Professional Documents
Culture Documents
Designing For Augmented Reality in Aviation E-Commerce: MEÅ Niversity
Designing For Augmented Reality in Aviation E-Commerce: MEÅ Niversity
M ASTER T HESIS
Author: Examiner:
Alexander Gunillasson Thomas Mejtoft
July 1, 2020
i
UMEÅ UNIVERSITY
Abstract
Designing for Augmented Reality in aviation E-Commerce
by Alexander Gunillasson
In-flight entertainment (IFE) is facing exciting times ahead as more and more tech-
nological advancements find its way into the IFE-systems. With the increasing hard-
ware capabilities of IFE-systems, one exciting technology which could be seen in
future IFE-systems is augmented reality (AR). AR could allow for virtual try-on
(VTO) in a webshop, where passengers can view virtual representations of for in-
stance, glasses or headpieces, augmented onto themselves through a camera on the
IFE-screen. However, integrating VTO in a webshop comes with new challenges,
both for the designer and the developer. This thesis has examined design guidelines
for AR in a literature study, along with general principles for user experience (UX).
Based on the findings from the literature study, a prototype for a webshop with
integrated VTO, targeting IFE, has been build in an iterative design process. Fur-
thermore, a small application has been built using Google’s AR-framework ARCore
for Android, where users can try different virtual glasses. The literature study, along
with usability testing in each design iteration, showed that there are no guidelines
which can cover everything when designing for AR; however, the design guidelines
presented by Google and Apple are a great place to start. Furthermore, the usability
testing showed that people have certain expectations of a webshop, which indicates
that VTO should be integrated seamlessly into the application without interfering
with the layout. The application development showed that ARCore certainly is a po-
tent tool when it comes to AR-development. However, ARCore is not open source,
which could make it less appealing for companies in need of highly flexible solu-
tions.
ii
UMEÅ UNIVERSITY
Sammanfattning
Designing for Augmented Reality in aviation E-Commerce
av Alexander Gunillasson
In-flight entertainment (IFE) står inför spännande tider framöver. Den snabba tekn-
iska utvecklingen öppnar upp många framtida möjligheter för IFE-systemen. Med
ökad hårdvarukapacitet är augmented reality (AR) en teknik som skulle kunna an-
vändas i framtida IFE-system. AR kan möjliggöra virtual try-on (VTO) i en webb-
shop, där passagerare kan se virtuella representationer av till exempel glasögon eller
huvudbonader, projicierade på sig själva genom en kamera på IFE-skärmen. Att in-
tegrera VTO i en webbshop medför dock nya utmaningar, både för designern och
utvecklaren. Detta examensarbete har fokuserat på att undersöka designriktlinjer
för AR i en litteraturstudie, tillsammans med allmänna principer för user experience
(UX). Baserat på information och lärdomar från litteraturstudien har en prototyp
för en webbshop med integrerad VTO, riktad mot IFE, byggts i en iterativ design-
process. Dessutom har en mindre applikation byggts med Googles AR-ramverk
ARCore för Android, där användare kan prova olika virtuella glasögon. Litter-
aturstudien, tillsammans med användbarhetstester i varje designiteration, visade
att det inte finns några riktlinjer som kan täcka allt när man designar för AR, däre-
mot ger designriktlinjerna som presenteras av Google och Apple en bra riktlinje
till att börja med. Användbarhetstesterna visade dessutom att människor har vissa
förväntningar på en webbshop, vilket tyder på att att VTO bör integreras friktions-
fritt i applikationen utan att besvära layouten. Applikationsutvecklingen visade att
ARCore är ett kraftfullt verktyg när det gäller AR-utveckling. ARCore är dock inte
open source, vilket kan göra det mindre tilltalande för företag med behov av flexibla
och anpassningsbara lösningar.
iii
Acknowledgements
There are many people who have contributed to this thesis in some way. I would like
to thank all of them for their help and support throughout working on this thesis.
Örjan Strömberg - My external supervisor at Tactel. Thank you for helping me
shape the direction of this thesis and for the overall support and guidance through-
out this thesis.
Ulrik Söderström - My supervisor at Umeå University. Thank you for your guid-
ance and for answering any questions regarding the thesis which arose during the
process.
Magnus Emanuelsson, Martin Sjölund, and Isak Gustafsson - Peer review group.
Thank you for your feedback through the whole process.
Lotta Westin and the employees at Tactel - Special thanks to Lotta Westin for wel-
coming me into Tactel and to all the employees at Tactel for the welcoming and
positive atmosphere.
