Effect of User Roles on the Process of Collaborative 2D Level Design on

Large, High-resolution Displays

Anton Sigitov

, Andr

e Hinkenjann

and Oliver Staadt

Institute of Visual Computing, Bonn-Rhein-Sieg University of Applied Sciences, Sankt Augustin, Germany

Institute of Computer Science, University of Rostock, Rostock, Germany

Keywords:

Computer-supported Cooperative Work, Co-located Collaboration, Groupware, 2D Level Design, Group

Behavior, User Roles.

Abstract:

This paper presents groupware to study group behavior while conducting a creative task on large, high-

resolution displays. Moreover, we present the results of a between-subjects study. In the study, 12 groups with

two participants each prototyped a 2D level on a 7m x 2.5m large, high-resolution display using tablet-PCs for

interaction. Six groups underwent a condition where group members had equal roles and interaction possibil-

ities. Another six groups worked in a condition where group members had different roles: level designer and

2D artist. The results revealed that in the different roles condition, the participants worked signiﬁcantly more

tightly and created more assets. We could also detect some shortcomings for that conﬁguration. We discuss

the gained insights regarding system conﬁguration, groupware interfaces, and groups behavior.

1 INTRODUCTION

Large, high-resolution displays (LHRDs) are highly

suitable for collaborative work. Due to their large size

and a vast amount of pixels, they allow visualization

of large data sets. They foster the exploration of data

sets through intuitive and effective physical naviga-

tion (Ball et al., 2007; R

adle et al., 2013); they al-

low users to better overview the content, and to make

use of spatial memory while searching for items on

the display (Andrews et al., 2010). Additionally, the

large size provides much better support for multi-user

support in comparison to standard desktop displays.

Polley and McGrath (1984) listed four collabora-

tive task types: planning, creative, intellective, and

contest. The creative process was investigated on

LHRDs to a limited extent only. The tasks used in

the previous research set constraints that did not al-

low to unfold creativity. For instance, jigsaw com-

position (Azad et al., 2012; Scott et al., 2003), text

composition (Ryall et al., 2004), the arrangement of

photos (Hinrichs et al., 2006), outline task (Tse et al.,

2004), or interior layout planning (Scott et al., 2004),

allowed only to layout assets, but not create them.

Additionally, in the case of the jigsaw and text com-

position, the outcomes were more or less predeﬁned.

Moreover, researchers investigated the creative pro-

cess mostly on tabletops. On the vertical displays, the

analytical process was more in focus, e.g., Andrews

et al. (2010), Vogt et al. (2011).

In this work, we developed groupware (collabora-

tive software that allows multiple users to work on a

common task) to support and investigate the 2D level

design process on LHRDs using tablet-PCs for inter-

action. The process combines at least two collabora-

tive tasks: planning and creative. The groupware al-

lows not only to place/layout assets but also to create

them. We also conducted a study to evaluate the de-

veloped groupware in combination with our LHRD.

We assessed users attitude towards the system, ana-

lyzed the usage of the system, and looked into behav-

ioral patterns in terms of territoriality, collaborative

coupling, and workspace awareness.

Previous research detected that during collabora-

tion on LHRDs users occasionally undertake different

roles to approach the task in a more efﬁcient man-

ner (Vogt et al., 2011; Rogers and Lindley, 2004).

As a result, researchers suggested to consider user

roles when building groupware and provide group-

ware mechanisms that will support them. One pos-

sibility to support user roles is to provide interfaces

that consider all possible roles and allow for dynamic

switching between them. Another way is to provide

different interfaces where each interface focuses on

a speciﬁc role. In both cases, users must coordinate

themselves to distributed roles as needed. However,

in the case of universal interfaces, coordination is vol-

118

Sigitov, A., Hinkenjann, A. and Staadt, O.

Effect of User Roles on the Process of Collaborative 2D Level Design on Large, High-resolution Displays.

DOI: 10.5220/0008933801180129

In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2020) - Volume 2: HUCAPP, pages

118-129

ISBN: 978-989-758-402-2; ISSN: 2184-4321

 2022 by SCITEPRESS – Science and Technology Publications, Lda. All rights reser ved

untary and therefore, might never take place. Special-

ized interfaces, on the other hand, might be perceived

as too restrictive and awkward, similar to Rogers and

Lindley (2004).

In our study, we investigated the effect of different

interface types on the collaboration process. Twelve

groups two participants each participated in the study.

All groups had the same task. However, the partici-

pants of the ﬁrst six groups had the interface which

allowed them to create tiles and place tiles on the

LHRD. Thus, the participants could decide on their

own who will undertake what role in case they de-

cided to distribute the roles. The participants of the

other six groups had different interfaces which either

allowed to create assets or to place them. Thus, the

participants had to undertake a speciﬁc role.

Our contribution is twofold. First, we present a

groupware system for collaborative level design on

LHRDs. The system is based on the Unity engine

and could serve as basis groupware not only for level

design tasks but also for other creative tasks. We pro-

vide the results of the evaluation, as well as insights

regarding the acceptance of the system setup and level

design task in the given system environment. Second,

we provide insights regarding the effects of the ex-

plicit distribution of user roles on the process of col-

laborative 2D level prototyping. The results could be

generalized to any other collaborative design process

on LHRDs.

Our research extends the theoretical basis and im-

proves understanding of collaborative processes dur-

ing creative work on LHRDs. It reveals and cate-

gorizes behavioral patterns of groups and provides

knowledge of how group behavior can be guided. Ad-

ditionally, it provides insights useful for the profes-

sional practice of building collaborative systems and

groupware for creative tasks.

2 RELATED WORK

In this section, we provide a brief overview of re-

search that considered planning, creative tasks, and

games on LHRDs, as well as research that investi-

gated group behavior during co-located collaboration

on LHRDs.

Planning and Creative Tasks on LHRDs. Azad

et al. (2012) conducted a controlled experiment where

the participants had to solve jigsaw puzzles in non-

collaborative and highly-collaborative conﬁgurations.

They looked into on-display behavior, off-display be-

havior, and combined behavior. As a result, they

could develop some guidelines for groupware design

regarding territoriality and reaching and sharing inter-

actions.

Jakobsen and Hornbæk (2016) conducted a study

on an LHRD comparing touch input to mouse input.

