Team, Visitors, External Collaborators
Overall Objectives
Research Program
Application Domains
Highlights of the Year
New Software and Platforms
New Results
Bilateral Contracts and Grants with Industry
Partnerships and Cooperations
XML PDF e-pub
PDF e-Pub

Section: New Results


Participants : Axel Antoine, Marc Baloup, Géry Casiez, Stéphane Huot, Edward Lank, Sylvain Malacria, Mathieu Nancel, Thomas Pietrzak [contact person] , Thibault Raffaillac, Marcelo Wanderley.

Production of illustrative supports

Trace figures are contour drawings of people and objects that capture the essence of scenes without the visual noise of photos or other visual representations. Their focus and clarity make them ideal representations to illustrate designs or interaction techniques. In practice, creating those figures is a tedious task requiring advanced skills, even when creating the figures by tracing outlines based on photos. To mediate the process of creating trace figures, we introduce the open-source tool Esquisse (Figure 3). Informed by our taxonomy of 124 trace figures, Esquisse provides an innovative 3D model staging workflow, with specific interaction techniques that facilitate 3D staging through kinematic manipulation, anchor points and posture tracking. Our rendering algorithm (including stroboscopic rendering effects) creates vector-based trace figures of 3D scenes. We validated Esquisse with an experiment where participants created trace figures illustrating interaction techniques, and results show that participants quickly managed to use and appropriate the tool [18].

Figure 3. Workflow of Esquisse: (left) facilitating staging of 3D scenes with templates, anchor points, direct UI import, kinematic manipulation and automated posture tracking, (center) Esquisse's algorithm rendering the tracing images, and (right) final vector-based trace figures.

Impact of confirmation modes on expert interaction techniques adoption

Expert interaction techniques such as gestures or keyboard shortcuts are more efficient than traditional WIMP techniques because it is often faster to recall a command than to navigate to it. However, many users seem to be reluctant to switch to expert interaction. We hypothesized the cause might be the aversion to making errors. To test this, we designed two intermediate modes for the FastTap interaction technique, allowing quick confirmation of what the user has retrieved from memory, and quick adjustment if she made an error. We investigated the impact of these modes and of various error costs in a controlled study, and found that participants adopted the intermediate modes, that these modes reduced error rate when the cost of errors was high, and that they did not substantially change selection times. However, while it validates the design of our intermediate modes, we found no evidence of greater switch to memory-based interaction, suggesting that reducing error rate is not sufficient to motivate the adoption of expert use of techniques [25].

Effect of the context on mobile interaction

Pointing techniques for eyewear using a simulated pedestrian environment

Eyewear displays allow users to interact with virtual content displayed over real-world vision, in active situations like standing and walking. Pointing techniques for eyewear displays have been proposed, but their social acceptability, efficiency, and situation awareness remain to be assessed. Using a novel street-walking simulator, we conducted an empirical study of target acquisition while standing and walking under different levels of street crowdedness. Results showed that indirect touch was the most efficient and socially acceptable technique, and that in-air pointing was inefficient when walking. Interestingly, the eyewear displays did not improve situation awareness compared to the control condition [23].

Studying smartphone motion gestures in private or public contexts

We also investigated the effect of social exposure on smartphone motion gestures. We conducted a study where participants performed sets of motion gestures on a smartphone in both private and public locations. Using data from the smartphone’s accelerometer, we found that the location had a significant effect on both the duration and intensity of the participants’ gestures. We concluded that it may not be sufficient for gesture input systems to be designed and calibrated purely in private lab settings. Instead, motion gesture input systems for smartphones may need to be aware of the changing context of the device and to account for this in algorithms that interpret gestural input [26].