iv
Contents
Abstract i
Sammanfattning ii
Acknowledgements iii
1 Introduction 1
1.1 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Research questions . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Tactel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Background 4
2.1 In-flight entertainment system . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.2 Future of In-flight entertainment . . . . . . . . . . . . . . . . . . 5
2.2 Augmented Reality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.2 The concept of Virtual Try-on in E-Commerce . . . . . . . . . . 6
3 Theoretical framework 8
3.1 User Experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1.2 User-Centered design . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Designing for Augmented Reality . . . . . . . . . . . . . . . . . . . . . 12
3.2.1 Research and design guidelines . . . . . . . . . . . . . . . . . . . 12
Overview of Google’s and Apple’s AR design guidelines . . . . 13
3.2.2 Google ARCore . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.3 Prototyping for AR . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Using the camera . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4 Method 16
4.1 First iteration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.1.1 Literature study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.1.2 Design proposals . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.1.3 Low-fidelity prototype . . . . . . . . . . . . . . . . . . . . . . . . 17
4.1.4 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Test group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2 Second iteration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.1 Prototyping tools . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.2 Wireframe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.3 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.4 Test group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3 Third iteration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
v
5 Result 20
5.1 Iteration one . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.1.1 Low fidelity Concepts . . . . . . . . . . . . . . . . . . . . . . . . 20
5.1.2 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Tasks for design concept one . . . . . . . . . . . . . . . . . . . . 24
General feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.1.3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.2 Iteration two . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.2.1 Wireframe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.2.2 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
General feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.2.3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.3 Iteration three . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.3.1 Mock-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.3.2 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5.4 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
6 Discussion 37
6.1 Design process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.2 Suitability of testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7 Conclusions 42
7.1 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
vi
List of Figures
2.1 Reality-Virtuality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
List of Tables
List of Abbreviations
AR Augmented Reality
MAR Mobile Augmented Reality
VTO Virtual Try-On
UX User eXperience
UCD User Centered Design
HCI Human Computer Interaction
IFE In-Flight Entertainment
API Application Programming Interface
SDK Software Development Kit
1
Chapter 1
Introduction
Augmented Reality (AR) is a technology which has seen a significant rise in popular-
ity over the last decade. The concept of AR was envisioned as early as 1901 [1]. With
technology evolving and hardware becoming ever more capable, the early envisions
of this technology has now become a reality. One concept which has gained traction
over the last decade is the ability to integrate AR in E-Commerce. E-Commerce, like
AR, continues to rise in popularity and contributes to a significant percentage of the
total revenue for clothing companies. The sales for online shopping of apparel and
accessories have seen a significant increase over the last years and show no signs
of slowing down. Just in the US sales are expected to reach 145 billion by 2023, for
comparison the sales reached 96 billion in 2017 [2]. However, online shopping comes
with one big problem for the consumers, that is the consumer’s inability to view a
fashion item on themselves before buying the product, this is often referred to as the
suit, fit and match dilemma [3]. Virtual Try-on (VTO) tries to address this problem
by allowing consumers to virtually try-on fashion items before making a purchase
decision. It has been found that applying AR in an online shopping context can in-
crease the shopping experience and entertainment value, and it can also influence
the brand attitude [4]. Moreover, with research suggesting that AR and VTO can in-
crease the consumer’s willingness to buy a product [5], a strong case could be made
for companies applying this technology to their online shops.
E-Commerce is now starting to appear across multiple industries, since the poten-
tial of increase in revenue it provides. The aviation industry is one area where on-
line shopping is getting more attention. The In-flight entertainment systems (IFE)
which are present on most airline’s longer flights, have become increasingly more
advanced since the introduction, allowing for more entertainment sources to the
passengers. Airlines have realized the value in e-commerce, seeing how popular it
is on the ground. However, while e-commerce exists on many airlines today and to
some, is a valuable source of revenue, it still underperforms [6]. Connectivity could
be a game-changer in changing this, as well as personalized shopping [7, 6]. Further-
more, with the ever-increasing capabilities of hardware, AR is a technology which
could find its way into IFE.
With increasing use cases of AR-technology, researchers try to keep up with the new
unique challenges it provides [8]. One area that is currently getting attention is the
Chapter 1. Introduction 2
1.1 Objective
This Master Thesis will examine design principles for augmented reality and build
a prototype targeting virtual try-on for a webshop within an in-flight entertainment
system. Furthermore, the objective is to implement a simple use case, where a user
can test sunglasses by utilizing Google’s AR-framework ARCore.
1.2 Tactel
This master thesis is done in collaboration with Tactel3 . Tactel is a digital interaction
agency. The company is based in Sweden, was founded in 1995 and became part of
Panasonic Avionics4 in 2015. Tactel creates digital solutions for their clients around
the world. The office based in Umeå mainly focuses on developing entertainment
systems for clients within the aviation industry.
3 Tactel: https://tactel.se/sv/
4 Panasonic: https://www.panasonic.aero/
4
Chapter 2
Background
Before we dive into User Experience (UX) and Augmented Reality (AR) and all its
different aspects, we first need to give some context to the topic of this thesis. Firstly
we need to look at in-flight entertainment systems (IFE-systems) to get a better un-
derstanding of what it is and what capabilities exist today. However, there are lots of
different IFE-systems, but since this thesis is done in collaboration with Tactel, which
is a part of Panasonic Avionics. The main focus will be on their system, particu-
larly their next-generation IFE-system, which comes with new features and updated
hardware. The reason for this is because, in theory, this system could be capable
of running AR-applications. Having more capable hardware has led to the idea of
integrating Virtual try-on (VTO) in the webshop.
2.1.1 History
In-flight entertainment systems have seen a rapid growth since the introduction.
The first appearance of in-flight entertainment dates back as early as the year of
1921, where the first instance of in-flight movies was recorded, according to White
[16]. However, the exact date of when in-flight entertainment was introduced in air-
planes is a controversial topic, and subject to interpretation [16, 17]. White states
that “entertainment” was not necessarily the motivation behind the early adoptions,
he further describes that “the term "inflight media event" is a much more accurate
label to describe the purpose of the first onboard entertainment efforts.” [16]. The
evolvement of theses systems continued, specifically targeting passengers being able
to watch movies. In the 1960s, the phenomenon started to become well recognized,
and in-flight movies were becoming an established industry, this was still long be-
fore the notion of a IFE-system. Instead, the acronym “IMP” (Inflight Motion Pic-
tures) started to gain traction and was frequently used [16]. However, it was not
until the year of 1988, where modern IFE-systems started to form. During this pe-
riod, the concept of having a personal screen was introduced. This medium was
first introduced by integrating a 2.7" LCD-screen in the back of every first-class seat,
entertainment on an individual level had arrived [15]. Since then, as hardware has
Chapter 2. Background 5
become more capable, this concept of having personal entertainment has continued
to progress rapidly.