The study consisted of two tasks: the newspaper task

and the puzzle task. In both tasks, participants had

to layout either puzzle pieces or articles. They found

that LHRDs can enable equal participation regardless

of input, yet touch input can frequently become an

interference source.

Liu et al. (2017) executed a study where the partic-

ipants had to ﬁnd similarities or connections between

the pictures and arrange them in a meaningful way.

They used the scenario to evaluate touch gestures for

cooperative work. They also enabled users to perform

the gestures on multitouch tablets, thus allowing for

indirect interaction. However, remote gesturing was

not accepted well by all participants.

Some other researchers investigated layout tasks

in tabletop environments. For instance, Ryall et al.

(2004) let the participants assemble target poems us-

ing word tiles. The focus of the study was on the ef-

fects of the table size on social interaction and the ef-

fects of the group size on task performance and work

strategies. Another study with a layout was conducted

by Scott et al. (2005). In this study, they evaluated a

storage bin mechanism for passing objects in tabletop

environment.

The research of planning and creative tasks on

LHRDs shows that 2D level prototyping on LHRDs

is possible and worth investigation. However, many

questions regarding interaction, visualization, and

handling of social phenomena are open and require

answers. Moreover, many studies considered creative

and planning tasks isolated from a context. Thus,

researchers must investigate how the combination of

proposed visualization and interaction techniques will

perform together.

Gaming on LHRDs. Sabri et al. (2007) imple-

mented one of the ﬁrst LHRD games, which was a

real-time strategy game. They run the game on a dis-

play that consisted of nine displays arranged in a 3x3

matrix with a total resolution of 8 Megapixels. They

compared gamers performance on the LHRD to the

performance on smaller displays, investigated usabil-

ity issues, and evaluated different display form fac-

tors and input possibilities. The results revealed that

players performed signiﬁcantly better on the LHRD

in comparison to standard desktop displays.

PyBomber, developed by Machaj et al. (2009), is,

to the best of our knowledge, was the ﬁrst multiplayer

LHRD game. The game run on the GigaPixel display

that consisted of 50 displays with a total resolution of

96 Megapixels. Machaj et al. (2009), used the game

to investigate the effect of the LHRD environment as

Effect of User Roles on the Process of Collaborative 2D Level Design on Large, High-resolution Displays

119

well as different player conﬁgurations on social inter-

actions and physical navigation. They could observe a

shift in social dynamics with an increasing number of

players. Additionally, they provided some guidelines

for the design of LHRD games.

Miners is another multiplay game for touch-

sensitive LHRDs developed by von Zadow et al.

(2016). Miners is a fast-paced, collaborative game

that must be played precisely by four players. Each

player has a unique role with speciﬁc abilities in the

game. It is a cooperative game, which means that the

players have to bundle their forces to beat the game.

The participants found the game enjoyable and ex-

posed a high level of engagement. The observation

results showed as well that participants suffered from

limited awareness of other participants because of the

employed interaction technique that allowed only for

direct interaction with the display from up close.

The studies show that gaming on LHRDs not only

possible but also well accepted by players and makes

fun. Thus, it represents a potential new development

branch for computer games. The prototyping pro-

cess of game levels for LHRDs on standard desk-

top displays might be tedious because of the small

workspace they provide. Therefore, it can be of ad-

vantage to develop LHRD groupware that will allow

for level prototyping directly on LHRDs.

Territoriality. In the context of computer-

supported cooperative work, territoriality aims for al-

location of particular workspace areas for particular

activities, for own work, and assets protection, e.g.,

Isenberg et al. (2010), Scott et al. (2004), Tang et al.

(2006).

Tang (1991) conducted one of the ﬁrst researches

on territoriality in the context of CSCW. Two territory

types were detected: group and personal. Later, Scott

et al. (2004) conducted an extensive study within a

non-digital tabletop environment to gain a deeper un-

derstanding of territoriality. As a result, they detected

a new territory type: storage territory. Additionally,

Scott et al. described in detail the characteristics of

individual territory types. Many other researchers in-

vestigated territoriality in tabletop environments, e.g.,

Tang et al. (2006), and some interactions techniques

were introduced, e.g., Moellers et al. (2011), Scott

et al. (2005), to support this concept.

Signiﬁcantly less research on territoriality took

place in the context of vertical displays. Azad et al.

(2012) investigated territoriality on public wall-sized

displays. In addition to the known territory types, they

detected a new territory type that describes display

regions avoided by the user. They called these ter-

ritories unused. Jakobsen and HornbÆk (2014) ob-

served territoriality on a large, vertical display. They

noted that participants frequently worked in parallel

without negotiating for space and shared the display

evenly. They also stated that territories are more criti-

cal for loosely coupled work than for tightly coupled.

Wallace et al. (2016) conducted an empirical study

where they explored users’ personal spaces around

large public displays and conﬁrmed the emergence of

different territory types.

Previous research shows that users tend to apply

territoriality even if no explicit tools are provided and

that users able to handle territoriality using social pro-

tocols and norms. However, the concept of territori-

ality is still not well understood. In particular, it is

not clear if the concept has a signiﬁcant impact on

intra-group interactions and if it should be consid-

ered during the design process of collaborative sys-

tems. On the one hand, providing supporting tools

and techniques might make the system more user-

friendly. On the other hand, implemented rules and

workﬂows might limit the interaction freedom of the

user resulting in decreased acceptance of the system.

Collaborative Coupling. Collaborative coupling

describes the process of user-user interaction for task

accomplishment and can be expressed, among others,

through collaboration tightness, coupling styles, user

roles, and task subdivision strategies. In general, re-

searchers subdivide collaborative coupling into two

ranges: tightly and loosely (Gutwin and Greenberg,

1998; Morris et al., 2004; Tse et al., 2004; Gutwin

and Greenberg, 2002). Within these, the intensity

level may vary depending on a coupling style. Orig-

inally, tightly coupled work was deﬁned as work that

barely could take place without user-user interaction,

while loosely coupled work describes rather a work-

ﬂow where users act independently, e.g., Gutwin and

Greenberg (1998, 2002), Salvador et al. (1996).

Tang et al. (2006) adjusted the term collaborative

coupling as how ”collaborators are involved and oc-

cupied with each other’s work” to highlight social

aspects of the phenomena. They conducted two ob-

servational studies in the context of the collabora-

tive exploration of ﬁxed spatial data around tabletops.

