A multimedia system product for use in physical fitness training including a method and system for producing customized physical fitness programs through an internet-based environment. The user is allowed to select the individual components of the exercise video and manipulate motion capture data by using a graphics engine to customize the "look and feel" of the customized workout through available training avatars. The individual components eligible for inclusion are included in a database containing a plurality of audio and visual media files of personal trainers performing exemplar exercises. The compiled instructional video is then available for download and installation on a video-enabled personal media player.