An approach is provided in which the approach displays a main video feed and multiple secondary video feeds on a display. The multiple secondary video feeds are embedded in the main video feed at multiple locations based on visual characteristics sets of the multiple secondary video feeds. The approach dynamically detects a level of interest of a user at a first one of the locations in the main video feed based on a set of physical characteristics of the user. The approach identifies a first one of secondary video feeds that corresponds to the first location and, in turn, resizes the first secondary video feed based on the detected level of interest.