They could detect and describe six coupling styles

the groups used during the work (e.g., same problem,

same area).

Following, Isenberg et al. (2010) conducted an-

other exploratory study around a tabletop system,

where participants had to analyze 240 documents.

The study revealed eight different coupling styles that

were described based on participants’ data view and

personal interactions.

Jakobsen and HornbÆk (2014) recreated the ex-

ploratory study of Isenberg et al. (2010) using a mul-

titouch wall-sized display. Jakobsen et al. used codes

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

120

for visual attention and verbal communication to de-

scribe coupling.

Rogers and Lindley (2004) investigated group be-

havior around both vertical and horizontal interactive

displays. They observed that in the vertical display

scenario, the participants more frequently transitioned

to loosely-coupled work in comparison to the hori-

zontal display scenario. They could also observe the

interactor user role (the person who interacts with the

system).

Vogt et al. (2011) investigated group behavior dur-

ing collaborative sensemaking on an LHRD. They de-

scribed group behaviors concerning activities (e.g.,

extract and cluster) and user roles. In total, they iden-

tiﬁed two user roles: sensemaker and forager.

Collaborative coupling is a useful tool for descrip-

tion and understanding of intra-group behavior. Ex-

traction of collaborative coupling patterns might open

a possibility to create adaptive collaborative systems

(Sigitov et al., 2018). These systems will be able to

recognize intra-group states and adjust automatically

internal parameters to provide better support for indi-

vidual states. Another research direction would be to

create collaborative systems that are apt to manipulate

the collaborative process motivating or even enforcing

users to adopt a particular coupling pattern.

Workspace Awareness. Another essential ad-

vantage of LHRDs is that they are apt to provide ex-

tensive support for workspace awareness - the ability

to ”maintain awareness of others’ locations, activi-

ties, and intentions relative to the task and space”

(Gutwin and Greenberg, 1996). That allows all team

members to overview the entire process of level proto-

typing, see the produced results immediately, and in-

tervene rapidly if some problems occur. On the other

hand, workspace awareness can become a source of

recurrent distractions (Sigitov et al., 2016), thus af-

fecting the experience and efﬁciency of designers

negatively.

3 LHRD GROUPWARE FOR 2D

LEVEL PROTOTYPING

The software infrastructure we developed for collab-

orative 2D level prototyping on LHRDs consists of

four main components (see Figure 1): server, mobile

client, display client, and level editor. Mobile clients

can run on a mobile device, laptop, or a conventional

desktop computer. It provides an interactive inter-

face to users and allows them to create visual artifacts

in the form of map tiles. Display clients take com-

mands and messages from the server and pass them to

the level editor. The level editor application encapsu-

lates the display client component for communication

purposes. The server is a connecting link between

clients; therefore, clients communicate only with the

server or through the server, and no client does know

anything about other clients.

The system uses the following protocol to regis-

ter clients on the server. On each new connection, the

server puts the client into the waiting pool ﬁrst. Next,

it asks the client to identify itself; after that, the client

responses with its type and name. Depending on the

type of the client, the server chooses a communica-

tion protocol and allocate appropriate data structures

for the client. Subsequently, the server responses and

conveys a session-id to the client. That means that

the server registered the client. Finally, the client can

send data.

3.1 Mobile Client

The mobile client (Figure 1) represents an interaction

interface between the user and the system. Using a

device with the mobile client application running the

user can log in on the server, move her pointer on dis-

play, create map tiles, and put or delete tiles on the

map. Additionally, the mobile client application per-

forms data transformation of the user input. The data

transformation is necessary since the application can

run on different devices, e.g., tablet-PC or laptop.

The mobile client application consists of three

panels. The user can switch between Controller,

Craft, and Settings. The Settings panel allows the user

to adjust her pointer’s color and speed on the LHRD.

The Craft panel allows the user to create new tiles.

The panel contains the palette with different colors,

slider for brush size adjustment, canvas to draw, a text

ﬁeld to name the tile, and two buttons save and reset.

The Controller panel contains the menu bar, tiles bar,

interaction ﬁeld, recycle bin, and drop area. The menu

bar allows the user to hide and show the Settings and

Craft panels, quit the application, or delete a tile on a

map. The tiles bar contains all the tiles created by the

user or her co-workers. The user can drag a tile from

the tiles bar into the recycle bin to delete it, or into the

drop area to put the tile onto the map.

In our system, we implemented the discrete

coarse-and-precise pointer interaction technique simi-

lar to Nancel et al. (2013). The precise pointer has the

form and the size of a single tile, so it marks the area

on the map a tile will be put in. The coarse pointer,

on the other hand, has the size of a display unit, thus

providing a possibility for fast navigation between re-

mote areas on display. The precise pointer always

remains within the coarse pointer, thus moving the

coarse pointer will also move the precise pointer.

Effect of User Roles on the Process of Collaborative 2D Level Design on Large, High-resolution Displays

121

Figure 1: (left) System overview; (right) the mobile client application running on the tablet-PC.

The interaction ﬁeld in the mobile client applica-

tion allows controlling of both pointers. The user can

move the precise pointer by making the swipe gesture

with one ﬁnger, while with two ﬁngers, the user can

move the coarse pointer.

3.2 Server

The server application is responsible for multiple

tasks. It logs in a database every piece of information

that was received or sent by it, thus allowing for an in-

depth analysis of client interactions. It also backups

the current state of the system and distributes it to the

newly connected clients, thus ensuring the integrity

and reliability of the system.

The server also manages communication between

clients. Every time the server receives a data package,

it decides ﬁrst what clients are affected by this new

information, for instance, if the server receives a tile

then all clients in the environment have to be updated.

If, however, a mobile client reports on movement in

the interaction area, then only the display clients re-

ceive the notiﬁcation. Additionally, it can perform

data transformation, means transform data from a

client into an appropriate form for other clients. For

instance, the mobile client application can report on

one-ﬁnger swipe gesture passing a relative delta since

the last frame. In that case, the server has to accu-

mulate the values and command the display client to

move the precise pointer of the user one tile left if the

accumulated value exceeded a given threshold. Thus,

the continuous values from the mobile client become

a discrete value that can be understood by the display

client.

