is close to the real-life environment. Accordingly, studies suggested that borrowing cognitive characteristics of the target environment in designing virtual environments alongside quantitative information would bring more realism into the created space. Krupinski used those insights in his study by proposing a rendering algorithm for lighting in virtual environments that is easily controlled based on users' cognitive feedback (Krupiński 2020).

This study aims to create a VR-supported ambient lighting sketch tool tailored to car interior design by implementing precedent studies and insights. The study highlights two aspects from the precedent studies: intuitive usability and cognitive interactivity. The VR sketch tool created for this study incorporates a drag-and-drop technique for lighting components to provide more intuitive usability. It also uses the Wizard-of-Oz method to control lighting elements and manipulate them to derive plausible design outcomes. In addition, an actual vehicle interior and its ambient lighting elements are presented with virtual environments that reflect lux levels of actual in-vehicle lighting: 5 lux in the dark and almost invisible during daylight conditions (Caberletti et al. 2010). Finally, designers are invited to immersive and interactive design spaces to freely sketch the vehicle's ambient lighting with suggested settings determined in the study.

3. Designing in-vehicle lighting using VR

3.1. VR facility

The study utilized digital contents of car models, environmental scenes, and lighting modules for a VR workshop on designing in-car lighting. The 3D modelling of Tesla ‘model 3 2018’ was purchased from hum3d.com (Hum3D 2021). The scenes used in the study involve both static 360-degree pictures and 360-degree videos to provide better simulation. Light modules provided in the workshop are straight-line shapes for linear modules and spheres for point modules in various colours and sizes. The drag-and-drop function in Unity 2019 for Windows was used for the lighting modules (Unity 2021b). All the modules were placed together in the VR scene setup. Finally, the study used SteamVR (Unity 2021a) to allow an operator to switch between two different perspectives simultaneously: one for duplicating the participant's viewpoint, another to observe the participant's placement. Figure 1 illustrates the two concurrent and different viewing perspectives.

Figure 1. The workshop setup of the VR-supported system for designing in-vehicle lighting. The participant designed the lighting contents wearing the ‘HTC VIVE’ headset, while the wizard observed simultaneously watching the identical view generated through the Steam VR with a GTX1080 graphics card. At the same time, the wizard used two other windows to operate the Unity VR contents-and-control interface.

3.2. Lighting inventory

The study provided a set of basic lighting modules: points, lines, and bulbs with colour variations, as presented in Figure 2. All participants adopted drag-and-drop manipulation of lighting modules relatively quickly. Technically, the ‘Throwable’ script was applied to implement this interaction in Unity. In addition, the workshop utilized ‘High-Definition Render Pipeline’ in Unity to quickly render the quality of light modules to render the lit circumstance as realistically as possible. The technical procedure is illustrated in Figure 2 on the right. Creators know the rendering platforms of the automotive industry, VRED (Autodesk 2021) or Vray (Chaos 2021), for modelling lighting components. However, despite their high-quality renders, these platforms might not be optimal for a real-time-based rendering suitable for an academic workshop. These platforms require high-powered computers and take a long time to render. Therefore, Unity was used in the workshop to achieve real-time manipulation and sufficient rendering quality.

Figure 2. (A) The default basic light inventory, (B) Point light modules, (C) Line light modules, (D) Bulb light modules, (E) Schematic illustration of making a light module for the VR workshop).

3.3. Manipulation applying Wizard-of-Oz method

The study adopted the Wizard-of-Oz method in moderating the workshop. The ‘wizard’ of the Wizard-of-Oz method is an operator who rapidly realizes the participant's requests. By convention, the wizard holds a sufficient level of expertise so that they may proactively process the participant's request and adequately respond to it (Maulsby, Greenberg, and Mander 1993; Dahlbäck, Jönsson, and Ahrenberg 1993). For example, An and his research team proposed a sketch-based VR environment, where the wizard transformed participant's paper prototypes into VR contents in real time (Ann et al. 2017). In the study done by Dow et al., the wizard offered a set of interfaces deliberately selected to be minimal (Dow et al. 2005). The wizard manually excluded less relevant interfaces during the VR workshop. Also, Wong advocated the Wizard-of-Oz method as an excellent compromise to the imperfect technical stage of VR (Wong et al. 2019). The study employed the Wizard-of-Oz to operate the workshop with an expert-supported co-design method.

