Upload
oscar
View
212
Download
0
Embed Size (px)
Citation preview
CollARt: a Tool for Creating 3D Photo Collages Using Mobile Augmented Reality
Asier Marzo Public University of Navarre
Pamplona, Spain
Oscar Ardaiz Public University of Navarre
Pamplona, Spain
ABSTRACT
A collage is an artistic composition made by assembling different
parts to create a new whole. This procedure can be applied for
assembling tridimensional objects. In this paper we present
CollARt, a Mobile Augmented Reality application which permits
to create 3D photo collages. Virtual pieces are textured with
pictures taken with the camera and can be blended with real
objects. A preliminary user study (N=12) revealed that
participants were able to create interesting works of art. The
evaluation also suggested that the possibility of itinerantly mixing
virtual pieces with the real world increases creativity.
Categories and Subject Descriptors
H.5.m [Information Interfaces and Presentation]: Misc
Keywords
Collages; mobile; augmented reality; creativity
1. INTRODUCTION The collage technique consists in sticking together different
pieces. This technique has been used by some famous artists such
as Pablo Picasso and Marcel Duchamp. Nowadays it is properly
established and used by both art professional and amateurs.
Collages are composed of pieces from other works of art, pictures,
newspaper clips or bits of colored papers. All of these elements
are glued on a canvas forming a shallow creation. Nonetheless,
pieces may be glued tilted or perpendicularly in order to create a
tridimensional work. In the same way, stacking clippings of
pictures gives some depth to collages. Extending this idea, some
collages are made with physical objects or mock-ups; however,
their assemblage is more difficult.
Digital technologies can be used for reducing required technical
level to execute certain creative activities or to increase the
effectiveness and quality of creations. Technology lowers barriers
or gives new possibilities to artistic creations. It makes very
interesting the idea of creating digital counterparts of traditional
artistic techniques. For instance, Pottery [16] permits to mold and
decorate different kinds of virtual pots, vases or jugs. Digital
adaptations may not have the same haptic response as the real
activity. Nevertheless, they are endowed with additional features.
Options such as undo, redo and quick preview of outcomes
improve exploration. Additionally, collaboration with other users
and sharing through social networks are common features of
present applications. Exploration, collaboration and open
interchange are essential for creativity support tools [17].
Augmented reality is a visualization technique which overlaps
virtual content over the real world [1]. Advances made in the
recent years permit to implement this technique in handheld multi-
touch devices.
We have developed CollARt, a Mobile Augmented Reality
(MAR) application for creating 3D photo collages. A paper
marker is used for determining the device position. The paper is
the canvas in the real and in the virtual world. Therefore, the
application provides intuitive navigation and placement of new
pieces for the user. Furthermore, mobile platforms are widely
available and support an itinerant way of creating works of art.
2. RELATED WORK A number of applications allow the users to create collages
automatically. For instance, CollageMachine [11] assembles
images fetched from the Internet based on the preferences and
interests of the user. Similarly, StainedGlass [7] extracts frames
from a video and creates a digest of it in form of a photo
composition. Setting a target illustration an algorithm can place a
set of images in an empty canvas trying to fit the shapes, colors
and textures of the target picture [9]. In the same way as
Arcimboldo did in his paintings. The same principle can be
applied for assembling a set of tridimensional objects to conform
a target 3D volume [6]. PhotoPieces is a mobile application [15]
in which the users create photo collages filling with pictures a
template made of different regions. Templates are chosen from a
library or made by segmenting a photo.
Augmented Reality has been used for painting virtually over the
surface of real objects using a head mounted display (HMD) and
an electronic pencil [8]. Different textures can be used for the
stroke as well as solid colors or patterns captured from the real
world. In addition, virtual objects with premade textures can be
attached to the real entities. A field study conducted with a similar
system in an architecture school [10] revealed several benefits
derived from its tangibility; however a projector was used instead
of a HMD. Additionally, the system permitted to attach different
virtual objects to tangible markers improving the configurability
[2]. Small planar markers or panoramic views can be augmented
with virtual objects [12] and shared with other users. That system
is composed of a mobile augmented reality application and a
server which stores the creations. ColAR [5] recognizes the
outlines of a picture from a set of predefined ones. It can
reproduce a 3D animation using the real filling that the users
painted as textures for the objects. ARPaint [13] can augment a
picture permitting to trace strokes on the plane which contains it.
