The strongest point of Visual Touchpad is the way it performs segmentation of the hand regions from the video frames and augment them transparently on the screen. This not only provides a really compelling direct manipulation experience, but also help with self-occlusion issues.
Although projected hands on the screen provide a direct manipulation experience with significantly less self-occlusion, typing tasks on Visual Touchpad does not give haptic feedback. This not only makes text entry difficult and awkward, but also prone to errors.
Since the cameras are assumed to be above the touchpad surface facing downward, I would like to work on adding an extra input dimension using simple hand gestures. This additional input dimension could bring the concept of hovering over the keyboard, an expression which originally belongs to mouse interactions.