As a rule, assigned designers with expertise in light and vehicle design act as the wizard. Therefore, the wizard's role is to guide the designer's decision-making regarding lighting features and their effect. Also, the wizard was fluent in operating Unity 2019 to respond to participants' requests in real time. The wizard also provided technical help and expert knowledge. An exemplary protocol follows:

participant: ‘(Indicating the foot area) Can I have here some more point lighting in orange colour? Well would it be too much?’

wizard: ‘Sure, would three point lighting be enough? Usually orange nuanced light induces relaxing atmosphere.’

Furthermore, the wizard tried to respond to the participants' requests about issues other than light configurations. For example, some participants wished to place small accessories, such as cups or electronic gadgets, near themselves. Alternatively, they asked to modify the shapes or colours of the interior items. Finally, some wanted to reconfigure the car components by opening the door or uncovering the sun visor.

3.4. Contextual scenarios to conceptualize in-car lighting

Designers are exposed to diverse driving situations to stimulate them to develop a novel conceptualization of the lighting design. The composition of the given situations is based on three factors: driving mode (two types: autonomous driving or not), driver's emotional states (four types: angry, sad, excited, and neutral), and external environment (six situations). The three factors were independent of each other while directly influential on lighting contents. Instead of providing all possible scenarios by combining three factors, the study purposefully designed eight scenarios by coupling two driving modes and four emotion types. Such a setting allowed participants to choose an appropriate environment from the six situations. As shown in Figure 3, diverse contexts were considered so that the participant designers might have sufficient alternatives to support the eight scenarios.

Figure 3. The eight environments used in the tool.

4. Design workshop

4.1. Participants

The workshop intends to ask the participants to compare the VR context with conventional design procedures. Recruited designers have a design education background and more than 2 years of automobile industry experience. The detail information for designers described in Table 1. As a result, six designers (five men and one woman, mean age $=27.5$ years, standard deviation $=3.45$ years) volunteered for the workshop. All confirmed no VR motion sickness for wearing a VR headset for around an hour. In addition, they were rewarded with $60 for participation.

Table 1. Profile of the car designers in the study.

Participant	Work experience(years)	Expertise
P1	6	Exterior styling
P2	2	User experience
P3	2	Human machine interface
P4	2	User experience
P5	3	Exterior styling
P6	2	Human machine interface

4.2. Procedure

The workshop session involved one participant and a wizard as a team. Each workshop was planned for an hour to avoid any physiological or psychological discomfort caused by VR. Each participant was assigned two workshops within a week. Then, each participant produced eight lighting design sketches (four in the first workshop and the second four in the second workshop) in total (Figure 4).

Figure 4. Overview of the workshop procedure.

The first-time workshop participants were given tutorials on operating the tools for about 10 min before the workshop session. Then, they were offered four scenarios of the eight: two driving modes and four emotion types to minimize the sequential effect. These scenarios were randomly presented to the participant. The participant selected a suitable environment from the six scenery options for each scenario (Figure 3). In the second round of the workshop, participants designed the remaining four scenarios.

A self-assessment and a semi-structured interview were conducted after the workshop. The participants were asked to fill out short subjective questionnaires for about 5 min in this session. Based on Jeong et al. (2018), and Ann et al. (2017), six questions (Usability, Learnability, Implementation ability, Iteration ability, Sketches satisfaction, Willing to use the tool in the future) were asked to respond on the 7-point Likert Scale (−3: very poor to +3: very good). The additional interview took about 30 min, and the replies were transcribed. The interview questions were structured based on related works about lighting (Jeong et al. 2018; Noh et al. 2020) and virtual reality (Ekströmer et al. 2019; Wong et al. 2019). As summarized, we focussed on the following concerns:

• Usability evaluation

  • How was locating ambient lighting through a drag-and-drop interface?

  • How were changing attributes of ambient lighting through wizard using voice interaction?

• Immersive experience during the VR workshop

  • Were the system components sufficient for the sketching activity?

