TRACKED VIDEO ZOOMING

Systems, methods, and instrumentalities are disclosed for dynamic picture-in-picture (PIP) by a client. The client may reside on any device. The client may receive video content from a server, and identify an object within the video content using at least one of object recognition or metadata. The metadata may include information that indicates a location of an object within a frame of the video content. The client may receive a selection of the object by a user, and determine positional data of the object across frames of the video content using at least one of object recognition or metadata. The client may display an enlarged and time-delayed version of the object within a PIP window across the frames of the video content. Alternatively or additionally, the location of the PIP window within each frame may be fixed or may be based on the location of the object within each frame.