You are on page 1of 6

Natural Gestures for Multi-Touch Applications

Elizabeth Starnes Paul Varcholik


University of Central Florida University of Central Florida
Orlando, Florida Orlando, Florida
lizzystarnes@gmail.com pvarchol@bespokesoftware.org
Charles Hughes
University of Central Florida
Orlando, Florida
ceh@eecs.ucf.edu

may change their choice of gestures. For those


Abstract who do not own/use any multi-touch device,
what would be their preferred gesture for zoom-
Multi-touch interfaces are becoming com- ing or resizing an image? What if the user is not
monplace on smart phones and mouse pads, high- from a technical background? This paper investi-
end monitors and a number of notebook comput- gates which gesture would be the most natural
ers. Unfortunately, most of these commodity sys- for zooming and rotating an image depending on
tems have small form factors and support only the hemispheric dominance of the user and
two simultaneous touches. Examples of larger, whether the user owns a multi-touch device.
true multi-touch devices more commonly occur Most research for gestures in multi-touch is very
in research and demonstration settings. The focus specific. The scope of this research is broad
of this project is to investigate a range of multi- enough to help programmers decide which ges-
touch gestures (static and dynamic). Observations tures to include for future applications, especially
were made through a group of students as to what those running on non-Windows based GUI inter-
gestures users find the most “natural” for zoom- faces.
ing, rotating, etc. when interacting with picture
based multi-touch applications. Such data can
help programmers to consider gestures for future 2 Previous Work
programs based on the idea that users may come
In the 1970’s, Myron Krueger pioneered research
from artistic backgrounds versus technical back-
in the areas for augment reality through Videop-
grounds. Programmers should also consider the
lace. It was a room where people could interact
size of the table used with their applications and
with the walls by movement and see their sil-
whether the majority of the users own handheld
houettes and other objects projected on the wall.
multi-touch devices.
The idea of “pinching” or stretching an image by
a hand gesture was mainly based off his work
(Krueger, Gionfriddo and Hinrichsen 1985).
In the realm of multi-touch, most HCI re-
1 Introduction
search pertaining to gestures is specific to certain
The way people interact with multi-touch dis- topics such as robotics (Micire, Desai et al 2009),
plays is through gestures. A gesture is a physical ShadowGuides (Freeman, Benko et al 2009),
motion that maps to an action in an application. hand shapes (Epps, Linchman and Wu 2006),
For example, if a person were to touch an icon to etc. For example, researchers at the University of
open a file, the action would be the file opening Massachusetts focused on natural gestures relat-
on the screen. The pinching gesture for zooming ing to robotic teams. Users were asked to com-
on an iPhone or other handheld multi-touch de- plete a series of tasks with no information on
vices has become the norm for many people. For which gestures to utilize. The users who
most people who own these devices, when pre- owned/used an iPhone tend to use the “pinching”
sented a multi-touch table, they may prefer this gesture that is so commonly known for zooming.
gesture over another. But, the size of the table Users who played video games also used certain
gestures pertaining to mouse interaction. The
researchers mainly concluded that the well-
known mouse driven interfaces need to carry
over into the multi-touch domain (Micire, Desai
et al 2009).
Koskien et al. (2008) acquired applicable in-
formation on gestures by a user study in Finland.
Their focus was on power control rooms and the
eventual change to multi-touch systems instead
of analog or digital. Since it was such a unique
Figure 1 shows the prototype table for TACTUS. Pic-
study where users did not have tasks associated
ture from Bespoke Software.
with desktops, users did not typically chose ges-
tures specific to the mouse. Users tend to use the The FTIR design was based on and named
multi-touch gestures when there was a task that for the idea of frustrated total internal reflection
required a larger area to initiate. in the field of optics. A diagram of the FTIR
Other examples of research pertaining to design for multi-touch tables is shown in figure
gestures include entail shadows, shapes and user 2. When light travels from one medium (such as
defined gestures. Hilliges et al. (2009) investi- glass and water) there is no angle of refraction
gated depth to gestures by shadows. Their exam- and the light is totally reflected (Han 2005).
ple shows a person picking up a ball by merely
hovering over the table, not by touching the ta-
ble. Epps et al. (2009) explored hand shapes for
interacting with a table top. The majority of their
subjects used index finger-based shapes and the
set up happened to be Windows-oriented for
icons, etc..
Most of the research for gestures involved
specifically asking people what gestures they
prefer; an entire gesture set is created. This paper
mainly focuses on a few common actions as op-
Figure 2 displays the FTIR design proposed by Jeff
posed to an entire gesture set. These actions are
Han, image from (NUI Group Authors 2006).
also general and not geared to a specific genre of
multi-touch applications. In the user study, stu-
In the case of multi-touch tables, infrared
dents were asked to show on the table which ges-
light is projected inside the acrylic table top and
ture to use pertaining to an action such as zoom-
is reflected. When a person touches the top of the
ing or moving an object.
acrylic, some of the infrared light is transfer to
the fingertip and hence the total internal reflec-
3 Hardware tion is frustrated (NUI Group Authors 2006). A
good example of frustrated total internal reflec-
The user study was conducted on the prototype tion in everyday life is fingerprints on a glass of
for TACTUS, a multi-touch research testbed. water. An infrared camera can detect this scatter-
The picture of the table is seen below in figure 1 ing of light, which shows up as a blob. A blob is
(Varcholik, LaViolia and Nicholison 2009). representation of the contact point of the table
TACTUS uses the FTIR (Frustrated Total Inter- and a fingertip to the camera. Relatively simple
nal Reflection) design proposed by Jeff Han in vision-based algorithms can find the coordinates
2005 as an inexpensive way to create a multi- of these blobs based on camera images. User
touch table based on a simple principle of phys- applications, such as the picture manipulation
ics. one for this study, can then be built using the
data available from these algorithms.