Our work differs from previous ones as CollARt is a multi-touch
mobile augmented reality application which merges virtual objects
with real objects. Furthermore, it permits to create tridimensional
photo collages naturally and its workflow encourages exploration.
Permission to make digital or hard copies of all or part of this work for personal
or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice
and the full citation on the first page. Copyrights for components of this work
owned by others than ACM must be honored. Abstracting with credit is
permitted. To copy otherwise, or republish, to post on servers or to redistribute
to lists, requires prior specific permission and/or a fee. Request permissions
from [email protected].
MM’13, October 21–25, 2013, Barcelona, Spain. Copyright © 2013 ACM 978-1-4503-2404-5/13/10...$15.00.
http://dx.doi.org/10.1145/2502081.2502154
585
3. COLLART CollARt is a mobile application which allows the user to create
3D compositions. Using augmented reality, virtual objects can be
blended with the real world. Pieces are textured with pictures
taken with the camera (Figure 1). When the paper marker is
detected by the application it shows the live feedback of the
camera and the virtual pieces overlapped with it. By touching the
screen the menu icons toggle visible, permitting to undo, redo,
clear, select or add a new piece.
Pressing the add icon makes the camera view to come out for
taking a picture. Afterwards the placement options permit you to
rotate, scale, adjust the texture or select between different shapes
for the current piece (Figure 2). Planar pieces such as squares,
circles or rectangles as well as tridimensional volumes, namely
cubes, spheres or pyramids are available. For the rotation mode
we used the interaction technique defined in [14]. The position
and rotation of the piece relative to the device will remain
constant. Additionally, one finger can be used to perform an
Arcball rotation in view space, and two fingers to rotate or change
the depth of the object across the view axis. Adjusting the texture
allows the user to modify the part of the previously taken picture
which is mapped to the current piece.
Whenever the marker is detected by the camera and placing mode
is active, the tick button appears. Pressing it will place a piece on
the current position. As a result, the marker is only needed to
place pieces or to observe the ongoing creation. Pressing the finish
button will return to the main menu. By pressing the select icon
the application switches to select mode. In this mode a crosshair is
shown, permitting to select any object for modifying it.
Figure 1. Tridimensional pieces textured with pictures taken
with the camera.
Figure 2. Placing one piece using the interface as seen by a
user on the mobile device screen.
Virtual objects are blended with real objects. Therefore, the real
pieces occlude and clip virtual objects (Figure 3). Visual
recognition in real time of the world objects would require a
considerable time of processing or to paint them with markers.
Consequently, several premade scenes were created, defining
virtual replicates of the real objects.
CollARt has been implemented using C++, OpenGL and GLSL.
The target platform is iOS and the library Vuforia [17] has been
used for tracking the marker.
Figure 3. A creation which blends virtual pieces with a real
canister.
4. USER EVALUATION A preliminary evaluation was conducted with 12 people (3 women
and 9 men), aged between 18 and 43 (M=27.75, SD=7.08).
Participants had heterogeneous professions and computer
experience in some cases. Two users were left-handed. Four of the
participants had formal artistic education and two of them worked
as professional artists.
The main purpose of the evaluation was to check if blending
virtual elements with real objects and itinerant creations are
interesting properties to the users. As a general objective we tried
to determine whether users were able to create original collages
with artistic value. Additionally, we aimed to detect usability
problems and annotate interesting behaviors of the users.
Searching for more concrete uses of the tools was another goal of
the study.
Participants performed the evaluation individually and without
time constraints. At the beginning they had to complete the
training activity. In it, users were asked to place four cubes at each
corner of the canvas and two squares perpendicularly between
them on the middle. The purpose was to teach the users how to
use the basic functions of the application. Accordingly, they were
assisted during it.
Four tasks were designed: table, vertical, cylinder and on the
move. In the table and vertical tasks the users had to create a
collage having no restriction on shapes, pictures or time. In the
table task the canvas rested horizontally on a table whereas in the
vertical task it was stuck to a wall, this simple orientation change
may affect usability and creations. The cylinder task consisted of
placing a real canister in the middle of the canvas. The real
cylinder clipped and hid virtual pieces influencing the collage as it
586
was another existing piece. In the on the move task, using indoor
and outdoor pictures was mandatory.
Users were split randomly into two groups. The first group
performed the training activity followed by the table and vertical
tasks. The second group did the training activity and subsequently
the cylinder and the on the move tasks. Therefore, group 2 was
more focused on exclusive features of the application such as
blending virtual pieces with reality and on the move use.
