A method, system, and device for processing video shooting are described. The method includes: shooting a subject in a background sample, the background sample, and a target background respectively, to generate a first video recording the subject and the background sample, a second video recording the background sample, and a third video recording the target background respectively, the shooting time lengths of the first video and the second video being smaller than the shooting time length of the third video; and comparing the first video and the second video, extracting images of the subject from the first video, generating a subject image frame sequence with a transparent background, sequentially superimposing each subject image frame in the subject image frame sequence to the third video, and generating a video file recording the subject and the target background.