ReferenceAuthors :Mathieu Nancel, Julie Wagner, Emmanuel Pietriga, Olivier Chapuis, Wendy Mackay
Affiliations: LRI -Univ Paris-Sud & CNRS,INRIA ,F-91405 Orsay, France
Presentation: CHI 2011, May 7–12, 2011, Vancouver, BC, Canada
Summary
Contents
As complex tasks such as pan-zoom navigation have received little attention on large sized high resolution display,building upon empirical data gathered from studies of pan-and-zoom on desktop computers and studies of remote pointing, the paper identifies three key factors for the design of mid-air pan-and-zoom techniques;uni- vs. bi- manual interaction, linear vs. circular movements, and level of guidance to accomplish the gestures in mid-air. 32 LCDs each 30 inch diagonal were used to create the display and the software was ZVTM toolkit run on Mac OS X.
Hypothesis
The paper presents the following seven hypotheses
Two- handed gestures will be faster than one-handed gestures because panning and zooming are complementary actions, integrated into a single task . Two-handed gestures should also be more accurate and easier to use.Linear gestures should map better to the zooming component of the task, but should eventually be slower because of clutching.Users would prefer clutch-free circular gestures.Techniques using fingers (1D Path and 2D Surface conditions), should be faster than those requiring larger muscle groups. 1DPath gestures would be faster, with fewer overshoots than techniques with lesser haptic feedback. Finally, they predicted that 3DFree gestures would be more tiring.
Methods
User study was done among 12 participants for the following task.
Participants navigate through an abstract information space made of two groups of concentric circles: the start group and the target group until they get to the target, after they are given a brief introduction of the procedure. Each technique is presented in a block of 18 trials consisting of 6 replications at each distance.They measure movement time MT and number of overshoots for each of 2592 trials: 2 GESTURE × 2 HANDedness× 3 GUIDANCE × 3 DISTANCE × 12 participants × 6 replications. Participants also answer questions, based on a 5-point Likert scale, about their perceived performance, accuracy, ease of learning, ease of use, and fatigue.Each session lasted between 30 and 90 minutes, depending on techniques and participant.
Results
On all other equal basis, two-handed techniques are consistently faster than one-handed techniques. It is more advantageous when the degree of guidance for achieving gestures is low. Movement time and number of overshoots were perfectly correlated with the overall performance measures.Involving smaller muscle groups improved performance.Linear gestures are generally faster and preferred than Circular ones.
1D Path (avg. 9511 ms) was significantly faster than 2DSurface which was faster than 3DFree .For Circular gestures, 1DPath guidance is faster than both 2DSurface and 3DFree.For Linear gestures, there is no significant difference between 1DPath and 2DSurface, but a significant difference between 2DSurface and 3DFree. For 1DPath guidance there is no significant difference between Circular and Linear gestures, but there is a significant difference between Circular and Linear for 2DSurface and 3DFree guidance.
There is a significant difference between Linear and Circular gestures for 2DSurface and 3DFree, but not 1DPath.2DSurface gestures exhibit more overshoots than 1DPath and 3DFree gestures.
4 participants preferred the high degree of guidance provided by 1DPath techniques, compared to only 1 for both of 2D Surface and 3D Free techniques.1DPath guidance was least tiring while 3DFree caused the most fatigue. Participants generally preferred TwoHanded to OneHanded techniques and Linear to Circular gestures which was less tiring as well.Four different groups were created based on the techniques preferred by individuals.
Discussions
I was overall satisfied with the results. Some of the things regarding preference of path, fatigue looked obvious and were further bolstered by the results. Some results were interesting, like the test results for hypothesis 3. The paper has provided a basis to talk about while designing new applications with wall sized interaction spaces and as well with all different forms of gestures combined. As movies have been showing, hopefully this research can lead us to technology where we can perform gestures to control the large displays using gestures rather than other devices.
Comments
Post a Comment