Figure 4. Various collages made by the participants.
Afterwards, all the participants filled in a questionnaire for
measuring the system usability (SUS) [3] and a series of question
for determining CollARt’s value as a creativity support tool (CSI)
[4]. Both SUS and CSI range from 0 to 100, the higher the value
the more usable or creativity supportive. In addition, they had to
score the ease and originality of each task using a one to five
Likert scale. Various open questions were answered by the users,
namely: what aspects have been more interesting?, what
difficulties have arisen using the tool?, in what places or context
would you use the tool? and can you devise a game to be played
with the tool?. T-tests were used for comparing both CSI and SUS
scores between the first group and second group. One-way-
ANOVA was used for the Ease and Originality scores between the
tasks as each participant performed only two of them.
The average CSI score reported by the participants was 71.66
(SD=7.52) for the group 1 and 73.88 (SD=7.57) for the group 2.
The score is higher in the group 2; however, there was not a
significant difference between the groups (t(10) = -0.50, p > 0.05).
The average SUS of the questionnaires was 72.08 (SD=11.77) for
the group 1 and 80.83 (SD=15.46) for the group 2. The score is
higher in the group 2. Nonetheless, there was not a significant
difference between the groups (t(9) = -1.10, p > 0.05).
The easiness of completing each task was 4.08 (SD=1.24) for the
training, 4.16 (SD=0.40) for the table task, 4.16 (SD=0.75) for the
vertical task, 4.16(SD=1.16) for the cylinder and 4.33 (SD=0.81)
for the on the move task; there was no significant difference
between the four tasks (F3,20 = 0.06, p > 0.05). The perceived
originality of the creation for each task was 2.25 (SD=1.05) for
the training, 3.5 (SD=1.37) for the table task, 3.66 (SD=1.36) for
the vertical task, 3.83 (SD=0.98) for the cylinder and 4.33
(SD=1.03) for the on the move task; there was no significant
difference between the four tasks (F3,20= 0.43, p > 0.05).
Mean time to perform the evaluation was approximately half an
hour. Users spent in average 10 minutes per task. Creations
represented buildings, creatures or abstract shapes (Figure 4). The
training period lasted five minutes in average and all users
managed to work with the application. Nevertheless, one user
created collages without sense of space as pieces were placed
randomly. In the same way five users did not use all the features,
specifically they did not scale pieces or adjusted the textures. All
of the participants took more than one picture and expressed to
enjoy the activity. Furthermore, six of them asked if the
application was available for later use.
5. DISCUSSION Features such as mixing virtual content with the real world and on
the move creations seem to provide better support for creativity
than other features of the application as the group 2 obtained
higher scores. Nonetheless, the difference was not significant and
since the evaluation was preliminary, CSI results were not
weighted. SUS scores were higher in the second group. The
reason is that the real cylinder intersects the virtual objects.
Therefore, it provides a better perception of the position and
orientation of the ongoing piece placement. According to the
questionnaires, placing the objects and perceiving their true
position was the most difficult aspect of using the application.
During the on the move task the marker was smaller, consequently
the users did not need to be so far from it. In that task they did not
have to connect anything with the cylinder. Those may be the
reasons of why they reported the task as the easiest one to
perform. The training was slightly more difficult as it required
placement of objects with a certain layout. Owing to the fact that
placing objects is the most problematic aspect, future
improvements will add snap to grid or vertex features. Another
possibility would be to add physic simulation or shadows to the
collage pieces. Professional 3D software may give more precise
results; however, the time required for teaching the users to
perform the same tasks is more elevated.
Originality was higher for the on the move task. Probably the
users had more objects to photograph. The cylinder task was not
the most original. Apparently the cylinder shape conditioned the
placement of the pieces. All the users agreed that the most
interesting aspect was to take pictures and apply them to
587
tridimensional objects using a quick workflow. Several of them
pointed out that integrating virtual objects with the cylinder was
interesting. Furthermore, one said that it would be interesting to
do it with famous buildings or paintings.
Suggested games could be grouped as map editors, interactive
simulations or replacement of images in traditional board games.
Map editors for platform games, Minecraft (Figure 5) or Sims’
homes and interactive simulations such as Rude-Goldberg
machines or Jenga using the pieces. CollARt could be used for
playing Pictionary or Taboo but using tridimensional collages
instead of pictures. Two users attached the marker to some part of
their body and used it for playing (Figure 6). General
improvements recommended were to add multiplayer support or
the ability to share creations on Facebook.
