The advancement of information technology along with the proliferation of communication and multimedia, have increased the demand of digital contents. Video data of digital contents such as VOD (video on Demand), NOD (News on Demand), Digital Library, IPTV (Internet Protocol Television), and UCC (User Created Contents) are getting more permeated in various application fields. Video data have sequential characteristic besides providing the spatial and temporal information in its 3D format, making searching or browsing ineffective due to long turnaround time.In this paper, we suggest ATVC (Authoring Tool for Video Contents) for solving this issue. ATVC is a video editing tool that detects key frames using visual rhythm and insert metadata such as keywords into key frames via XML tagging. Visual rhythm is applied to map 3D spatial and temporal information to 2D information. Its processing speed is fast because it can get pixel information without IDCT (Inverse Discrete Cosine Transform), and it can classify edit-effects such as cut, wipe, and dissolve. Since XML data save key frame information via XML tag and keyword information, it can furnish object-based retrieval or content-based retrieval.