Content uploaded by Kangsoo Kim
Author content
All content in this area was uploaded by Kangsoo Kim on Nov 14, 2017
Content may be subject to copyright.
The Impact of Avatar-Owner Visual Similarity on Body
Ownership in Immersive Virtual Reality
Dongsik Jo
VR/AR Research Group, Electronics and
Telecommunications Research Institute (ETRI)
dongsik@etri.re.kr
Kangsoo Kim
Gregory F. Welch
Department of Computer Science, University of Central
Florida
Woojin Jeon
Yongwan Kim
Ki-Hong Kim
VR/AR Research Group, ETRI
Gerard Jounghyun Kim
Department of Computer Science and Engineering, Korea
University
gjkim@korea.ac.kr(Correspondence)
ABSTRACT
In this paper we report on an investigation of the eects of a self-
avatar
'
s visual similarity to a user
'
s actual appearance, on their
perceptions of the avatar in an immersive virtual reality (IVR) expe-
rience. We conducted a user study to examine the participant
'
s sense
of body ownership, presence and visual realism under three levels
of avatar-owner visual similarity: (L1) an avatar reconstructed from
real imagery of the participant
'
s appearance, (L2) a cartoon-like
virtual avatar created by a 3D artist for each participant, where the
avatar shoes and clothing mimic that of the participant, but using
a low-delity model, and (L3) a cartoon-like virtual avatar with a
pre-dened appearance for the shoes and clothing. Surprisingly,
the results indicate that the participants generally exhibited the
highest sense of body ownership and presence when inhabiting the
cartoon-like virtual avatar mimicking the outt of the participant
(L2), despite the relatively low participant similarity. We present our
experiment and main ndings, also, discuss the potential impact of
a self-avatar's visual dierences on human perceptions in IVR.
CCS CONCEPTS
•Human-centered computing →Virtual reality
;
•Comput-
ing methodologies →Virtual reality ;
KEYWORDS
Virtual reality, self-avatar, presence, body ownership, HMD
ACM Reference Format:
Dongsik Jo, Kangsoo Kim, Gregory F. Welch, Woojin Jeon, Yongwan Kim,
Ki-Hong Kim, and Gerard Jounghyun Kim. 2017. The Impact of Avatar-
Owner Visual Similarity on Body Ownership in Immersive Virtual Reality.
In Proceedings of ACM VRST conference (VRST ’17). ACM, New York, NY,
USA, 2 pages. https://doi.org/10.1145/3139131.3141214
Permission to make digital or hard copies of part or all of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for prot or commercial advantage and that copies bear this notice and the full citation
on the rst page. Copyrights for third-party components of this work must be honored.
For all other uses, contact the owner/author(s).
VRST ’17, November 2017, Gothenburg ,Sweden
©2017 Copyright held by the owner/author(s).
ACM ISBN 978-1-4503-5548-3/17/11.
https://doi.org/10.1145/3139131.3141214
Figure 1: Three levels of self-avatar similarity to its human
owner in IVR. (Left) L1-High: a realistically reconstructed
avatar from real imagery of the owner. (Middle) L2-Medium:
a cartoon-like virtual character wearing the same colored
out�t as the owner. (Right) L3-Low: a cartoon-like virtual
character with the pre-determined default out�t.
1 INTRODUCTION
In immersive virtual reality (IVR), a self-avatar representing the
owner (human user) has shown to have positive eects in his/her
sense of presence and the quality of interaction [Steed et al
.
2016].
While self-avatars are typically generated using a pre-determined
form with developed by 3D artists, it has become more practical to
create a self-avatar model precisely and instantly via an automatic
3D reconstruction process, thanks to technical innovations in si-
multaneous depth sensing and human motion tracking [Feng et al
.
2017]. In other words, it has gotten easier to create a self-avatar
that looks the same as its owner in IVR via a head-mounted display
(HMD), e.g., an avatar with the same colors and materials as the
owner's outt.