2.2.1 Definition
A famous early definition came in 1994 by Milgram and Kishino in their Reality-
Virtuality (RV) Continuum [20], where they describe a Mixed Reality (MR) rang-
ing from an environment consisting of real-world objects to a Virtual Environment
consisting of virtual objects (see Figure 1). As shown, AR can be viewed as an “in-
termediar” between virtual reality and the real world. However, a couple of years
later in 1997, a more thorough definition came by Ronald T. Azuma, who defined
it as: “Augmented reality (AR) is a variation of virtual environments (VE), or Vir-
tual Reality as it is more commonly called. VE technologies completely immerse a
user inside a synthetic environment. While immersed, the user cannot see the real
world around him. In contrast, AR allows the user to see the real world, with virtual
objects super-imposed upon or composited with the real world.” [21].
Chapter 2. Background 6
the framework are a set of design guidelines for designing AR-applications. Those
guidelines combined with previous research will function as a baseline with regards
to prototyping a webshop with VTO-features. In order to evaluate design choices,
an iterative design process will be adopted wherein each iteration a prototype will
be created and tested.
8
Chapter 3
Theoretical framework
The theoretical framework in this thesis aims to assist in answering the research
questions. Furthermore, help provide a better understanding of key concepts and
existing research within the context of this thesis.
on forming users first impression of a website [32]. It was found that 50ms is all
it takes for users to form a first impression on a website [32]. However, the study
mentions that precisely what this first impression entails needs further research [32],
but it’s still essential for the designer to keep this in mind. A couple of years later,
Google confirmed this 50ms result in conducting their study in the context of first
impressions and aesthetic appeal [33]. They found that the first impressions could
even happen as quickly as 17ms. The result showed that people find websites with
less visual complexity and high prototypicality (refers to how representative a de-
sign is within a particular category) more appealing [33]. People have expectations
on how a particular type of website should look, a key takeaway from the research
done by Google is that designers should aim to meet that expectation. If a web-
site for e-commerce is being designed, it should follow typical design patterns for
e-commerce. On the topic of first impressions for a website, it has also been found,
through analyzing eye-tracking data, that it takes about 2.6 seconds for the eyes to
settle on key areas of the web page[34]. It was further discovered in the study that
good design choices inherently leads the users to stay longer on the web page.
In summary, there are numerous aspects a designer need to pay attention to when
designing for a good user experience. It simply isn’t enough to aim for good usabil-
ity; the designer needs to make an effort in aesthetically making the product/service
appealing as well. By designing for a good user experience, it will not only lead to
less frustration for the user, but it will also increase the likelihood of the user com-
ing back, this could be especially useful in the context of e-commerce. Users spend
most of the time on e-commerce pages browsing through products, having a good
user experience could be the difference between users doing just browsing or them
buying the product [31]. Finally, in order to try and clarify the concept of UX even
further, let us take a look and two definitions of UX from different perspectives.
“The first requirement for an exemplary user experience is to meet the exact needs of the
customer, without fuss or bother. Next comes simplicity and elegance that produce products
that are a joy to own, a joy to use. True user experience goes far beyond giving customers
what they say they want, or providing checklist features. In order to achieve high-quality
user experience in a company’s offerings there must be a seamless merging of the services of
multiple disciplines, including engineering, marketing, graphical and industrial design, and
interface design.” [35]
“User experience is not about the inner workings of a product or service. User experience is
about how it works on the outside, where a person comes into contact with it. When someone
asks you what it’s like to use a product or service, they’re asking about the user experience.
Is it hard to do simple things? Is it easy to figure out? How does it feel to interact with the
product?” [31]
Researchers have said that UX shouldn’t be restricted to interaction with a product
or artefact [28]. However, within the context of this papers research topic, the main
focus will be on user interaction with a product. More specifically, interaction with
a product having a user interface (UI).
Chapter 3. Theoretical framework 10
Factor Description
Useful To whom is the product useful? if the product isn’t useful to
someone then it won’t be able to compete with other products.
Usable Making sure that the end objective of the product is effectively
and efficiently reached.
Findable The product itself is easy to find and easily attainable. This also
refers to how findable content within the product is. If a user
can’t find the product, then of course they won’t use it.
Credible This refers to the end users being able to trust the product that’s
been provided. The product provides information that is accu-
rate, it does what it’s suppose to do and will do so for a long
period of time.
Desirable “Desirability is conveyed in design through branding, image,
identity, aesthetics and emotional design.” [36]. A cheap watch
might fulfill every other UX criteria, however when presented
with the choice of having the that or a Rolex, most people would
choose the Rolex, it’s more desirable.
Accessible The product should be accessible to a wide range of user abilities.
Valuable The product must deliver value, both to its users and to the busi-
ness.
According to [39] each phase in the process can be defined as shown below.
Phases Description
Context Understand in what context the system will be used. Who will
use the product, what will they use it for and under what circum-
stances.
Requirements Specify user requirements. Which requirements or user goals
must be met in order for the product to be successful.
Design Develop design solutions. The design will evolve in each itera-
tion, starting from rough concepts working towards a complete
design.
Evaluate Evaluation of the design solution, preferably done through us-
ability testing.
