Abstract
The number and size of 3D object repositories have been increasing at a considerable rate. Consequently, finding a specific 3D model in such collections is getting more difficult. Current 3D search engines do not take advantage of novel interaction technologies, usually presenting query results in grids of thumbnails. This greatly hinders objects' interpretation and allows little to none manipulation at all. Immersive environments are believed as promising solutions for displaying 3D models, allowing complete representations of the models. These environments can be enhanced with multimodal techniques for a more natural interaction. In this paper we present a prototype that uses immersive visualization and mid-air interactions to explore query results in a dataset with 3D objects, using one of four different visualization modes. We evaluated these modes with 29 users and concluded that our immersive approaches are preferred by users and, albeit novel, perform at par relatively to traditional bi-dimensional grids with thumbnails.

Abstract
Manipulating objects is an essential aspect in virtual environments. Nonetheless, object positioning in immersive virtual environments relying in direct and natural approaches is still difficult. Previous research concluded that degrees-of-freedom separation in mouse and touch interfaces led to positive results. In this document we present a user evaluation to assess if explicit separation of degrees-of-freedom also benefits mid-air manipulation tasks. We implemented a virtual widget based technique that allows users to control a single DOF, and compared it against a direct approach and the PRISM technique, which adjusts the ratio between the hand and object movement. The results of our assessment suggest that full DOF separation benefits precision in spatial manipulations, at the expense of additional time for complex tasks. From these results we proposed a new technique that combines different aspects from the three techniques compared in our assessment.