The sinking costs of producing digital video and its growing presence on the Internet suggest that it has potential for use in web-based learning technologies. However, there have been few investigations into how the kind of interaction one has with video impacts subsequent learning. In this in-progress study participants are asked to watch video of an expert taking apart a toaster and describing how it works. The recorded event is the same for all participants, but the event is presented in one of three different modalities: (1) digital video shot from a free-standing camera (2) digital video shot from a free-standing camera that has been annotated in a video markup application called DIVER and (3) digital video shot from a head-mounted camera. A number of different assessment tasks are used to characterize the quantity and type of learning that is supported by a particular mode of video interaction.