In the early stages of the design process, design-solutions can be made just by paper
and pencil drawings [40]. It is vital to get feedback from users in every iteration, even
if thorough examinations have been made beforehand, conducting user tests allows
Chapter 3. Theoretical framework 12
for discovering additional information that might have been overlooked or missed
in early research [40]. As the design process progresses, the prototype evolves in
each iteration based on feedback working towards the final design of the product.
As earlier described, how important it can be to present the user with an aesthetically
beautiful product, testing the usability of a product is also extremely important in
achieving a great user experience. Usability testing means evaluating a product by
testing it on potential users [41]. Usability testing can be broken down into seven
key steps [41].
1. Create a prototype
2. Decide on a test plan
3. Choose test participants
4. Perform the test
5. Analyze the test results
6. Document the data
7. Report your findings
case for a VTO-system in IFE. Still, there are some useful guidelines presented both
by Google and Apple, which can be utilized.
Although UX and AR have received some attention (in the context of retail) by re-
searchers (see chapter 2). Much of the research mentioned focuses on mobile aug-
mented reality (MAR). However, it has been pointed out that general AR design
guidelines can be considered valid independent of device platforms or context of
use [43]. A study made an effort in amongst other things, to evaluate the UX in
AR. The evaluation was performed on Ray-bans website where there were ranging
versions of AR-integration available. Based on the result it was implied that “A well-
designed AR should be practical, easy to use, easy to learn, organized, symmetric,
attractive, and pleasant if it is to provide effective and relevant information to users,
and it should contemplate the emotional aspects of UX.” [22].
Another study conducted two case studies where they tested two AR-application,
one which could guide new students around the campus area and another, which
was a commercial MAR application [8]. The study aimed to evaluate how users
perceive MAR applications emotionally, discover challenges, and opportunities as
well as best practices for UX and MAR. Their findings indicated that users value
MAR applications based on its functionalities, usefulness and ease of use [8]. It was
further pointed out that “A major reason as to why designers should utilize AR in
mobile applications is to provide positive experiences to users by engaging them
emotionally with the application.” [8].
One study has looked at expected user experiences of MAR services [44], specifically
within the context of shopping centers. The researchers conducted interviews to try
and determine expectations of MAR services, as well as user requirements central
within the context, which was deemed as a potential area of use for MAR services.
The study was not done on any existing MAR-application, but rather to evaluate the
expected experiences and requirements beforehand. Similarly to the studies men-
tioned above, it was found that MAR is expected to create an emotionally enhanced
experience. It should create a playful, stimulating and pleasant experience. More-
over, it was concluded that in terms of UX, expectations were often utility-based.
Meaning that the use of AR needs to fill a purpose; it needs to be useful to the user
and provide them with relevant information.
In summary, many design aspects of an AR-application falls in under the umbrella of
UX. However, there are some unique challenges specific to AR. One being that UX-
designers usually create prototypes in a 2D-environment, with AR comes the new
challenge of designers having to sketch 3D-objects and scenes viewed through a 2D-
perspective which is the camera [8]. Advancements have been made in this area.
However, the amount of technologies available for rapid creation of low and high-
fidelity prototypes is still quite limited [8]. However, with a VTO-system specifi-
cally targeting the facial region, much of the prototyping difficulties can be avoided,
which is discussed later in this chapter.
only applies if there are both objects which supports VTO and objects which does
not. Clarity is stressed in Apple’s design guidelines, expressing “Use badging only
when your app contains a mixture of objects that can be viewed in AR and objects
that cannot.” [12]. While a user has activated the VTO on a pair of sunglasses,
how should the user be able to switch between different colors of the glasses and
should any information be present about the product during the VTO? Google has
mentioned that if a standard UX interaction model, such as tapping or swiping, can
be utilized, it should [11]. They further mention that it is important to keep the user
emerged in the experience, also to keep the experience simple, do not overload the
user with information. To keep the user emerged, Google expresses that “if users
need to select, customize, or share an AR object, try to figure out a way they can do
it without leaving AR.” [11]. Additionally, it is also mentioned by Google that it is
important to let the user know if something is offscreen.
more advanced. For instance, Samsung now has something called AR Emoji avail-
able within their camera application. AR Emoji allows the user to create an avatar
based on themselves. For prototyping VTO in a webshop, this feature can be bene-
ficial. Aside from being able to create an avatar, it is further possible to customize
the avatar in several different ways. For instance, it is possible to apply glasses of
different kinds as well as different types of headpieces. When creating a prototype,
this avatar can be used to demonstrate the visual idea of virtually trying on different
accessories.
16
Chapter 4
Method
In developing a prototype for this thesis, an iterative design process have been
adopted. The design process is based on UCD (see section 3.1.2), where design
evaluations are done in each iteration. The design evaluations are done through
usability testing, and the design evolves in each iteration based on feedback from
tests. Early on in the start-up phase of this thesis, the idea was to conduct a survey
to specify user needs and requirements for a VTO-system. However, it was found
that researchers have already evaluated augmented reality (AR) and virtual try-on
(VTO) in several different aspects, which is mentioned both in Chapter 2 & 3. With
the knowledge of that, it was decided, this thesis would build on existing research
and focus entirely on the design aspect of VTO.
Further, it is essential to mention that this thesis focuses primarily on the AR-interactions
within a webshop, not the webshop itself. The reason behind this is that Tactel has al-
ready made a prototype for a webshop targeting in-flight entertainment (IFE). With
that said, a webshop will still be designed but only to clarify how AR will be used
within that context.