  • Was a virtual environment sufficient to engage in the design space?

• Anticipated role of wizard

  • Did the wizard collaborate well with your design implementation progress?

  • Did the wizard collaborate well with your design iteration progress?

• General impression

  • Are you satisfied with the whole design process using this tool?

  • Do you wish to use this tool in your future design work?

4.3. Data collection and analysis method

Because the number of ideas and time consumed in developing ideas is similar among designers, the data were gathered and analyzed on all the design outcomes, the workshop's verbal records, and the post-interview verbal records. Furthermore, the post-interview verbal records were processed for in-depth analysis. Through the thematic coding process, authors individually conducted an initial round of open coding, which provided a general overview of the themes. Then, four researchers reviewed the transcripts again to emerge additional coding in the second round. Rearrangement and reclassification of coded data continued until the definitive emergence of three major themes shown in Table 2.

Table 2. Themes and codes for the final coding scheme.

Theme	Codes
Usability of the tool	Learnability of the tools
	Technical barriers
Realistic virtual environment	Implementable light modules
	Iteratable light attributes
	Similarity with vehicle mock-up
Immerse into the design space	Similarity with an actual driving
	Expectation of music/sound effect

4.4. Results

4.4.1. Design outcome

Forty-eight in-car lighting designs from twelve workshops were gathered (Figure 5). On average, each design contained 5.63 light modules, and the linear modules were the most frequently used. Generally, participants spent between 1 min 40 s and 11 min 25 s to create a light design for a given scenario. Thus, on average, it took 5 min 16 s to complete one lighting design for the car interior using our VR setup. There is a similar preference of lighting design across three emotion categories: ‘excited,’ ‘angry,’ and ‘desperate.’ In ‘neutral’ emotion, the workshop found that ‘point’ light modules were used relatively more frequently than others. The point modules were frequently found below the rear mirror or door pockets to highlight small areas. Also, they were used to deliver information about the driving state. Table 3 presents the frequency of light modules across emotion categories, the shapes of light modules, and levels of autonomous driving.

Figure 5. Lighting sketches from the workshop.

Table 3. Light modules used in designing in-car ambient lighting from 12 workshops.

	Car without autonomous mode			Car with autonomous mode
	Line	Point	Bulb	Line	Point	Bulb
Angry	12	9	5	20	13	7
Sad	21	6	5	23	4	3
Excited	19	9	6	14	6	7
Neutral	26	15	2	22	13	4

Within the same emotional category, the participants showed a similar tendency in designing the lighting. When the scenarios included an excited or angry mood, the light modules were placed within the FOV, near the sun visor or steering wheel area. In the case of desperate scenarios, the participants designed indirect lighting. The light modules were placed outside FOV under the door or foot lamps while the reflection was imminent and visible. The examples are shown in Figure 6.

Figure 6. Lighting designs for sad emotions in an autonomous vehicle context. Indirect lighting was placed outside the field-of-view. The lights are highlighted with yellow circles.

At the same time, each participant showed a consistent style tendency. P1 used built-in lighting focussing on the steering wheel. P2 used point modules at small areas, such as a door pocket or cup holder. P5 placed the light modules at ventilation components to create a moody atmosphere across all eight scenarios. The examples are presented in Figure 7.

Figure 7. Lighting designs that reflect personality. For these examples, the special focuses were the steering wheel (P1), small pockets (P2), and ventilation (P5).

4.4.2. Finding 1: ‘VR was easy and quick, but the wizard was thankful’

All participants agreed on the intuitive use of VR as the benefit, although they were not fluent in it. As shown in Table 4, designers responded that this system was easy to use and valuable in iterating progress while sketching lighting. They also noted that arranging the light modules by the drag & drop tool allowed them to realize by duplicating, arranging instantly, or linking light modules by simply requesting the wizard for assistance. For example, P4 and P5 mentioned that ‘it was convenient and quick to visualize the concept without considering the circuit plan. I just placed the light module to brighten the area. Then, call out, “light here!”’ Also, participants often compared the VR setup with physical mock-ups. Attaching a light module acquires several technical concerns, whereas the VR setup ignores those issues. Participants entirely focussed on shaping and styling the in-car interior lighting using unlimited light modules.

