Application Methods of Deep Learning in Enhancing Visual Effects in Film and Television Post-production

Post-production of film and television is an important link related to the quality of film and television works. In response to the traditional method of relying on manual post-production of film and television works, this article uses artificial intelligence to improve visual enhancement and video editing of film and television works. Firstly, this article analyzes the current status of post-production in film and television. Three consecutive frames of images are selected from the video sequence image in film and television post-production, and the differential image of adjacent two frames is determined. Logical operations are performed on the two differential binary images to automatically obtain the ROI within the scene of the film and television post-production video image. Then, in order to achieve high-quality video restoration while automatically repairing the computational complexity during the video restoration process, a method of optical flow propagation based on global matching and Transformer encoder is proposed to effectively improve the accuracy of optical flow restoration in videos. Finally, this article takes a certain video clip from daily promotional activities as the experimental object, and uses the artificial intelligence post-production method proposed in this article to improve the visual effect of film and television works.