4 User Study
Sixteen volunteers from the University of Central
Florida participated in a user study – six females
and ten males, between the ages of 20 and 29. hand held multi-touch device before stuck to us-
One volunteer was a graduate student and anoth- ing two fingers for most of the tasks of zooming
er was graduated. Majors ranged from Digital and rotating. This was most likely because they
Media, Art, Engineering, Computer Science, were accustomed to their handheld multi-touch
Music Education and Event Management. Three devices, though only a fourth of those students
of the participants were left-handed. used the pinch gesture with one hand. One stu-
The students were placed in front of the dent commented that, since the table was so
prototype for TACTUS and shown a display that large, he would not use the pinch gesture with
contained seven pictures. The display contained one hand. If the table had been a bit smaller, he
no other icons and the GUI of Windows was not said his first reaction would be to use the one-
fully shown, except for the toolbar at the bottom. handed pinch gesture to zoom.
A screenshot of what the students saw is below
in Figure 3. The users were asked a series of 5.1 Zooming an image
questions before interacting with the table. They Students came up with typical and different types
were asked if they have used or owned an iTouch of gestures to zoom on an image. One unique
or a handheld multi-touch device before and gesture was the four point zoom in which stu-
whether they have interacted with a large multi- dents used two fingers on each hand, specifical-
touch display such as a wall, table or computer ly. These students commented that the table was
screen. 87.5% of users have used a handheld quite large and the use of multiple fingers on
multi-touch device before and 37.5% owned each hand felt necessary to accomplish the task.
such devices. Students were also asked if they A few students also displayed a particular
have interacted with a multi-touch device larger gesture of double tapping to zoom. This idea
than handheld, such as a monitor, table, wall, etc. probably came from double clicking on images
44% of the students described interactions with for a usual computer interface with a mouse. One
larger types of multi-touch devices. student’s gesture for zooming was to press down
The students were then asked to show on for a period of time (selection) and then to zoom
the multi-touch table the gestures they would use out they tapped the screen and moved their finger
to perform actions on the pictures shown. They downwards in a “scrolling” manner. The student
were asked to zoom in and out, rotate and move had never used a multi-touch device before and
the pictures. They were also asked to clear all the preferred a button rather than doing a certain
pictures from their sight, which usually resulted gesture. Another student suggested a double tap
in a “flicking” motion. to zoom in and a single tap to zoom out on a
picture. Both of these students were from
artistic/creative disciplines. A more visual expla-
nation of this gesture and the other gestures are
shown in figure 4 below.

Figure 3 displays the pictures the students were asked


to describe how to manipulate through various ges-
tures.