Many researchers have studied the eects of an avatar
'
s char-
acteristics on human perceptions of the avatar, such as a visual
quality of the avatar [Wu et al
.
2014]. Although some have studied
the eects of appearance between avatar and its owner [Jo et al
.
2017], more comprehensive studies looking into inuence (e.g. to
sense of ownership and presence) of variety of aspects including the
avatar-owner are needed. Therefore, we experimentally investigate
how a self-avatar
'
s visual similarity to its owner, particularly in
the appearance of clothes and shoes, can inuence the perceptions
with respect to the avatar. In our experiment, we examine each
Dongsik Jo, Kangsoo Kim, Gregory F. Welch, Woojin Jeon, Yongwan Kim, Ki-Hong Kim, and Gerard Jounghyun Kim. 2017.
The Impact of Avatar- Owner Visual Similarity on Body Ownership in Immersive Virtual Reality.
In Proceedings of ACM VRST conference (VRST ’17). ACM, New York, NY, USA, 2 pages.
https://doi.org/10.1145/3139131.3141214
VRST ’17, November 2017, Gothenburg ,Sweden D. Jo et al.
participant
'
s sense of body ownership, presence, and avatar realism
in an IVR environment via a head-mounted display (HMD), under
three dierent levels of avatar-owner visual similarity as shown in
Fig. 1 below and described next.
2 EXPERIMENT
We designed a single-factor (i.e., levels of avatar-owner similarity)
within-subjects experiment. The three test conditions by the simi-
larity are: (L1) an avatar comprising a 3D mesh model reconstructed
from real imagery of the participant, whose raw point cloud data
was captured by a depth sensor, (L2) a cartoon-like virtual character-
based avatar created by a 3D artist using colors/patterns that match
the participant
'
s real outt, (L3) a cartoon-like virtual character
wearing a pre-determined outt. See the examples in Fig. 1. We
initially hypothesized that participant
'
s (or owner
'
s) perceptions in
terms of the avatar
'
s visual similarity would be the highest in L1
and the lowest in L3.
For the experiment, the participant wore an HMD (Oculus Rift)
to experience an IVR version of a large virtual cave, with a life-
sized self-avatar seen from a rst-person viewpoint. During the
experiment, each of the 12 participants were asked to perform one
condition per day in a random order to eliminate bias, and follow
the avatar
'
s body postures operated by an experimenter behind the
curtain (Wizard-of-Oz) to help see the participant
'
s body in detail.
Then, they were also asked to observe a red sphere that randomly
appeared on the body of the avatar, in an attempt to make them
aware of the avatar
'
s visual appearance, and hence the similarity
to their real appearance.
To assess participant
'
s sense of body ownership, presence, and
avatar realism among the dierent conditions, we used a subjective
survey instrument comprising 7-point Likert scale questions. Ex-
amples of the survey questions include “I felt like the body I saw in
the virtual world was my body,“for the sense of body ownership
from [Lugrin et al
.
2015] to evaluate visual resemblance such as
a robot-typed avatar, and “How strong was your sense of ‘being
there‘in the virtual environment?“for the sense of presence from
[Dinh et al
.
1999] . For the overall avatar realism, we simply asked
whether they felt the avatar was realistic or not (1: very unrealistic
to 7: very realistic).
3 RESULTS AND DISCUSSION
Prior to the experiment, we expected that L1 (a reconstructed avatar)
would induce the highest sense of presence, realism and thus body
ownership. First, a one-way ANOVA analysis for the three condi-
tions revealed signicant main eects of all factors on all dependent
variables, such as body ownership (F(2,33) = 39.41, p<0.05), presence
(F(2,33) = 13.49, p<0.05), and Overall visual realism (F(2,33) = 11.68,
p<0.05). Surprisingly, the results showed that L2, with relatively
lower similarity compared to L1, induced a higher sense of own-
ership and presence among the three tested conditions. See Fig. 2.