3.3 Level Editor and Display Client

The level editor application was implemented using a

Unity plug-in for LHRDs (Sigitov et al., 2015). The

level editor contains a simple 2D tilemap which is

shown on display and can be edited by users using

mobile clients. To conﬁgure the editor, the devel-

oper must provide the following values: the size of

tiles (which is also the size of the precise pointer); the

size of the display units (which is also the size of the

coarse pointer); width and height of the map.

Using the Unity plug-in for LHRDs, a set of vir-

tual cameras that mimic the physical conﬁguration of

the display is deﬁned. Each virtual camera becomes

assigned to an individual display unit. The level ed-

itor application consists therefore of n instances with

n equals the number of display units. All instances

are synchronized using mechanisms provided by the

plug-in.

One instance of the level editor application (called

manager) encapsulates the display client. The display

client is responsible for receiving commands from

the server and triggering the appropriate command in

the level editor (e.g., show/hide coarse/ﬁne pointer,

place/delete tile). The commands are triggered only

on the manager. It, in turn, performs the desired ac-

tion, subsequently changing the internal state of the

virtual world, and ﬁnally distributes the new internal

state, among other instances.

4 USER STUDY

With the study, we aimed to assess users’ attitude to-

wards LHRD-based systems for co-located collabora-

tion in the context of creative tasks, and investigate

the collaboration process in the context of creative

tasks. We also wanted to identify how a user inter-

face that forces users to adopt a speciﬁc role within

the team will affect the collaboration process, and get

feedback on the developed groupware for 2D level

prototyping to identify potential improvements. We

utilized the system described in section 3 and let par-

ticipant dyads build 2D game levels for 30 minutes.

Apart from the time constraint, we did not set any

other constraints.

4.1 Study Design

We used a between-groups study design. The in-

dependent variable was Condition with two levels,

namely Equal Roles and Different Roles. In the

Equal Roles condition, participants were provided

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

122

with identical interfaces. Thus both group members

could perform the same set of manipulations. In that

case, the participants could ignore the concept of user

roles entirely, or they could dynamically (re-)assign

roles.

In the Different Roles condition, the participants

were provided with two different interfaces. One in-

terface allowed only for the creation of visual assets,

while the other interface allowed for the placement

of visual assets only. Therefore, the participants had

either undertake the role of a 2D artist (creation of vi-

sual assets) or a level designer (placement of visual

assets). We used the same mobile client application

as in the Equal Roles condition. For 2D artists, how-

ever, we deactivated modules that are responsible for

interaction with the LHRD, while for level designers

we disabled the Craft panel.

During the study, we gathered qualitative data that

included surveys, ﬁeld notes, and video recordings.

In total, we gathered 368 minutes of video/audio data.

Moreover, we obtained quantitative data that encom-

passed the participants’ position, pointer positions,

and task-related system events.

4.2 Apparatus

We conducted the study on a large, curved tiled-

display (see Figure 2) comprising 35 displays (hence-

forth display units) ordered through a seven (column)

by ﬁve (row) grid. Each column has a relative angle

difference of 10 degrees along the Y-axis to adjacent

columns, as such, creating a slight curvature. The dis-

play units were 46” panels with a 1080p resolution.

The application for the display was implemented us-

ing the Unity game development platform with a spe-

cial plug-in for large displays (Sigitov et al., 2015).

We used an array of seven infrared cameras to

track users’ heads through head-worn helmets within

an area of around 20 square meters directly in front

of the display. For interaction purposes, we utilized

two identical Amazon Fire HD 10 tablets. The tablets

ran an application to control the pointer position on

the display and create visual assets. Use of the tablets

as interaction devices allowed us to untether partici-

pants, thus foster physical navigation, and to extend

operative display’s real estate by a highest and the

lowest row. In the case of touch-capable large dis-

plays, these areas would be hard or even impossible

to reach.

4.3 Procedure

A supervisor guided participants through the entire

study. First, the supervisor asked the participants to

Figure 2: Large, curved tiled-display used for the study. The

display shows a 2D level prototype created by a group of

participants.

ﬁll in a consent form and a demographics question-

naire. Next, the supervisor explained the task, the

procedure of the study, what equipment the partici-

pants will wear, and how and what data the system

will gather. In the case of Different Roles condition,

the supervisor also asked the participants to decide

who will undertake which role. If participants could

not decide, the roles were assigned randomly. Af-

terward, the supervisor demonstrated how the group-

ware works and invited the participants to try it out.

The participants had as much time as they needed

to get to know with the system. Finally, the super-

visor equipped the participants with helmets which

were used for position tracking and let them build a

2D game level for 30 minutes. During the study, the

supervisor observed the process and made ﬁeld notes.

After the time expired, the participants ﬁlled in the

questionnaires and supervisor conducted a short oral

interview.

4.4 Participants

We performed the study with 12 groups of two partic-

ipants each. Six groups must accomplish the task un-

der Equal Roles condition while the other six groups

accomplished the task under Different Roles condi-

tion. The participants of the Equal Roles condition

were aged between 21 and 41 years (M = 27.58;

SD = 6.05). The participants of the Different Roles

condition were aged between 20 and 35 years (M =

24.67; SD = 4.23). All participants had a normal or

corrected-to-normal vision. There were four female

and eight male participants in both conditions. Each

participant took part only once in the study. All par-

ticipants had an academic background (students or re-

search associates). Each participant received 10 Euros

for taking part in the study.

Effect of User Roles on the Process of Collaborative 2D Level Design on Large, High-resolution Displays

123

Figure 3: (left) Visualization of display usage [front view] and user movements [top view] in Equal Roles [ER] and Different

Roles [DR] condition. Gray grid represents the LHRD from the front. The green and red dots on the grid depict the position

of the tiles. The blue curved line represents the LHRD from the top. The yellow lines outline the tracking area in front of the

LHRD. The green and red lines inside the tracking area depict the movements of the users during the study. (right) Number

of created tiles in different conditions.

5 RESULTS AND DISCUSSION

In this section, we present and discuss the results ac-

quired through the analysis of gathered qualitative and

quantitative data. We used questionnaires to evalu-

ate users’ attitude towards the system and to receive

feedback on developed groupware. The majority of

questions provided a 7-point Likert scale for the an-

swer (from 1 = bad / strongly disagree to 7 = good /

strongly agree). There were also three questions with

free text answers where the participants could explain

what groupware functionality they missed, what inter-

action device they would prefer to use instead of the

provided one, and also leave some free comments.