5 Results
A majority of students used two hands to per-
form the tasks of zooming and rotating images.
For moving or clearing the pictures, one hand Figure 4 shows the gestures the students used to
was preferred The students who owned or used a
zoom in and out on a picture. Illustrations reproduced One student emulated a mouse by tapping to se-
from Gestureworks. lect an image and dragging it with one finger.
The gesture for clearing the images was al-
5.2 Rotating an image
so asked and needed to be clarified as to whether
There was a variety of gestures to rotate a pic- it pertained to one image as opposed to all of the
ture. Figure 5 shows the visual representation of images shown. Also, the subjects were confused
the gestures used for rotating an image. There as to where the pictures would end up when they
were a few forms of two point gestures, such as were cleared and they wanted a place to throw
the pivot and rotate. One finger would stay sta- them into instead of using just the screen as the
tionary on the picture while the other hand/finger reference. Students also came up with a five
rotated around the pivot finger. There was no point flick, which was described either moving
feedback given for students to persuade them to left or right.
use this gesture, but it happened to be the gesture
recognized by the program, though that recogni-
tion was turned off while the students performed
their gestures. The other form of the two point
gestures (two hands) was moving both points
instead of keeping one as a pivot point. These
points would move in opposite directions and in
an arch motion.
The art students again came up with inter-
esting gestures such as using a lasso (one point in
a circular fashion) to rotate a picture. Such a ges-
ture has mainly been seen to group images or
select multiple images. This data was seen in the Figure 6 shows the students’ gestures for clearing and
study for controlling robot teams (Micire, Desai moving pictures. Illustrations reproduced from Ges-
et al 2009). Another student suggested tapping tureworks.
on an image, then moving their finger in a way
like the lasso. The second gesture choice was 6 Picture Application
five points on one hand and merely rotating the After the students were asked the questions for
hand left or right depicting which way the picture the user study, they were allowed to interact with
would rotate. the picture application that was created for the
project. The picture application was initially
supposed to be the user study in which students
would only interact with the program and their
actions observed. There were many issues with
the application regarding hardware and software.
The students were able to zoom with the pinch
gesture, two point/two hand gesture and a ten
point hand gesture (same idea as two point/two
hand but with five points for each hand). They
were also able to rotate pictures using the pivot
and rotate hand gesture. When the program was
having issues, subjects tended to stick with one
Figure 5 shows the students gestures for rotating an
image. Illustrations reproduced from Gestureworks.
handed gestures to compensate for the difficulty
with the table. We obtained a lot of feedback that
5.3 Moving and Clearing images will be useful in future experiments.
The students were asked to show the gesture for 7 Discussion
how to move a picture and typically they would
use one or a few points. Figure 6 below displays The data found from the students showed that
the gestures described for moving and clearing programmers should consider certain demo-
pictures. Surprisingly, some students used two graphics and whether the majority of users own a
points specifically to move an image, though multi-touch device. Users may come from differ-
they said up to five points should be recognized. ent disciplines such as an artistic background as
opposed to a technical one. Users may also have ti-touch devices, a user literally rotates the device
never interacted with a handheld multi-touch de- to rotate an image. This may be where the rota-
vice or have never owned one before. The stu- tion gestures come from, rather than being based
dents from artistic backgrounds came up with on some natural or normative gesture. 44% of
very unique gestures to perform typical tasks. students showed a two point/two hand gesture to
Also, the students that have not owned a hand- rotate a picture that involved both points moving.
held multi-touch may use variations of typical They specifically requested to rotate pictures at
gestures for the same tasks. The zoom gesture for the corners rather than the middle with this ges-
an iTouch, iPad, etc. did not prove to be the most ture. Figure 8 shows a graphical representation of
natural gesture on a large surface like a multi- the number of students who chose certain ges-
touch table. Students commented that if the sur- tures to perform the action. The second most
face was smaller they would use the pinch ges- popular gesture was the five point gesture, fol-
ture, but instead an adaptation of using two lowed by the pivot gesture and lasso.
hands came out of the study because of the size
of the table.
7.1 Zooming
The students who did not own a multi-touch
device still used typical zooming gestures. The
graph in figure 7 shows the gestures used by
students who do not own a handheld multi-touch
device. Only 12.5% of students who do not own
multi-touch devices used the pinch gesture. Most
of these students expressed an adaptation of the
pinch by using the two point/two handed gesture. Figure 8 shows the number of students who expressed
They commented that, if the surface was smaller, certain rotation gestures.
they would use the pinch gesture. The students
from academic disciplines such as Digital Media 7.3 Moving and clearing pictures
or Art came up with the unique gestures of press- The most used gesture for moving was simply
ing down, single tap, etc. People from localities using one point. This concept was most likely
that are more art-centered may have a natural intuitive because one point relates to a mouse
inclination to some creative gestures for tasks, cursor simply moving across the screen. There
especially when manipulating pictures. Pro- were interesting results for clearing pictures off
grammers may want to consider these demo- the screen. The usual gesture for doing such a
graphics when considering where the multi-touch task is using a one point flick because of the
display will be located. handheld multi-touch devices, but the majority of
students used five points. In particular, half of
the students who own a handheld multi-touch
device used five points to flick an image.

