1. Introduction 1
2.4 Gestural Control 27
2.4.1 Multi-‐touch scrolling 29
A classic form of scrolling in a multi-‐touch application is to place a single finger or two fingers on the item to be scrolled and then moving the finger up or down the screen in a flick motion to swiftly scroll the item; or in a slow motion for finer control over the scroll (Lao. S., et. al., 2009).
A problem occurs if the item to be scrolled is nested within another element that uses the same scrolling gestures for a translation effect. In this instance it may be more suitable to find a separate window to use for scrolling purposes. Or a multi-‐ touch scroll bar.
A multi-‐touch system DTLens (Forlines, C., 2005), uses a resize handle in the lower right of a window that allows the user to zoom into and out of an image by touching and dragging the scroll bar.
This is a useful addition as the zoom slider affects the items nested within an element (Figure 2.17).
Figure 2.17 – DTLens showing the scroll bar in the bottom right of the window. The user can touch and drag the scroll bar to adjust the zoom parameter in the
main window. (Forlines et. al., 2005)
2.5 3D Interaction
A potential issue arises when the user wishes to manipulate 3D objects using a multi-‐touch system and still keep the smooth natural interaction process. A 2D object can be manipulated by rotation, zooming and panning – tasks that can be easily performed using two hands. The notion of ‘Degrees of Freedom’ (DOF) can be introduced that defines which planes of movement an object can be moved in. A 2D object can be moved in 3 DOF, as described above, but a 3D object requires 6 DOF:
1. Moving up and down 2. Moving left and right
3. Moving forward and backward 4. Tilting forward and backward 5. Turning left and right
6. Tilting side to side (Martinet et al., 2009)
Hancock et al. have defined a form of 3D interaction by using a shallow-‐depth z plane in a 3D environment where the 3D object can be rotated about the x and y axes by touching it with a finger and moving the finger over the interactive surface. The standard interactions performed on a GUI are deemed shallow depth as windows can be stacked on top of each other and riffled through which makes this method advantageous for users of these systems.
A 5 DOF movement can be achieved by a single touch by pinning the touch action to the object through point of contact. Touching the point works like a sticky finger in that the contact point will rise to the surface. Rotating the object requires the user to touch a slide and drag it. A retouch may be required to keep rotating the object to view occluded sides (Figure 2.18) (Hancock et al., 2007).
Figure 2.18 – A sequence of motion using one-‐touch interaction in shallow-‐depth 3D. The black dot represents the point of contact of the user’s finger. (Hancock et
al., 2007)
Although this method allows the user to rotate an object in any direction it is often necessary to place constraints on the rotation such as movement in one plane only. This can be done by drawing a doughnut around the shape that allows rotation around that axis only by touching and dragging around it.
Five or six DOF can be achieved by using only 2 points of contact. The first point will allow for free rotation and translation in the x and y-‐axis as described above but a second point will add the ability to pitch and roll. Two contact points also allows for movement in the z-‐axis by changing the distance between them in a similar way to zooming 2D pages.
By introducing a third touch point pitch and roll adjustment are included in addition to the two previous movements. Although this allows for greater control over the object the combination of each touch point is immediately quite confusing for a user without practice of each individual point.
An empirical study was carried out to determine how each touch input affected speed, accuracy and user preference. The results showed that three-‐finger touch is faster (average 13.3s completion time) than two-‐finger touch (average 15.7s completion time), which is faster than one-‐finger (average 18.9s completion time) (Hancock et al., 2007). The same trend was seen with incomplete trials with the three-‐finger input obtaining the lowest number. On the user preference
questionnaire the three-‐finger input obtained the highest score, on average, with a higher preference seen on ease-‐of-‐use and expectation compare to the other inputs.
Another approach for 3D interaction examines how a traditional desk is used. 2D objects can be manipulated, such as paper, as well as 3D objects, such as pens and books. Knowing this an analogy can be assumed for multi-‐touch tables that allows the use of tangible objects for interaction (Hancock et al., 2009).
A device has been created, TableBall, incorporating a trackball that allows for 5 DOF – 3 DOF is tracked by the position and rotation of the device on the table and 2 DOF are provided by the trackball. Placing the TableBall on an object selects it and sliding and rotating the device across the table also moves the object in relation. The trackball on top of the device allows for precise object rotation around the x and y axis.
A user study was carried out to further explore the useful extend of tangible interactions. The users were required to dock a 3D pyramid with another pyramid
so that the vertices matched up. The pyramid was considered docked if the vertices were aligned to within 6cm (126 pixels).
The results showed that the users preferred using direct touch instead of TableBall. Direct touch was quicker than using TableBall for planar movements (where the object was only moved over a 2D area). Completion times for planar movements were 6.5 seconds for direct touch and 15.2 seconds when using TableBall. However, when full 3D rotation and translation was required the TableBall
technique was slightly faster with a completion time of 17.1 seconds as opposed to 17.6 seconds for direct touch.