Users seem to be more pro-e cient with virtual handles since it provides more comprehensible feedback of the rotation axis, while reducing time to change rotation axis - because users do not need to change views, as observed in (1), or define the axis manually, as required by (2). User evaluation showed that (3) outperforms the remainder approaches. These techniques only allow to control rotation over one DOF at a time, something we found desirable in our evaluation of LEGO applications. Concerning the rotation of objects, we developed three techniques: (1) uses the RST applied to the current translation plane, i.e., while moving the object, a second touch rotates the object around the first (2) is similar to the Opposable Thumb, but the rotation axis is one of the scene axes (3) uses virtual handles (similar to object handles in ), the first touch selects the rotation axis and the second touch rotates the object. When the object intersects the construction grid the arrows become red. The arrows depict the possible directions in which the object can move, which describes the current translation plane. Visual feedback is provided by four arrows and the shadow of the object. If the user intends to change the translation plane, no camera movements are required instead, a tap with a second finger alternates between a horizontal and a vertical plane (Figure 4.c). After picking up the object (Figure 4.a), the user translates it in the plane defined by the camera (Figure 4.b), allowing to take the most of the current view. Our solution, combines the aforementioned approaches into one, as depicted in Figure 4. Result analysis denoted (1) and (3) as the more e cient approaches, although without statistically significant di ↵ erences between them, which suggests that a combination of the two might result in the desired approach (since they are combinable). Also, we found that manipulating object depth with the scrolling touch was sometimes misunderstood as a scaling gesture. Test results showed that users experienced di culties when translating objects in a plane parallel to the view, suggesting that an orthogonal (to one of the scene axes) translation plane is desirable. We compared these approaches amongst each other and with the Z-Technique.
For all techniques, dragging the object will move it in the current translation plane thus, the user manipulates no more than two DOFs with one finger, which maintains a strict relation between the two-dimensional input and the two-dimensional translation. We developed three translation techniques: (1) uses a translation plane in which the normal direction is closer to the view vector and orthogonal to one of the scene axes (2) moves the object accordingly to a horizontal plane and, by scrolling a second touch, changes the object depth relatively to the camera (similarly to the Z-Technique) and (3) uses a horizontal translation plane which, by tapping a second finger, changes to a vertical plane. Complementing the object manipulation, we present our camera manipulation and cloning metaphor. In the following sections we briefly describe the comparison of these techniques and draw conclusions upon the selected approach. In order to assess the best approach for 3D object manipulation on multi-touch tabletops, with unconstrained view- points, we carried out a user evaluation with 20 users comparing several approaches for translation and rotation. As our test subjects denoted in their commentaries, this gesture was found to be very familiar and understandable. This gesture is resemblant of picking up a real LEGO brick and placing it on the desired location. Instead, we used a “pick” metaphor to grab an object and then, without releasing, move it, as depicted in Figure 4. To manipulate bricks, we aimed at a natural and familiar scenario for LEGO users, moving further from the traditional selection concept - which is to drag objects with a single touch. A final analysis of our interface shows that tabletop design guidelines have been taken into consideration, accounting for: maximizing the interactive area (77% of the area is dedicated for building the models) dominant hand is taken into consideration (as it provides more ergonomic comfort throughout the interaction) and graphical widgets have an appropriate size (enough to be touched and remain partially visible, thus minimizing occlusion). The right toolbar is a color palette, that can be used to change the color of a brick or the whole brick list. Furthermore, it also allows to temporarily hide bricks or to show previously hid- den bricks.
save the current model and quit the application.