This disclosure pertains to devices, methods, systems, and computer readable media for generating and/or interpreting image metadata to determine input parameters for various image processing routines, e.g., filters that distort or enhance an image, in a way that provides an intuitive experience for both the user and the software developer. Such techniques may attach the metadata to image frames and then send the image frames down an image processing pipeline to one or more image processing routines. Image metadata may include face location information, and the image processing routine may include an image filter that processes the image metadata in order to keep the central focus (or foci) of the image filter substantially coincident with one or more of the faces represented in the face location information. The generated and/or interpreted metadata may also be saved to a metadata track for later application to unfiltered image data.