8 Conclusion
Students from the University of Central Florida
were asked what gestures they would use for cer-
tain tasks when faced with a multi-touch picture
manipulation application. The data found by this
study can potentially help programmers choose
appropriate gestures based on demographics such
Figure 7 shows the number of students who do not as if the users don’t own multi-touch devices and
own multi-touch devices and what gestures they ex- if the users are artistically inclined. Users with
pressed to rotate an image. technical backgrounds versus artistic background
7.2 Rotating will most likely interact with an application dif-
ferently.
Since most multi-touch devices use other means
to rotate images such as, for most handheld mul-
Acknowledgments 7. Hilliges, O., Izadi, S., Wilson, A. D., Hodges,
S., Garcia-Mendoza, A., and Butz, A. 2009. Inte-
Partial support for this work was provided by the ractions in the air: adding further depth to inter-
National Science Foundation Scholarships in active tabletops. In Proceedings of the 22nd An-
Science, Technology, Engineering and Mathe- nual ACM Symposium on User interface Soft-
matics (S-STEM) program under award No. ware and Technology (Victoria, BC, Canada,
0806931 and grant No. 0638977. Any opinions, October 04 - 07, 2009). UIST '09. ACM, New
findings and conclusions or recommendation York, NY, 139-148.
expressed in this material are those of the authors
an do not necessarily reflect the views of the Na- 8. Koskinen, H. M., Laarni, J. O., and Honka-
tional Science Foundation. Partial support for maa, P. M. 2008. Hands-on the process control:
this work was also provided by Workforce Cen- users preferences and associations on hand
tral Florida and the UCF Office of Research and movements. In CHI '08 Extended Abstracts on
Commercialization. Human Factors in Computing Systems (Florence,
Italy, April 05 - 10, 2008). CHI '08. ACM, New
York, NY, 3063-3068.
References
9. Krueger, M. W., Gionfriddo, T., and Hinrich-
1. BespokeSoftware, sen, K. 1985. VIDEOPLACE—an artificial reali-
http://www.bespokesoftware.org ty. In Proceedings of the SIGCHI Conference on
Human Factors in Computing Systems (San
2. Dang, C. T., Straub, M., and Andre, E. Francisco, California, United States). CHI '85.
(2009). Hand Distinction for Multi-Touch ACM, New York, NY, 35-40.
Tabletop Interaction. In Proceedings of the ACM
international Conference on interactive Table- 10. Micire, M., Desai, M., Courtemanche, A.,
tops and Surfaces (Banff, Alberta, Canada, No- Tsui, K. M., and Yanco, H. A. 2009. Analysis of
vember 23 - 25, 2009). ITS '09. ACM, New natural gestures for controlling robot teams on
York, NY, 101-108. multi-touch tabletop surfaces. In Proceedings of
the ACM international Conference on interactive
3. Epps, J., Lichman, S., and Wu, M. 2006. A Tabletops and Surfaces (Banff, Alberta, Canada,
study of hand shape use in tabletop gesture inte- November 23 - 25, 2009). ITS '09. ACM, New
raction. In CHI '06 Extended Abstracts on Hu- York, NY, 41-48.
man Factors in Computing Systems (Montréal,
Québec, Canada, April 22 - 27, 2006). CHI '06. 11. NUI Group Authors. (2006). Frustrated
ACM, New York, NY, 748-753. Total Internal Reflection(FTIR). Retrieved May
26, 2010, from NUI Group Wiki:
4. Freeman, D., Benko, H., Morris, M. R., and http://wiki.nuigroup.com/
Wigdor, D. 2009. ShadowGuides: visualizations
for in-situ learning of multi-touch and whole- 12. Varcholik, P., LaViola, J., Nicholson, D.
hand gestures. In Proceedings of the ACM inter- (2009). TACTUS: A Hardware and Software
national Conference on interactive Tabletops Testbed for Research in Multi-Touch Interaction.
and Surfaces (Banff, Alberta, Canada, November Proceedings of the 13th International Confe-
23 - 25, 2009). ITS '09. ACM, New York, NY, rence on Human-Computer Interaction (HCI
165-172. International 2009).

5. Gestureworks, http://www.gestureworks.com

6. Han, J. Y. 2005. Low-cost multi-touch sens-


ing through frustrated total internal reflection. In
Proceedings of the 18th Annual ACM Symposium
on User interface Software and Technology
(Seattle, WA, USA, October 23 - 26, 2005).
UIST '05. ACM, New York, NY, 115-118.

You might also like