Most of the participants noticed gaps in the point cloud comprising
the reconstructed L1 model, and stated that they felt the avatar was
rather heterogeneous or inharmonious to the surrounding virtual
environment, seeming to indicate an eect related to the Uncanny
Valley [Mori 1970]. These results could indicate that it is sucient
to establish body ownership by just capturing a set of essential
Figure 2: Survey results on the sense of (a) body ownership,
(b) presence, and (c) overall visual realism.
salient qualities/features for the user (e.g. clothing, main facial fea-
tures, body size), and taking into consideration the plausibility of
the avatar appearance in the context of the virtual environment.
4 CONCLUSIONS
In this paper, we experimentally investigated the eects of a self-
avatar
'
s visual similarity (in three levels) to the sense of body own-
ership, presence, and overall visual realism. Our study showed that
participants generally exhibited a relatively high sense of body
ownership and presence in condition (L2) where the self-avatar
was modeled in a cartoon-like fashion, when other physical features
were preserved, such as the features of the participant
'
s outt. The
results could indicate a counterintuitive eect where, like the Un-
canny Valley, the most realistic appearance does not always provide
the highest sense of presence, and how one should consider the
plausibility of the avatar appearance in the context of the virtual
environment.
In the near future, we will continue to explore the eects of
other related factors with the signicant number of users, such as
avatar-owner motion similarity and realistic behavioral qualities
(visual-motor synchrony). We will also explore more quantitative
and objective measures, e.g., physiological signals.
ACKNOWLEDGMENTS
This work was supported by IITP grant funded by the Korea gov-
ernment (MSIT) (No.R0118-16-1007, Immersive content experience
enabling multi-user interaction based on high-speed and precise
tracking in an indoor VR space) and also in part by IITP/MSIP (No.
2015-0-00565, Development of vulnerability discovery technologies
for IoT software security)
REFERENCES
H. Q. Dinh, N. Walker, C. Song, A. Kobayashi, and L. F. Hodges. 1999. Evaluating
the Importance of Multi-Sensory Input on Memory and the Sense of Presence in
Virtual Environments. In In Proceedings of the IEEE Virtual Reality. 222–228.
A. Feng, E. Suma, and A. Shapiro. 2017. Just-in-time, viable, 3-D avatars from scans.
Computer Animation and Virtual Worlds 28, 3-4 (2017), e1769.
D. Jo, K.-H. Kim, and G. J. Kim. 2017. Eects of avatar and background types on
users
'
co-presence and trust for mixed reality-based teleconference systems. In In
Proceedings the 30th Conference on Computer Animation and Social Agents. 27–36.
J.-L. Lugrin, J. Latt, and M. E. Latoschik. 2015. Anthropomorphism and illusion of virtual
body ownership. In In Proceedings the 25th International conference on arti�cial
reality and telexistence. 1–8.
M. Mori. 1970. The Uncanny Valley. Energy 7, 4 (1970), 33–35.
A. Steed, Y. Pan, F. Zisch, and W. Steptoe. 2016. The impact of a self-avatar on cognitive
load in immersive virtual reality. In Proceedings of the Virtual Reality. 67–76.
Y. Wu, S.-V. Babu, R. Armstrong, J.-W. Bertrand, J. Luo, T. Roy, S.-B. Daily, L.-C. Dukes,
L.-F. Hodges, and T. Fasolino. 2014. Eects of virtual human animation on emotion
contagion in simulated inter-personal experiences. IEEE Transactions on Visualiza-
tion and Computer Graphics 20, 4 (2014), 626–626.
Dongsik Jo, Kangsoo Kim, Gregory F. Welch, Woojin Jeon, Yongwan Kim, Ki-Hong Kim, and Gerard Jounghyun Kim. 2017.
The Impact of Avatar- Owner Visual Similarity on Body Ownership in Immersive Virtual Reality.
In Proceedings of ACM VRST conference (VRST ’17). ACM, New York, NY, USA, 2 pages.
https://doi.org/10.1145/3139131.3141214