To capture and analyze the collaboration process

and effects of the different interfaces on the process,

we used video recording, observations, ﬁeld notes,

oral interviews, and system logs. There was also a set

of 7-point Likert scale questions to acquire the par-

ticipants’ subjective perception towards the collabo-

ration process. Table 1 summarizes the questionnaire

results. In the following subsections, we ﬁrst present

results regarding the system and groupware. Next,

we go over to the observed design process and con-

sider results regarding physical navigation, territorial-

ity, collaborative coupling, and workspace awareness.

We provide the results separately for the Equal Roles

(ER) condition and Different Roles (DR) condition.

5.1 System Evaluation

Overall System and Groupware. The participants

found the system rather easy to use. They acknowl-

edged that the system is easy to learn and that one

can quickly become skillful with it.

Most participants stated that it was fun to use the

system; that the system makes the work more inter-

esting; and that they found that the system is suitable

for the provided task.

The participants, however, stated that they were

only partially satisﬁed with the system and that it only

partially worked the way they wanted. The oral inter-

views revealed that it was due to some missing func-

tionality the participants would prefer to have in the

system. The results also revealed that the participants

of the Different Roles condition felt more comfortable

working with the system, and that the participants of

the Equal Roles condition would rather prefer to have

a sitting accommodation while the participants of the

Different Roles condition could rather spare it.

The results suggest the provided groupware as a

solid basis for future development. It can be extended

to provide more sophisticated tools for users in terms

of drawing and asset placement. So, for instance, the

participants wished to replace the ﬁeld with the pre-

deﬁned colors with the RGB-wheel, to allow cloning

and editing of created tiles, the possibility to set mul-

tiple at once, and a eye-free technique for setting tiles.

Additionally, the groupware can be modiﬁed to sup-

port other creative tasks, like interior design. It is

also a good idea to provide sitting accommodation for

longer sessions. Overall, the participants expressed a

positive attitude towards the system and the system

suitability for the design task.

Large, High-resolution Display. The partici-

pants stated that the display is suitable for the pro-

vided task and for the co-located collaboration. Yet,

the participants of the Equal Roles condition were not

sure if a common desktop display would be more

suitable for the task. However, the participants of

Different Roles condition rather disagreed that a com-

mon display would be a better choice.

The participants felt in general comfortable work-

ing with the display. Yet, some participants felt phys-

ical discomfort during the task. Additionally, 11 out

of total 24 participants found the display rather too

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

124

Table 1: Questionnaire results: the participants assessed the system using a 7-point Likert scale with 1 = bad / strongly

disagree to 7 = good / strongly agree.

Equal Roles Different Roles

Mean SD Mean SD

Overall system and groupware

easy to use 5.33 1.30 5.66 1.37

easy to learn 6.41 0.67 6.67 0.65

I quickly become skillful with the system 5.58 1.31 5.83 1.19

The system is fun to use 5.58 0.99 5.58 1.73

The system makes the work more interesting 5.58 1.24 6.08 1.24

The system is suitable for the provided task 5.25 1.13 6.08 1.08

I am satisﬁed with the system 4.08 0.99 4.83 1.33

I felt comfortable working with the system 5.41 1.08 6.00 0.60

I would prefer to have a sitting accommodation 5.25 1.65 3.16 1.75

Large, high-resolution display

The display is suitable for the provided task 5.83 0.83 5.75 1.14

The display is suitable for co-located collaboration 6.08 0.51 6.25 0.62

A common desktop display would be more suitable 3.75 1.28 2.75 1.42

I felt comfortable working with the display 5.41 1.08 6.00 0.60

I felt physical discomfort while working with the display 3.25 2.05 2.42 1.16

I think the display is too large for the task 3.83 1.94 3.08 2.02

Interaction device

The interaction device is suitable for the task 5.4 1.07 5.08 1.38

It was easy to master the interaction device 6.25 0.75 5.50 1.17

I felt comfortable working with the interaction device 4.83 1.19 5.33 1.56

I felt physical discomfort while working with the interaction device 3.66 1.72 3.58 1.83

Perception of collaboration

I and my partner worked entirely cooperative 4.33 1.23 5.67 1.07

It was fun to work collaboratively 6.00 1.13 6.58 0.67

My partner distracted me during the work 1.75 0.62 1.33 0.65

large for the task.

Overall, the results showed that users consider

LHRDs as an acceptable visual interface for co-

located collaborative work and that they could imag-

ine working with such devices in the future. The par-

ticipants were satisﬁed with the display, its size, and

its curvature. Although some participants felt phys-

ical discomfort while working with the display, we

could detect during the oral interviews that it was due

to frequent switches from the tablet display to the

large display and vice versa. This problem is solv-

able through an eye-free technique for asset place-

ment. Generally, we can conclude that, if a system

provides a secondary display in addition to a LHRD,

then designers must ensure an interaction process that

keeps the number of switches between two displays

as low as possible.

During the oral interviews, the participants also

explained that in the given 30 minutes, it was practi-

cally impossible to utilize the entire display real es-

tate. Thus the display could be smaller, in their opin-

ion. If, however, they would have enough time, then

the size would be acceptable. We also asked the par-

ticipants if it would be better to have a ﬂat display

instead of curved. All participants stated that it would

be worse.

Interaction Device. Most participants agreed

that the provided mobile device is suitable for the

task and that it is easy to master. Most participants

felt rather comfortable working with the mobile de-

vice. Yet, some participants felt physical discom-

fort. Fifteen out of total 24 participants were satisﬁed

with the provided mobile device and would stick to

it; 5 out 24 participants would prefer to work with a

more lightweight tablet device and/or have a stylus

for better drawing; 4 out 24 participants would pre-

fer to have another interaction device/interface, e.g.

laser pointer, game controller, or touch capable large

display.

The results showed that the majority of users were

satisﬁed with the tablet PC as an interaction device.

Better acceptance could is achievable through the uti-

lization of more lightweight and more qualitative de-

vices with support for pen input. In contrast to large

displays with direct touch input, mobile devices pro-

vide better support for LHRDs in terms of reachabil-

Effect of User Roles on the Process of Collaborative 2D Level Design on Large, High-resolution Displays

125

ity of remote display areas and objects. Moreover,

they allow for interaction from a distance and mitigate

occlusion situations. Since user acceptance is high,

we would recommend designing interaction around

this kind of devices and as mentioned above, set focus

on eye-free interaction techniques.

