A virtual reality movie is a kind of movie that is planned, shot and produced using virtual reality technology and viewed by the audience using virtual reality devices. This paper proposes using a 3D virtual camera to improve the shooting of virtual reality movies on screens. The camera focal length is utilized to regulate the size of the scene view body, and the adaptive Kalman filter method is employed to compensate for the camera’s attitude data. Select the most commonly used optical motion capture technology for virtual reality movies, build a motion capture system model, screen the optical motion capture data, and introduce the decision tree learning method in machine learning to analyze the correct rate of virtual motion recognition matching. Collect the research data, analyze the gender significance difference of the researched people by using the independent sample test, and share the immersion effect of virtual reality movie works by combining the four-dimensional experience of the researched people’s sensory, interaction, ease of use, and emotion. The virtual reality movie works have good performance in the four aspects of user sensory experience, interaction experience, ease of use experience and emotional experience, among which the mean value of sensory experience is 4.3426, which is the most outstanding, indicating that the virtual reality movie that integrates 3D image generation technology, live preview technology and performance capture technology has strong immersive creation ability.