The magic lens paradigm, a commonly used descriptor for handheld Augmented Reality (AR), presents the user with dual views: the augmented view (magic lens) that appears on the device, and the real view of the surroundings (what the user can see around the perimeter of the device). The augmented view is typically implemented by rendering the video captured by the rear-facing camera directly onto the device’s screen. This results in dual perspectives—the real world being captured from the device’s perspective rather than the user’s perspective (what an observer would see looking through a transparent glass pane). These differences manifest themselves in misaligned and/or incorrectly scaled transparency resulting in the dual-view problem. This paper presents two user studies comparing (a) device-perspective and (b) fixed Point-of-View (POV) user-perspective magic lenses to analyze the effect of the dual-view problem on the use of the surrounding visual context. The results confirm that the dual-view problem, a result of dual perspective, has a significant effect on the use of information from the surrounding visual context. The study also highlights that magnification and not the dual-view problem is the key factor explaining the correlation between magic lens size and the increased intensity of the magic lens type effect. From the results, we derive design guidelines for future magic lenses.