5.2 Group Behavior

Physical Navigation. During the experiment, par-

ticipants did not frequently change their location in

front of the display (see Figure 3). Most of the time,

the participants positioned themselves in an overview

position, and physical navigation was reduced to head

and body rotations.

We ascribe the observed behavior to two factors:

the curvature of display and the size of tiles. Although

we made tiles relatively small (45px x 45px), the par-

ticipants could work well with that size from a dis-

tance. Additionally, the curvature of the display al-

lowed the participants to work comfortably on practi-

cally all display regions staying in the middle of the

working area. Thus, the participants did not depend

on location changes and could switch between work-

ing display regions display by merely rotating their

heads or bodies.

Overall, the impact of physical navigation was

marginal in both conditions.

Display Usage. Most participants preferred to

work in a comfortable display region. We could see,

however, that they utilized other display regions as

well. The observed behavior is not unexpected. Since

the participants had only 30 minutes to complete the

task, we did not expect that they would be able to

build a level which takes the entire display. We asked

them, however, to work as if they would have enough

time to do so. We did not prescribe where the par-

ticipants should start and in what direction the level

should expand. As Figure 3 depicts, most groups

(groups 2,3,4,5,6 in the Equal Roles condition, and

groups 1, 4, 6 in the Different Roles condition) worked

predominantly in the middle row of the display. Two

groups (group 1 in the Equal Roles condition and

group 3 in the Different Roles condition) tried to uti-

lize the entire display. One group started in the top

left corner of the display and the other group in the

bottom left corner.

Territoriality. The Different Roles condition did

not allow for territorial behavior on the shared display.

In the Equal Roles condition, however, we could ob-

serve distinctly territorial behavior in the early stages

of the experiment. Four out of six groups decided

to split the task by area after they have established

a common ground and agreed on what kind of level

they want to build. The participants divided then the

display into the left and right side and started to work

loosely from the outer display edges to the center.

Figure 3 depicts this behavior that was exposed by

groups 2,3,4, and 5 during the Equal Roles condition.

After the participants met, the territorial behavior di-

minished, and the participants split the task mostly

by a sub-task (e.g., one participant drew and placed

coins, while the other drew and placed plants).

We could not see any necessity for supporting ter-

ritoriality or providing special tools for territory man-

agement. Since the participants perceived the collab-

oration as tight, they did not feel threatened or dis-

turbed by the actions of the partner. One possible

improvement could be an overview territory on the

shared display to show all created tiles at once. That

will ensure a better workspace awareness and create a

new ground for more tightly coupled work.

Creation of Assets. Figure 3 depicts the ﬁnd-

ing regarding tiles creation. The analysis of the sys-

tem logs showed that the creation of assets took place

over the entire session. All groups in both condi-

tions demonstrated this behavior. The supervisor who

made ﬁeld notes also detected the behavior. We as-

cribe the behavior to the fact that the participants re-

peatedly received new ideas during the design pro-

cess, thus returning to the draw tile sub-task.

We could also detect that the groups in the Differ-

ent Roles condition created, on average, more tiles in

comparison to the groups in the Equal Roles condi-

tion. We assume that the distribution of roles created

a feeling of responsibility for the assigned task, thus

leading to more focused work which, in turn, results

in better productivity. However, the statistical analy-

sis using independent-samples t-test revealed that the

difference was insigniﬁcant.

Perception of Collaboration. The participants

felt working rather cooperatively and agreed that it

was fun to work collaboratively. They also did not

feel distracted by the partner. Only 3 out of 24 par-

ticipants stated that they would rather prefer to work

alone and that they would perform better if working

alone. All three participants belonged to the Equal

Roles condition.

Collaborative Coupling. To analyze the col-

laborative coupling behavior, we analyzed the video

recordings multiple times. We ﬁrst, identiﬁed peri-

ods of loosely and tightly coupled work, measured

the length of those periods, and calculated the over-

all time of loosely and tightly coupled work for each

group. Figure 4 shows the results of the analysis.

The visualization exposed noticeable difference

between the Equal Roles and Different Roles condi-

tions. In the Equal Roles condition the participants

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

126

Figure 4: Level design process for different conditions: (left) periods of loosely and tightly coupled work; (center) total time

for loosely and tightly coupled work; (top right) state diagram for the Equal Roles condition; (bottom right) state diagram for

the Different Roles condition

seemed to work more in a loosely coupled manner.

Thus, we conducted an independent-samples t-test

to compare the time the participants spent working

loosely coupled in different conditions. The analy-

sis revealed a signiﬁcant difference in the scores for

the Equal Roles condition (Mean = 1331.21, SD =

290.17) and the Different Roles condition (Mean =

908.48, SD = 174.75); t(10) = −3.06, p = 0.012. The

detected signiﬁcant difference indicates that utilized

interfaces can have a great effect on how users ap-

proach a common task and on how they shape their

collaboration. Especially in teams that contain users

who do not know each other well (or not at all) in-

terfaces that force tightly-coupled work might be of

advantage resulting in a more coordinated workﬂow

and more consistent results.

Next, we detected what sub-tasks the participants

approached during the identiﬁed periods. Based on

the gained information, we created state diagrams

that describe the collaboration processes for the Equal

Roles and Different Roles conditions. Figure 4 de-

picts the state diagrams. The results show that the

participants of the Equal Roles condition shaped the

collaboration process differently in comparison to the

Different Roles condition.

In the Equal Roles condition, the participants

started with establishment of a Common Ground.

They discussed what kind of level they want to build,

what tiles are necessary, who will create what tiles,

where the ground will be, and where they should start

to place tiles. Next, they moved on to the Draw a Tile

state working loosely-coupled. After the participants

created ﬁrst tiles, they could move on to the states

Place a Tile or Ponder Next Steps. In both states,

the participants worked in a loosely coupled manner.

They could also switch to the tightly coupled step Co-

ordinate Work, for instance, to share new ideas, ask

for advice, or propose new tiles.

