During our everyday lives, we regularly look for things, performing visual searches. At other times, signs seek to attract our attention. An Augmented Reality (AR)-based assistant system visually highlighting objects of interest would thus prove useful in a variety of situations. But how do we best guide users’ attention? On 2D displays with limited size, simple approaches such as coloring are often sufficient. However, in Augmented and Virtual Reality (VR), targets can be located in any direction, making highlighting more difficult. Many highlighting techniques for use in AR and VR have been proposed to solve this issue. As AR hardware still has some limitations, techniques are frequently evaluated in VR. Yet, these evaluations are often conducted in abstract, small, or static settings. This thesis aims to provide insight into the performance of highlighting techniques in a complex and high-fidelity environment, which users naturally traverse. For this, we analyzed previously proposed approaches. In VR, we implemented four highlighting techniques to compare (Color + Outline, Link, Arrow, and HiveFive). We further provided a high-fidelity virtual city center with various dynamic elements in which to compare the highlighting techniques. Using an Omnidirectional Treadmill, it is possible to traverse this environment while walking naturally. Results of our user study showed that in our high-fidelity environment, Link was the fastest highlighting technique, while Arrow was the slowest. Further, Color + Outline and HiveFive had an increased rate of missed targets. Regarding personal preference, Arrow performed the worst, while the other three were rather popular. Participants also expressed a preference for a combination of in-view and in-situ elements for visual highlighting. This suggests that different highlighting techniques are differently suited for finding objects in complex and high-fidelity environments.