Figure 5. A user tried to replicate a Minecraft landscape.
Figure 6. This user made a skull and performed Hamlet.
6. CONCLUSION CollARt is a mobile augmented reality application which permits
to create tridimensional photo collages. It can be used itinerantly
to create works of art that blend virtual objects with real world
elements. These features increase the creativity support index of
the tool. New techniques for improving the tridimensional
placement of objects should be evaluated. Nonetheless, users
generally used the application properly, enjoyed it and created
interesting works of art.
7. ACKNOWLEDGEMENTS This work has been partially funded by the Government of
Navarra through a research grant.
8. REFERENCES [1] Azuma, R., Baillot, Y., Behringer, R., Feiner, S., Julier, S.
and MacIntyre, B. Recent advances in augmented reality.
Computer Graphics and Applications, vol. 21, no. 6, pp. 34-
47. 2001.
[2] Binder, T., De Michelis, G., Gervautz, M., Jacucci, G.,
Matkovic, K., Psik, T. and Wagner, I. Supporting
configurability in a mixed-media environment for design
students. Personal and Ubiquitous Computing, vol. 8, no. 5,
pp. 310-325. 2004.
[3] Brooke, J. SUS: A 'quick and dirty' usability scale. Usability
Evaluation in Industry 189. Taylor and Francis. 1996.
[4] Carroll, E. A., Latulipe C., Fung, R. and Terry, M. Creativity
Factor Evaluation: Towards a Standardized Survey Metric
for Creativity Support. Creativity & Cognition Berkeley, pp.
127-136. 2009.
[5] Clark, A. and Dunser, A. An interactive augmented reality
coloring book. 3D User Interfaces (3DUI), pp. 7-10. 2012.
[6] Gal, R., Sorkine, O., Popa, T., Sheffer, A. and Cohen-Or, D.
3D collage: expressive non-realistic modeling. Non-
photorealistic animation and rendering, pp. 7-14. 2007.
[7] Girgensohn, A. and Chiu, P. Stained glass photo collages.
IEEE International Conference on Image Processing, vol.2,
pp. 871-874. 2003.
[8] Grasset, R., Boissieux, L., Gascuel, J.D. and Schmalstieg, D.
Interactive mediated reality. Australasian conference on User
interface, vol. 40, pp. 21-29. 2005.
[9] Huang, H., Zhang, L. and Zhang, H. Arcimboldo-like collage
using internet images. ACM Transactions on Graphics, vol.
30, no. 6, p. 155. 2011.
[10] Jacucci, G., Oulasvirta, A., Salovaara, A., Psik, T. and
Wagner, I. Augmented reality painting and collage:
evaluating tangible interaction in a field study. Human-
Computer Interaction-INTERACT, pp. 43-56. 2005.
[11] Kerne, A. Collage Machine: Temporality and Indeterminacy
in Media Browsing via Interface Ecology. SIGCHI Extended
Abstracts, vol. 2. 1997.
[12] Langlotz, T., Mooslechner, S., Zollmann, S., Degendorfer,
C., Reitmayr, G. and Schmalstieg, D. Sketching up the
world: in situ authoring for mobile augmented reality.
Personal and Ubiquitous Computing, pp. 1-8. 2011.
[13] Lee, S., Jung, J., Hong, J., Ryu, J.B. and Yang, H. AR Paint:
A Fusion System of a Paint Tool and AR. Entertainment
Computing, pp. 122-129. 2012.
[14] Marzo, A., Bossavit, B. and Ardaiz, O. Interacting with
multi-touch handheld devices in augmented reality spaces:
effects of marker and screen size. Interacción Persona-
Ordenador, p. 8. 2012.
[15] Marzo, A. and Ardaiz, O. A Smart Phone Tool to Create
Photo Collages, Procedia Computer Science, vol. 15, pp.
159-162. 2012.
[16] Pottery. 2010. http://www.idreams.pl/Pottery/
[17] Qualcomm. Vuforia. https://www.vuforia.com/
[18] Shneiderman, B., Fischer, G., Czerwinski, M., Resnick, M.,
Myers, B. and 13 others. Creativity Support Tools: Report
from a U.S. National Science Foundation Sponsored
Workshop. Int. J. of Human–Computer Interaction vol. 20,
no. 2, pp. 61-77. 2006
588