1. Researchgate1
2. Google Scholar2
3. Diva-Portal3
Except for above web sites, in order to find other resources from other sources than
researchers, Google’s regular search engine have been used as well which has al-
lowed for discovering both Apple’s and Google’s design guidelines for AR. Com-
bined with the literature study, current implementations using VTO has been stud-
ied. The reasoning behind this was to get a better understanding of different use
cases.
4.1.4 Evaluation
The evaluation of the lo-fi prototype was done through usability testing. The ap-
proach adopted was the “think aloud”-approach which encourages the test par-
ticipants to explain their thought process as they are evaluating the design. This
approach also allows the designer to observe how users interact with the product
carefully. The observation takes place as test participants are performing different
tasks given by the test leader. For each new task, once they “clicked” on, for in-
stance, a button the test leader presented a new page of the application. Keeping
the pages of the application on separate papers was done to avoid leading the test
participants into making a specific choice. This way, they would not have seen the
next page already. Finally, once the usability testing was done, the test leader asked
each participant a couple of questions regarding the evaluation and the design. The
questions targeted the user’s perceived experience of the product.
1 https://www.researchgate.net/
2 https://scholar.google.com/
3 http://umu.diva-portal.org/smash/search.jsf?dswid=-346
Chapter 4. Method 18
Test group
The test group consisted of persons with different levels of travel experience and
knowledge of an IFE-system. The test participants had no prior knowledge about
the content of this master thesis. As for the number of test participants, five persons
participated in usability testing. The choice was based on Jacob Nielsen’s recom-
mendation that a designer should strive towards qualitative data, not quantitative
when doing user research [47]. More specifically, Nielsen concludes that around five
participants in usability testing are enough.
4.2.2 Wireframe
The purpose of the wireframe is first off to make the prototype a bit closer to its final
representation. The wireframe was based on the lo-fi prototype combined with the
acquired feedback from the user testing in iteration one. The wireframe still does
not have any colors, images or animations in the design. The difference is that it was
made with AdobeXD rather than with paper and pencil, which made it possible to
make the prototype interactive. This way, the test users were able to interact with
the prototype more realistically.
4.2.3 Evaluation
Same as in iteration one, the evaluation was done through usability testing with a
“think aloud”-approach. However, as mentioned, the prototype was now an inter-
active prototype, which meant the tests were conducted on a laptop. Other than that
the test concept was the same, meaning the participants were given several differ-
ent tasks to complete while describing their thought process when performing the
tasks. Finally, this test also included the test leader asking each participant a couple
of questions regarding the experience and design.
4.3.2 Evaluation
In order to evaluate the final prototype, a different approach was adopted. In the
earlier iterations, the idea was to evaluate each prototype on people with no prior
knowledge on this thesis, as well as people with limited knowledge about IFE-
systems in general. For the hi-fi prototype, it was decided to let a UX-designer at
Tactel evaluate the design. The reason for this was that while the other two proto-
types were tested for the general navigation flow and logic, they did not provide
a result specific to just IFE-screens. The test participants evaluated the design in a
more general sense, because of that it was deemed to be a good idea to have someone
with experience and knowledge within the field of UX and IFE-screens to evaluate
the prototype.
4.4 Implementation
As a final part of the practical work in this thesis, a small application was build
using Android and ARCore. The focus in the application was on viewing a pair
of sunglasses augmented on to users face. This work was done using the official
Integrated development environment (IDE) for Android called Android Studio, and
the programming language used was Java. Furthermore, a 3D-framework called
Sceneform has been used to overlay 3D-models on to a detected face. Sceneform
makes it easier for java developers to import, render and build 3d assets. Finally, a
3D-modeling software called Blender has been used for positioning, orientation and
scaling of 3D-models.
The idea of the application was not to implement a fully developed webshop with
integrated AR which is outside the scope of this thesis, but rather to build an appli-
cation demonstrating the concept and how it could work in a real scenario. It was
also an opportunity to test the capabilities of ARCore.
20
Chapter 5
Result
In this section the results from each design iteration is presented. This section fol-
lows a similar structure to the method-section.
( A ) Categories ( B ) Accessories
( A ) Products page with toggle filter ( B ) Products page with try on icon
F IGURE 5.5: Start page and Virtual try-on menu choice selected
5.1.2 Evaluation
Below the evaluation is broken down in to sections for each of the two design con-
cepts. The design concept which is shown from figure 5.1 to 5.4 will be referred to
as “design concept one” and figure 5.5 & 5.6 will be referred to as “design concept
two”.
For the second task, once the participants were steered in the right direction, every-
one was able to select sunglasses successfully.
As for the third task, everyone understood how to add items to the basket. However,
only one understood where to go after the items had been added. It was mentioned
that there needs to be some feedback when an item has been added to the basket.
One participant said that “similar to an online shopping cart; it would be good to
have a small number that indicates how many items are in the basket”.
Once the basket was found, everyone understood how to activate the VTO. Every-
one also understood how to switch between the glasses.
General feedback
Everyone preferred design concept one where the VTO was highlighted through
an icon at the top left corner of the product. It was expressed here that having a
“try-on basket” would be a nice feature. However, the feature only really is useful
if the user already knows beforehand what he or she is looking for, then it would
be great to select the products which are appealing and test them. They elaborated
further that when searching for a product, they would prefer to navigate to that
product by searching or through filtering, however by doing so, it seems to remove
the possibility of trying on that product or adding it to the try-on basket.