In the Different Roles condition, the participants

started similarly with a discussion over general pa-

rameters. Subsequently, they moved on to the state

where 2D artist was drawing, and the level designer

was waiting for the tiles. After they created ﬁrst tiles,

the participants would either switch over to a state

where the level designer was placing the tiles, and

the 2D artist was waiting for new requests or to the

state where the level designer was building, and 2D

artist was continuing to draw more tiles. Additionally,

participants could also switch over to the coordina-

tion state to discuss what steps to take next. Overall,

the participants worked more coordinated and more

tightly in the Different Roles condition. Notably, the

participants playing the 2D artist role were involved

in the work of the level designer more deeply. It did

not always work the other way around. Some 2D

artist produced many assets without being asked for

them. That sometimes required additional coordina-

tion, since the level designer could not decipher the

meaning of tiles.

We also asked the participants of the Different

Roles condition if the idea of roles distribution does

make sense. Most participants (10 out of 12) found

the idea sensible on the basis that each of them could

focus on his/her sub-task and therefore provide more

qualitative results. They also stated that they did not

miss the functionality the co-participant had.

Groupware developers can utilize the concept of

User Roles to direct the collaboration process dur-

ing creative work and to push it towards more tightly

work. The roles must be well balanced to ensure that

all team members will have enough work for the en-

tire session. Otherwise, users might become a feeling

Effect of User Roles on the Process of Collaborative 2D Level Design on Large, High-resolution Displays

127

of being expendable, and the work process might be-

come less satisfying. We observed such a situation by

one group where the participant who was playing the

2D artist role experienced periods of idleness.

Workspace Awareness. Finally, we also

looked for evidence of workspace awareness during

the study. Two sources could potentially increase

workspace awareness of the participants: the large

display and the mobile devices. The mobile devices

were synchronized, which means that if one partici-

pant created a tile, it would appear in the library panel

on the tablet of the co-participant. Additionally, the

panel will automatically scroll down, so the tile be-

comes visible.

The supervisor could multiple times observe that

the participants laughed or smiled when a new tile,

created by the co-participant, appeared in their library

panel. In some cases, however, we could detect that

this also leads to interferences. For instance, in the

Different Roles condition, one 2D artist always waited

until the level designer was done with tile placement

before he submitted a new tile. Otherwise, due to the

auto-scroll function, the level designer must navigate

in the tiles library to the tile he was currently using.

During the oral interviews, We also asked the par-

ticipants if they occasionally observed what the part-

ner was doing and if that affected them by any means.

All participants answered in afﬁrmative. Most partic-

ipants (22 out of 24) stated that they were inspired by

what the partner was doing and received new ideas.

Two participants also stated that they observed the

partner actions to avoid interferences. One partici-

pant stated that he was in no way affected by what the

partner was doing.

6 CONCLUSION

In this paper, we presented groupware for co-located

collaborative 2D level design on LHRDs using tablet

PCs for interaction. The evaluation results revealed

a positive users attitude and high level of acceptance

towards the system. The results also identiﬁed the di-

rections of future improvements, like more sophisti-

cated tiles editor, more lightweight and more quali-

tative tablet-PCs, as well as eye-free techniques for

asset placement. The implementation of the pre-

sented groupware allows an arbitrary number of users.

Moreover, it can be easily modiﬁed to support differ-

ent design tasks.

Furthermore, we presented a study comparing

users behavior while working with and without spe-

ciﬁc roles distribution. The study showed that

whether users have to work with identical interfaces

or with different interfaces which force them to un-

dertake a speciﬁc role signiﬁcantly affects the collab-

oration process. Providing interfaces that enforce the

distribution of roles can help to shape a more tightly-

coupled and coordinated workﬂow as well as achieve

more consistent results. Moreover, the majority did

not perceive such limitations as awkward and wel-

comed the conﬁguration. When designing groupware

for co-located creative tasks on LHRDs, this may cre-

ate new opportunities. Groupware developers can ex-

ploit this knowledge to create systems which will be

able to affect the collaboration process.

REFERENCES

Andrews, C., Endert, A., and North, C. (2010). Space to

think: large high-resolution displays for sensemaking. In

Proceedings of the 28th international conference on Hu-

man factors in computing systems - CHI ’10, page 55,

New York, New York, USA. ACM Press.

Azad, A., Ruiz, J., Vogel, D., Hancock, M., and Lank, E.

(2012). Territoriality and behaviour on and around large

vertical publicly-shared displays. In Proceedings of the

Designing Interactive Systems Conference on - DIS ’12,

page 468, New York, New York, USA. ACM Press.

Ball, R., North, C., and Bowman, D. A. (2007). Move to

improve: Promoting physical navigation to increase user

performance with large displays. In Proceedings of the

SIGCHI conference on Human factors in computing sys-

tems - CHI ’07, number Figure 1, page 191, New York,

New York, USA. ACM Press.

Gutwin, C. and Greenberg, S. (1996). Workspace aware-

ness for groupware. In Conference companion on Hu-

man factors in computing systems common ground - CHI

’96, pages 208–209.

Gutwin, C. and Greenberg, S. (1998). Design for individ-

uals, design for groups: tradeoffs between power and

workspace awareness. In Proceedings of the 1998 ACM

conference on Computer supported cooperative work -

CSCW ’98, pages 207–216, New York, New York, USA.

ACM Press.

Gutwin, C. and Greenberg, S. (2002). A descriptive frame-

work of workspace awareness for real-time groupware.

Computer Supported Cooperative Work, 11(3-4):411–

446.

Hinrichs, U., Carpendale, S., and Scott, S. D. (2006). Eval-

uating the effects of ﬂuid interface components on table-

top collaboration. page 27. ACM Press.

Isenberg, P., Fisher, D., Morris, M. R., Inkpen, K., and

Czerwinski, M. (2010). An Exploratory Study of Co-

located Collaborative Visual Analytics around a Table-

top Display. In Visual Analytics Science and Technology

- VAST’10, Los Alamitos, CA, USA. IEEE Computer So-

ciety.

Jakobsen, M. R. and HornbÆk, K. (2014). Up close and

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

128

personal. ACM Transactions on Computer-Human Inter-

action, 21(2):1–34.

Jakobsen, M. R. and Hornbæk, K. (2016). Negotiating for

Space? Collaborative Work Using a Wall Display with

Mouse and Touch Input. In Proceedings of the 2016 CHI

Conference on Human Factors in Computing Systems -

CHI ’16, pages 2050–2061, New York, New York, USA.

