In this paper, we present a method for acquiring, spatially filtering and viewing annotated videos captured with a full field of view multi-head camera moving along a path. We describe our tailored ego-motion recovery algorithm used for calculating the trajectory path of the panoramic head. We then focus on sampling the plenoptic path efficiently according to geometric visibility events. Appropriate samplings allow us to filter and compress the panoramic images avoiding some redundancies in the image database. We present several applications and results of plenoptic paths either obtained from indoor shootings or perfectly rendered by computer graphics scripts.