5.1.3 Analysis
Having a try-on basket seemed to be a desired feature; however, with the current
design, it left some of the users a bit confused as to how it works. Everyone preferred
to browse the products by going through the categories section, which means that
the option of adding items to a try-on basket needs to be available from there. The
majority also preferred having an icon that indicates which items support VTO but
still offer the possibility of applying a filter which can filter out all the products
which support VTO.
Chapter 5. Result 26
5.2.1 Wireframe
The general design concept remained the same; however, some changes were made.
The menu button in the top left corner was removed as it did not serve any purpose.
Other than that the start page and categories pages have stayed the same, which can
be seen in figure 5.7 & 5.8. For the product page, having an icon indicating which
products support VTO was kept. The page displaying images and information about
a product stayed the same. The product and information page can be seen in figures
5.9 & 5.10. The biggest change is in the page demonstrating the try-on. This time the
feature is accessed directly from the try-on page. The selection of multiple products
is made through the addition of another button “View more”, see figure 5.11. The
button opens a view which offers the possibility of selecting multiple products, see
figure 5.12.
( A ) Categories ( B ) Accessories
5.2.2 Evaluation
Below is a summary of result from the test, along with the tasks which each test
participant where asked to perform.
Tasks
1. Navigate to sunglasses
2. Select a pair which supports Virtual try-on
3. Open the Virtual try-on feature
4. Select three glasses to try
5. Switch between the glasses
All of the test participants were able to find sunglasses and select a pair which sup-
ports VTO. One participant thought that the marker in the top left corner of a prod-
uct was a button rather than an icon.
Everyone was also able to open the VTO-feature without any problem.
Four out of five were able to find where to select more glasses to try. One participant
missed the button where this feature is activated, and it was required for the test
leader to steer the test participant in the right direction in order to find it.
Finally, everyone understood how to switch between the selected glasses. However,
one participant pointed out that it would be good with some text indicating how to
switch between the glasses.
Chapter 5. Result 30
General feedback
Similar to the previous iteration, everyone liked the idea of being able to select mul-
tiple glasses. However, the general consensus was that it was a bit diffuse to under-
stand. While most of the participants were able to find where the selection is made,
some felt it might be a feature which they probably would not use. A point which
was brought up is the fact that it does not seem to provide the option of just clicking
on a pair of glasses to view more information about them, it seems only to provide
the option of selection. Finally, one participant mentioned the option of deselecting
glasses, expressing that it was not clear how that was done.
5.2.3 Analysis
Based on feedback, the overall idea of having the option of selecting multiple prod-
ucts to try needed some consideration. Being able to select multiple products proved
again troublesome integrating into the design.
5.3.1 Mock-up
For the final design, the layout and navigation flow has remained the same. The
start page all the way to the product information page can be seen in figures 5.13 to
5.17. The major difference is in the Virtual try-on view, where the choice of selecting
multiple products have been removed. Instead, a user is now presented with the
choice of switching between similar products. Furthermore, in order to notify a user
that similar products can be shown by swiping or clicking the arrow, a text appears
above the card which describes this action, see figures 5.19 & 5.20 for the Virtual
try-on demonstration.
Chapter 5. Result 31
5.3.2 Evaluation
From the meeting with the UX-designer at Tactel, overall the feedback was positive,
and the layout did not need any corrections. However, one point which was brought
up was a scenario where this is used during a flight, sometimes there is turbulence,
and the flight is not perfectly smooth which means that it could be challenging to
click on buttons and menus if the components are too small. The screen is located
at a fixed distance from the passenger, so it is crucial to make sure that everything is
clear and easy to see. Another point which was brought up was the “sliding” card
in the VTO-view, which can be seen in figure 5.19 and 5.20. The view starts with the
card hidden, as demonstrated in figure 5.18. It could, however, be a better idea to
make the card visible right from the start as it holds information about the product
as well as navigation cues.
Important to note is that the mock-up presented above have already had corrections,
which means that most text fonts have been enlarged along with most “clickable”
components. The corrections were done based on the feedback from the meeting
mentioned above.
5.4 Implementation
Below is the result of the implementation phase of this thesis presented. As men-
tioned, the focus point was on viewing a pair of glasses augmented onto users face.
However, for demonstration purposes, it was decided that a product list page and
product information page would better put the Virtual try-on in the right context,
see figure 5.21. For the VTO it was initially planned to change glasses by swiping
left or right on the screen. However, after several attempts, it proved troublesome to
get this to work for some reason, so instead two buttons were added which can be
seen in figure 5.22 which also shows the VTO. For integrity reasons, the person has
been blurred out.
Chapter 5. Result 36
Chapter 6
Discussion
This chapter discusses the general findings of this study. First off the research ques-
tions of this study is answered based on the findings from the design process. Fur-
thermore, how well the result can be reflected in the theory section of this thesis
found in Chapter 3. Finally, the design process along with the implementation phase
is discussed in a broader sense.
Below is the main research question of this thesis.
How should AR-interactions be designed in an UX-friendly way for a web-
shop?
Part of this thesis was to look at research done in the field of user experience (UX)
and augmented reality (AR) and try to gather information about what is essential
when designing for AR, more specifically AR in the context of virtual try-on (VTO).
The two significant sources of inspiration have been Apple and Google’s design
guidelines for AR, but general principles for UX have been utilized as well.
The main question was divided into a number of sub-questions which is presented
and answered below.
How can current knowledge within the field of UX be applied to AR and
VTO?