ACM Press.

Liu, C., Chapuis, O., Beaudouin-Lafon, M., and Lecolinet,

E. (2017). CoReach: Cooperative Gestures for Data Ma-

nipulation onWall-sized Displays. In Proceedings of the

2017 CHI Conference on Human Factors in Computing

Systems - CHI ’17, pages 6730–6741, New York, New

York, USA. ACM Press.

Machaj, D., Andrews, C., and North, C. (2009). Co-

located Many-Player Gaming on Large High-Resolution

Displays. In 2009 International Conference on Compu-

tational Science and Engineering, pages 697–704. IEEE.

Moellers, M., Bohnenberger, R., Deininghaus, S., Zimmer,

P., Herrmann, K., and Borchers, J. (2011). TaPS Wid-

gets: tangible control over private spaces on interactive

tabletops. page 773. ACM Press.

Morris, M. R., Ryall, K., Shen, C., Forlines, C., and Vernier,

F. (2004). Beyond ”social protocols”: multi-user coordi-

nation policies for co-located groupware. In Proceedings

of the 2004 ACM conference on Computer supported co-

operative work Pages - CSCW’04, page 262. ACM Press.

Nancel, M., Chapuis, O., Pietriga, E., Yang, X.-D., Irani,

P. P., and Beaudouin-Lafon, M. (2013). High-precision

Pointing on Large Wall Displays Using Small Handheld

Devices. CHI ’13, pages 831–840, Paris, France. ACM.

Polley, R. B. and McGrath, J. E. (1984). Groups: Interac-

tion and Performance. Administrative Science Quarterly,

29(3):469.

adle, R., Jetter, H.-C., Butscher, S., and Reiterer, H.

(2013). The effect of egocentric body movements on

users’ navigation performance and spatial memory in

zoomable user interfaces. pages 23–32. ACM Press.

Rogers, Y. and Lindley, S. (2004). Collaborating around

vertical and horizontal large interactive displays: which

way is best? Interacting with Computers, 16(6):1133–

1152.

Ryall, K., Forlines, C., Shen, C., and Morris, M. R. (2004).

Exploring the effects of group size and table size on in-

teractions with tabletop shared-display groupware. In

Proceedings of the 2004 ACM conference on Computer

supported cooperative work - CSCW ’04, page 284, New

York, New York, USA. ACM Press.

Sabri, A. J., Ball, R. G., Fabian, A., Bhatia, S., and North,

C. (2007). High-resolution gaming: Interfaces, notiﬁca-

tions, and the user experience. Interacting with Comput-

ers, 19(2):151–166.

Salvador, T., Scholtz, J., and Larson, J. (1996). The Den-

ver model for groupware design. ACM SIGCHI Bulletin,

28(1):52–58.

Scott, S. D., Carpendale, M. S. T., and Habelski, S. (2005).

Storage bins: Mobile storage for collaborative tabletop

displays. IEEE Computer Graphics and Applications,

25(4):58–65.

Scott, S. D., Grant, K. D., and Mandryk, R. L. (2003). Sys-

tem Guidelines for Co-located, Collaborative Work on a

Tabletop Display. In Proceedings of the eighth confer-

ence on European Conference on Computer Supported

Cooperative Work - ECSCW’03, pages 159–178, Dor-

drecht. Springer Netherlands.

Scott, S. D., Sheelagh, C., and Inkpen, K. M. (2004). Ter-

ritoriality in collaborative tabletop workspaces. Pro-

ceedings of the 2004 ACM conference on Computer sup-

ported cooperative work - CSCW’04, pages 294 – 303.

Sigitov, A., Kruijff, E., Trepkowski, C., Staadt, O., and

Hinkenjann, A. (2016). The Effect of Visual Distractors

in Peripheral Vision on User Performance in Large Dis-

play Wall Systems. In Proceedings of the 2016 ACM

International Conference on Interactive Surfaces and

Spaces - ISS’16, pages 241–249. ACM Press.

Sigitov, A., Scherfgen, D., Hinkenjann, A., and Staadt,

O. (2015). Adopting a Game Engine for Large,

High-Resolution Displays. Procedia Computer Science,

75:257–266.

Sigitov, A., Staadt, O., and Hinkenjann, A. (2018). To-

wards Intelligent Interfaces for Mixed-Focus Collabora-

tion. In Adjunct Publication of the 26th Conference on

User Modeling, Adaptation and Personalization - UMAP

’18, pages 287–292, New York, New York, USA. ACM

Press.

Tang, A., Tory, M., Po, B., Neumann, P., and Carpendale, S.

(2006). Collaborative coupling over tabletop displays. In

Proceedings of the SIGCHI conference on Human Fac-

tors in computing systems - CHI ’06, page 1181.

Tang, J. C. (1991). Findings from observational studies

of collaborative work. International Journal of Man-

machine studies, 34(2):143–160.

Tse, E., Histon, J., Scott, S. D., and Greenberg, S. (2004).

Avoiding interference: how people use spatial separation

and partitioning in SDG workspaces. In Proceedings of

the 2004 ACM conference on Computer supported coop-

erative work Pages - CSCW’04, page 252. ACM Press.

Vogt, K., Bradel, L., Andrews, C., North, C., Endert, A.,

and Hutchings, D. (2011). Co-located collaborative

sensemaking on a large high-resolution display with mul-

tiple input devices. In Lecture Notes in Computer Sci-

ence (including subseries Lecture Notes in Artiﬁcial In-

telligence and Lecture Notes in Bioinformatics), volume

6947 LNCS, pages 589–604.

von Zadow, U., B

osel, D., Dam, D. D., Lehmann, A., Reip-

schl

ager, P., and Dachselt, R. (2016). Miners. In Pro-

ceedings of the 2016 ACM on Interactive Surfaces and

Spaces - ISS ’16, pages 235–240, New York, New York,

USA. ACM Press.

Wallace, J. R., Iskander, N., and Lank, E. (2016). Creating

Your Bubble. In Proceedings of the 2016 CHI Confer-

ence on Human Factors in Computing Systems - CHI ’16,

pages 2087–2092, New York, New York, USA. ACM

Press.

Effect of User Roles on the Process of Collaborative 2D Level Design on Large, High-resolution Displays

129