A video processing method, a terminal, and a server, where the method includes receiving a media presentation description (MPD) file sent by the server, where the MPD file includes region information of a region that can be independently decoded in the video, determining, according to the region information, a region used for playback on a terminal from the region that can be independently decoded, determining a to-be-acquired media segment according to the MPD file, acquiring a location in which data content is stored in the media segment, acquiring, according to the location in which the data content is stored in the media segment, the data content from the media segment stored in the server, and playing a picture of the region for the playback on the terminal. Therefore, it is not required to independently store a partial video on the server, thereby saving a storage resource of the server.