The result from the usability testing in first two iterations showed that it was crucial
that the webshop itself, excluding the VTO, had a pleasant and easy to understand
design. Even more obvious was the fact that people have expectations of what a
webshop should look like, which also was a key takeaway from a study made by
Google regarding first impressions of websites (see section 3.1.1). It became evident
that people have a preferred way of navigating a webshop or idea of what it should
look like, which means that if the design deviates too much from that it can detract
from the user experience. Making the design simple and familiar seemed to be the
preferred approach. Making the VTO easy to find and understand proved crucial as
well. For instance, as mentioned in section 3.2.1, Google mentions that if a standard
UX interaction model, such as swiping or tapping, can be utilized, it should. They
further mention that it is important to let the user know if there is something that can
be found offscreen. Having an icon indicate that there are more glasses to be viewed
by either swiping or taping an arrow icon proved useful in the usability testing as
everyone understood how to switch between glasses.
In conclusion, the result points towards keeping the design simple, and that VTO
should be integrated seamlessly into the application. However, one aspect which has
not been tested is what would happen if a user has knowledge about the VTO before
Chapter 6. Discussion 38
opening the webshop. With knowledge beforehand, it may be that user’s would be
more open to a layout which makes a more clear distinction between “normal” and
VTO-products right from the start page. With knowledge of VTO, it might even
be that user’s expect to find VTO directly when opening the webshop. This idea
certainly is something which would be interesting to explore further.
One reason why simplicity and familiarity seemed to be important might be that
VTO is still a new concept and far from every webshop has this feature available. It
could be that in the future it becomes common for a webshop to offer VTO, which
in turn might change people’s expectations of a webshop.
All in all, it is safe to say that when designing a webshop with integrated VTO, even
though VTO is a relative new concept, current knowledge within UX still applies.
Looking at some of points mentioned above, they tie in with what Garret says re-
garding UX in websites (see section 3.2.1), they don’t come with instructions so it’s
crucial that they are easy to understand. Utilizing design methods such as user-
centered design (UCD) is also a great tool for evaluating design choices, be that VTO
or just a webshop.
As for how well current knowledge within the field of UX applies to AR-applications
in general is not something which this thesis can answer. But as mentioned in section
3.1.1, researchers has mentioned that UX and AR is something which needs further
attention in the future. Use cases of AR can be quite different across different ap-
plications, and it might be that the current knowledge in the field of UX needs to
expand in order to keep up with a technology like AR.
Is the design guidelines for AR, presented by Google and Apple, applicable
when designing a webshop with VTO?
The two significant sources of inspiration have been Apple and Google’s design
guidelines for AR. Based on the three design iterations which were performed, it
is hard to say exactly how good those guidelines are. The problem is that not ev-
ery AR-application can use the guidelines to the letter; it depends on the context in
which the usage of AR is intended. For instance, a map application where AR is
used to present information based on different objects viewed through the camera
is very different from an application which allows a user to virtually try-on glasses.
With that said, regardless of the use case of AR, there is something there for every-
body. For instance, as mentioned in section 3.2.1, Google mentions the importance
of keeping the experience simple and the users emerged. Based on the usability
testing, there certainly seems to be grounds for that claim. Both design concepts in
iteration one and two explored the idea of selecting multiple glasses to try. However,
the importance of keeping it simple became evident when in both iterations people
struggled with this concept; it seemed to be adding too many steps, which ultimately
led to the removal of that concept. Another guideline which was useful came from
Apple regarding indicating products which supports AR, where they mention that
badging can be a useful tool. The result from the usability testing showed even
in the first iteration that badging seemed to be the desired approach when finding
products that supports VTO.
In summary, the design guidelines for AR by Google and Apple is a good starting
point when designing for AR. Overall, the focus of the guidelines is on AR where the
back-facing camera is used. However, using the back-facing camera in not how the
VTO was intended for the particular use case in this study. As the use case of VTO
in this study differ from the targeted AR-applications in the guidelines, quite a lot
Chapter 6. Discussion 39
cannot be used. However, as some design choices were based on the guidelines, and
also received positive feedback in the usability testing, they certainly can provide
some value when designing a webshop with integrated VTO.
One aspect of AR mentioned in other studies (see section 3.2.1) is that it should
be used to engage users emotionally and create a positive experience. Measuring
emotional experience is very difficult, and not something which has been attempted
in this thesis. However, it has been said by researchers that VTO can produce higher
user satisfaction along with increased shopping eperience (see section 2.2.2).
How can design concepts of AR-applications be created?
Creating a prototype for virtual try-on of sunglasses was initially a dilemma since
it would involve a 3D-environment. However, after doing some research, the in-
spiration came from Apple’s yearly World Wide Developers Conference (WWDC)
where they mention that using a smartphone’s camera is a great way of simulating
an AR-experience (see section 3.2.3). Through that came the realisation that Sam-
sung has a feature on their newer phones called AR Emoji which allows a user to
create an avatar based on themselves. Additionally, AR Emoji offers the possibility
of customizing the avatar, where different sunglasses can be placed on the avatar.
This turned out to be a useful feature when creating the hi-fi prototype to demon-
strate the virtual try-on of sunglasses. However, VTO is a very specific type of use
case of AR, for other AR-applications it might be better to utilize the camera in a
different way. There are prototyping programs out there which is made specific for
AR, but utilizing the camera along with for instance, paper or card board to simulate
information on the screen, seems to be a good option.
Worth mentioning is that the final mock-up did not go through usability testing, but
instead more of a usability review together with a UX-designer at Tactel. The Mock-
up was, however, demonstrated at the company and received positive feedback.