Table 4. Feedback criteria for the VR-supported ambient light sketching tool ($-3$: very poor, 3: very good).

Criteria	Mean (Standard deviation)
Usability	2.00 (0.63)
Learnability	2.17 (0.75)
Implementation ability	1.67 (1.21)
Iteration ability	2.17 (1.17)
Sketch satisfaction	1.67 (0.82)
Willingness to use the tool in the future	2.17 (0.75)

Moreover, they often placed light modules inside or beyond the virtual objects, such as seats and clusters, which is difficult to experiment with in a physical mock-up setting. These activities imply that the VR setup provided a more extensive design space for designers to explore new design ideas almost without technical barriers. The design procedure was an even quicker and more satisfying experience with the wizard's presence. The wizard speculated the outcomes after receiving the request from each participant. Accumulated experience resulted in a better performance.

4.4.3. Finding 2: ‘VR was realistic, but the wizard was magic’

Participants promptly tested the lighting effect in real time and immediately made incremental improvements with the wizard's support.

During the workshop, 270 light modules were created, and 186 modules were finalized after several modifications. We found that designers with professional experience showed difficulty verbally expressing the colour-related properties of lighting modules. Most colour terms and attributes refer to the object's colour or use essential colour names, such as brown or gray. Alternatively, some participants requested abstract colours, such as Aurora sky or ocean. Translating those concepts into light properties was difficult for the wizard.

At the same time, the participants found it remarkable that the wizard magically visualized their abstract concept through the requested light modules. P2 described the wizard as a magician who made her vague concept come true. P2 said, ‘In that virtual world, I should be alone. However, I did not feel alone at all. The wizard was supportive and so clever, reading my mind.’ Similarly, P5 commented, ‘The VR setup enabled me to explore design variations quite efficiently. But the wizard's support was more critical in this case.’

Like the design process with a physical mock-up, the participants evaluated the design from various viewpoints. Figure 8 presents the in-car light design observed from outside of the vehicle. Depending on the scene, participants' visual perception of the lighting design was highly influenced by the surroundings. P4, P5, and P6 validated the VR setup as a more powerful visualization tool than sketches or static rendering images in both 2D and 3D. Five of six participants reported that the VR workshop was comparable to the lighting design in practice in an actual car interior by showing a similarity of around 90%. In particular, P4 agreed on the similar impact of the 1:1 scale clay mock-up, based on his experience. The immersive perception was vital in the VR scene with static scenery, although design activity with moving scenery provided a unique experience.

Figure 8. Realistic aspects of VR in designing in-vehicle lighting.

4.4.4. Finding 3: ‘VR triggers empathy development’

Despite the imperfect visualization provided in the workshop, the realistic impression was entirely satisfactory, which resulted in participants' developing empathy with the provided scenarios. During the workshop, participants were instructed about eight scenarios composed of four types of emotion mapped with two levels of autonomous driving situations. The four emotion types were excitement, anger, despair, and neutral, and we intended to make the participants consider diverse emotions in various driving scenarios. However, participants tended to find themselves immediately immersed in presented emotions and scenarios when they have had a similar experience in driving in the past. For example, P3 said, ‘I often go for a drive to recover from the depressive mood. This made me get immersed into the despair scenario easily.’ Also, solid emotional engagement led to better empathy development for designing the lighting design of the car interior. P3 and P6 also requested sentimental background music to enhance the mood, especially with the despair emotion.

Moreover, empathy was additionally enhanced when the matching scene was provided during the workshop. Participants chose the most suitable environment among six variations. Participants agreed that the selected environment played a critical role in empathetic immersion. Regarding the despair emotion type, five of twelve lighting designs selected the ‘Night in Rural’ high-dynamic-range (HDR) image (see Figure 3), which illustrates calm but dark scenery in a rural area. The study found a benefit of deliberately combining the virtual object with the proper virtual environment. Depending on the match, the emotional experience can become more immersive, and thus the empathy is efficiently developed. A thorough match with the surrounding environment is critical when an emotional immersion is relevant.

