Human-computer interaction based on hand gesture tracking is not uncommon in Augmented Reality. In fact, the most recent optical Augmented Reality devices include this type of natural interaction. However, due to hardware and system limitations, these devices, more often than not, settle for semi-natural interaction techniques, which may not always be appropriate for some of the tasks needed in Augmented Reality applications. For this reason, we compare two different optical Augmented Reality setups equipped with hand tracking. The first one is based on a Microsoft HoloLens (released in 2016) and the other one is based on a Magic Leap One (released more than two years later). Both devices offer similar solutions for the visualization and registration problems but differ in the hand tracking approach, since the former uses a metaphoric hand-gesture tracking and the latter relies on an isomorphic approach. We raise seven research questions regarding these two setups, which we answer after performing two task-based experiments using virtual elements, of different sizes, that are moved using natural hand interaction. The questions deal with the accuracy and performance achieved with these setups and also with user preference, recommendation and perceived usefulness. For this purpose, we collect both subjective and objective data about the completion of these tasks. Our initial hypothesis was that there would be differences, in favor of the isomorphic and newer setup, in the use of hand interaction. However, the results surprisingly show that there are very small objective differences between these setups, and the isomorphic approach is not significantly better in terms of accuracy and mistakes, although it allows a faster completion of one of the tasks. In addition, no remarkable statistically significant differences can be found between the two setups in the subjective datasets gathered through a specific questionnaire. We also analyze the opinions of the participants in terms of usefulness, preference and recommendation. The results show that, although the Magic Leap-based system gets more support, the differences are not statistically significant.