In order to fully say that demonstrating VTO through an avatar with sunglasses
was a good approach, further testing would need to be performed. However, the
indication through demonstrating the mock-up along with the usability review was
that it was a good tool to use for the purpose.
Is it possible to implement virtual try-on using ARCore, and how difficult is
it?
Having utilized ARCore in the development of a small application which allows
a user to try on glasses, there are some critical takeaways. First off, ARCore does
provide a straightforward transition into AR-development. In order to develop a
full virtual try-on application from scratch, it would require extensive knowledge
about computer vision and machine learning as well. ARCore handles most of this
for you through high-level API:s, one such being Augmented Faces which is used
to detect faces and specific facial regions. Making use of the API:s means that it is
quite easy to get started. In order to render virtual objects such as 3d-models onto a
face, a graphics API called OpenGL can be used. OpenGL is, however, a low-level
API which is very powerful but requires significant knowledge about 2d/3d math
and everything needs to be handled manually. Fortunately, there is a 3d-framework
called Sceneform which handles most of the rendering of 3d-models for you. At
the time of developing the try-on application, Sceneform had just been released as
open-source. Although, while that is great, it was still in a transition phase where the
documentation was lacking and quite a lot of bugs where being reported. Therefore
Chapter 6. Discussion 40
the choice was made to use the last non-open-source version, which had excellent
documentation and guides in place.
Overall, Sceneform and ARCore made it quite easy to make this application, and the
result turned out better than expected. However, there is room for improvements.
For one thing, if a user looks straight into the front-facing camera the glasses look
decent, but from an angle it looks like the glasses are “floating”, meaning it does not
look like they are attached to the face. In order to solve this, one solution might be to
manually calculate different positions on the face and use those positions as anchor
points for the glasses. There are probably other ways as well, but with a limited
time frame, it was decided not to dive too deep into that. While Sceneform now
is open source, unfortunately, ARCore is not. That is a big drawback for ARCore
as it means that you are stuck with the public APIs which are provided and you
cannot tell what is going on “under the hood”. For a company who needs flexible
and adaptable solutions, going with a custom solution still might be a better choice.
In summary, yes, ARCore can be used to create virtual try-on for sunglasses. How-
ever, for the specific use case of testing virtual sunglasses it is not perfect. ARCore
does not provide good anchor points on the face to properly place the glasses which
means that in order to get a good result, manual calculation of points on the face
would be needed.
Finally, it is essential to note that this thesis specifically targets the virtual try-on
and designing a user interface (UI) for that purpose. However, to achieve that it was
deemed necessary to design a UI for the webshop, putting the VTO in better context.
VTO is still a relatively new concept, and overall few people are “experts” in this
area. Furthermore, as of today, the screens which this design are indented for does
not yet have the hardware capabilities to support VTO, at least not for the usage
of Google’s augmented reality framework ARCore. The lack of hardware support
coupled with the early stages of VTO in general means that it is uncharted waters,
which also was a factor in the decision not to include people who work with IFE
development and design in every iteration.
Based on the usability testing performed in this study, the result points towards
keeping the design simple, and that VTO should be integrated seamlessly into the
application. However, one aspect which has not been tested is what would happen if
a user has knowledge about the VTO before opening the webshop. With knowledge
beforehand, it may be that user’s would be more open to a layout which makes
a more clear distinction between “normal” and VTO-products right from the start
page. With knowledge of VTO, it might even be that user’s expect to find VTO
directly when opening the webshop. This idea certainly is something which would
be interesting to explore further.
Chapter 7
Conclusions
This study implies that when designing a webshop with integrated virtual try-on
(VTO), it is essential to keep the design and interactions as simple as possible. The
usability testing performed in the first two design iterations indicates that people
have certain expectations when it comes to a webshop. If those expectations are
not met, it does not matter how good or useful a feature specific to the VTO might
be in theory, people still might be thrown off if the layout does not correspond to
their expectations. Test participants almost exclusively preferred to find products
that support VTO through “traditional” navigation such as through categories and
subcategories. These findings suggest that VTO should be integrated seamlessly into
the application without interfering with the layout. It is, however, important to note
that only a few design ideas have been evaluated, it might very well be that there
are more complex solutions which would get just as good or even better feedback.
Still, the design ideas presented in this study, along with the feedback, can at least
indicate what seems to work and what does not.
Although the findings suggests a more seamless integration of VTO into the web-
shop, notifying a user of VTO before the webshop is opened could potentially change
that. However, pushing notifications to a user, informing them about VTO is not
something that has been evaluated in this study. This study specifically targeted
VTO in the webshop, but for future studies or implementations, one idea is to con-
sider VTO in a broader sense within the in-flight entertainment system.
Overall, by looking at research done in the area of user experience (UX) and aug-
mented reality (AR) it became evident that this area is still very new, even Google
and Apple’s design guidelines are in some cases very weak and open to interpre-
tation. When designing an AR-application, the guidelines by Google and Apple
should not be expected to cover everything; they are, however, an excellent place to
start. From that starting point, the designer will still need to perform tests in order
to get a feel for what users expect and prefers.
When implementing a virtual try-on application, ARCore certainly can make that
process much more comfortable, and a simple application can be built with little
to no knowledge about AR-development. For a company to use this framework,
some aspects need to be taken into consideration. As of today, ARCore is not open
source, which can make it less appealing for companies who needs highly flexible
and adaptable solutions. Therefore, unless ARCore can cover every need, it still
might be better to go with a custom solution.
Chapter 7. Conclusions 43
Bibliography