5. Discussion

This paper devised a VR setup for designing in-vehicle ambient lighting and conducted workshops with six experienced car designers. All participants of the workshop have professional industry experience. The study involved a drag-and-drop method of manipulating virtual light modules during the lighting design process. The study gathered feedback from these professionals on VR light designing setups compared to conventional approaches for designing in-vehicle lighting.

5.1. VR as a lighting sketch workplace

Only a handful of methods are offered for sketching and prototyping in-vehicle lighting, even though interest in the ambient lighting in the car interior is increasing with current vehicles. Further, designers are faced with physical barriers in prototyping as the lighting prototype requires technical knowledge about circuits and optics. Therefore, they have utilized either light prototyping tools designed explicitly for designers or adjusted manufactured products in their light design ideation process. Designers reflected in the post-interview that the simple interface of the lighting sketching process offered in the workshop reduced such technical barriers in the designing process. They also reflected that those light modules provided in the workshop also alleviate technical difficulties. As shown in Table 4, the self-assessment found that the tool proposed in this study is beneficial for iterating the lighting design of the in-vehicle environment.

Additionally, designers responded that the drag-and-drop interface allowed them to manipulate light modules easily and provided control in positioning light modules without any hassle about optical design. The number of light modules used in the lighting design workshop gives designers more freedom in design practices. Designers used an average of 5.63 light modules in one sketch workshop and 22.52 light modules in total in a workshop that took about an hour. This number is similar to the number of light modules used in an earlier study, where 19.5 light modules were used in an hour-long workshop (Jeong et al. 2018). This similarity in light modules used on average reflects that the proposed tool has resolved difficulties during the lighting sketching and design practices.

5.2. VR as an immersive design tool

VR-supported lighting sketches are studied primarily in architecture as creating physical mock-ups can be expensive. However, these studies highlighted that the advantage of using VR is to provide an immersive experience with a wide field of view (FOV) while resolving the cost spent on developing the physical models that are limited in providing an immersive design experience. Consequently, this study integrated VR into the in-vehicle ambient lighting design context. Designers reported that sketching with the VR lighting sketching tool provided an immersive experience during the design process. In addition, they pointed out that designing and sketching lighting in VR with a driving context such as autonomous driving environments allowed them to be more imaginative and creative while designing. Such immersion is reflected in the number of total light modules used in the workshops, as shown in Table 3: autonomous situations used 136 modules, and non-autonomous used 135 modules.

Furthermore, the VR environment makes it easy to recall the emotional experience of the driving context. The lighting sketches from the workshops indicated coherence with previous in-car ambient lighting research results. Usage of point light modules, 3.50 modules on average, increased in the neutral emotion scenario compared to other emotion scenarios that used an average of 1.96 modules. Designers expressed the role of point light in the neutral mode for conveying information. It is coherent with Harrison et al.'s study that defined the role of point light in the product (Harrison et al. 2012). In addition, designers utilized more lighting modules in the angry emotion scenario of autonomous driving (40 modules on average) than in a non-autonomous context (24 modules). Designers asserted that more light modules and their active response to the given context help calm the driver and passenger. This finding is in line with Zepf et al., who described that affective solutions to managing negative emotions like anger are likely to be increased in the autonomous driving context (Zepf et al. 2020).

5.3. Wizard as lighting sketch assistant

One researcher was assigned to act as an operator as a ‘wizard’ during the workshop. The wizard of the Wizard-of-Oz method allows our presented VR setup to be fully intelligent and autonomous (Maulsby, Greenberg, and Mander 1993; Dahlbäck, Jönsson, and Ahrenberg 1993). In particular, by having a sufficient level of expertise in lighting and product design, the wizard could respond to participants' requests during the workshop in real time. Indeed, all participants found the wizard was extremely valuable. For example, at the beginning of the workshop, participants tended to describe their lighting concepts with metaphors, pattern shapes, or irrelevant colour terms. Then, the wizard interpreted these vague expressions or unrealistic concepts into more precise and feasible concepts. In particular, the wizard translated the initial unclear concepts into manageable lighting properties. Like previous studies using the Wizard-of-Oz method to manipulate lighting, participants found the support from the wizard remarkable (Olsen and Spaulding 2017; Magielse and Ross 2011).

