Technologies for 3D input and output are rapidly advancing, and are becoming more common in home and workplace environments. However, viewing a stereoscopic display can cause eye strain and fully relying on 3D spatial interactions can be fatiguing and may be less efficient for common 2D tasks. In this paper we explore the design possibilities of transitioning between 2D and 3D modalities to best support the user’s current task. In a formal study, we demonstrate that a Hybrid interaction technique, that transitions between 2D display and input, to 3D, mid-task, outperforms 2D only and 3D only techniques. Guided by this result, we present HybridSpace, a simple modeling environment that combines the benefits of 2D and 3D input and output modalities within a single coherent interface.