5.4. Limitations and future work

This study considered implementing the VR setup for different purposes in design development. It mainly focussed on designers' exploration of lighting design using VR; however, the developed design results contribute valuable concepts and ideas to be evaluated in further design process steps. The study has not yet incorporated car interior materials and colours, even though the light directly influences the perceptions of the material quality in the rendering. This area will be visited with further implementation of the work. Moreover, the archived lighting distribution and configuration should be easily implemented into different car interior models. Designers can promptly reproduce and replicate the lighting design for the different cars as importing car models is done digitally. In this regard, future research will investigate the value and role of VR-based lighting design for the evaluation and agreement phases.

The wizard's presence was crucial in the workshop; however, the significant contribution of the wizard was a limitation of the VR-based design exploration, especially when domain-specific knowledge is essential, as shown in previous studies (Ann et al. 2017; Lehner and DeFanti 1997). The interview results indicate that VR-based creative practice will be greatly appreciated if an intelligent agent, like the wizard of our workshop, is provided when designing. We found that the quality of design outcomes was in proportion to the wizard's technical assistance and knowledgeable advice. Such dependency both addresses current limitations and presents a future opportunity. The advancement of intuitive interface and user experience (UX) designs will be incremental in the short term. Meaningful improvement will provide a more fertile environment for designers. The progress will be even more remarkable when a VR system is developed with a wizard-like intelligent system. Therefore, further studies are expected toward an intuitive and intelligent wizard who connects people and VR tools.

6. Conclusion

The study demonstrated a VR-based lighting design system and conducted workshops with six experienced designers. We devised a drag-and-drop tool to assist the participants in placing light modules in real time. The workshop focussed on designing in-car interior lighting, and eight scenarios were given. We adopted a Wizard-of-Oz co-design system for the workshop that involved participants and a knowledgeable operator. Based on our observation and interview answers, we unfolded three significant findings. First, the lighting design activity in VR was quick, and the quick visualization from verbal instructions satisfied the participants. The support from the wizard was essential for this. Second, the wizard was able to interpret the participants' personal tastes and intentions. Third, VR offered extensive flexibility, free from technical limitations, but the wizard was critical to obtaining satisfactory results. Last, the immersed visual experience led to empathy development that quickly affected their designs, especially when participants faced familiar emotional contexts. The empathetic immersion was more enhanced when the environmental scenery was appropriate. The design results depended on the wizard's technical fluency and domain-specific expertise. Future studies are expected to expand the application context to discover more value and potential for VR-based lighting design.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was supported by the National Research Foundation of Korea (NRF) funded by the Ministry of Education (MOE) (Nos. 4120200913638; 2018R1A1A3A04078934)

Notes on contributors

Taesu Kim

Taesu Kim received a B.S. in Industrial Design from Korea Advanced Institute of Science and Technology (KAIST), South Korea, in 2018, where he is currently pursuing Ph.D. His current research interest includes light design and emotion design focusing on automobiles.

Aigerim Shunayeva

Aigerim Shunayeva is a Master's student at Color Lab of Department of Industrial Design at KAIST. She received her BS in Computer Science from L.N. Gumilyov Eurasian National University in 2012. Her interest in using Virtual Reality for design processes and design research.

Gyunpyo Lee

Gyunpyo Lee is a Master's student at the Color Lab of the Industrial Design department at KAIST. He received his Bachelor of Fine Arts in Transportation Design from College for Creative Studies in 2015. He has experience as an automotive exterior designer at Changan European Design Center for about four years. His research interest lies in the visual semantic evaluation of both designers and users, aesthetic quality and cognitive evaluation of products, and emotional aspects of how people perceive forms pleasurable.

Hyeon-Jeong Suk

Hyeon-Jeong Suk is an associate professor at KAIST. She received BS and MS degrees in Industrial Design from KAIST and received a PhD degree in social science major in psychology from University of Mannheim in Germany. Her research group has professionally conveyed color research and design practice by intensively collaborating with diverse companies. Currently, she acts as an Editor-in-Chief of the Korea Society for Color Studies and a Dean of KAIST academy.