Deep Learning Cinema: Film Production in the Age of Generative Artificial Intelligence
Jeong, Chancheol (Hankuk University of Foregin Studies, S.Korea)
Since the turn of the 20th century, digital technology has completely reshaped the film industry. Now it seems that cinema is entering its next phase of techno-metamorphosis, which is coded by artificial intelligence technology, one that is much superior. The application of AI in film production might be considered still as part of the digital transformation but it is expected that it would end up with generating different ways films are produced and consumed. This paper briefly aims to examine the impact of artificial intelligence on the film industry and how AI technology is shaping the future of film production, by exploring examples of the application of AI in contemporary film production, ranging from pre-production to post-production, where deep learning has emerged as a game-changing technology.
First of all, I will examine how ScriptBook, one of the first AI-powered software to predict the success of movie, analyses key factors for movie profitability from the screenplay at the pre-production stage and could contribute to creating better scripts, limiting false decision-making, and improving financial decisions. In order to discuss the practical uses of AI in the post-production stage, I will examine neural network-based approaches used in film visual effects to create realistic and emotive digital characters and to make much faster VFX production environments by improving the accuracy and efficiency of labour-intensive VFX tasks. Some examples I consider in this paper include: Masquerade 2.0, which is Digital Domain’s ML driven facial capture system and was used in making Thanos’s face realistic and emotive in Avengers Endgame(2019); Arraiy, a ML-based visual effects software that are trained to accurately work out some of the most time-consuming and costly manual processing such as rotoscoping, camera match-moving, and object tracking. While discussing the practical uses of A.I in the film industry, this paper will look for co-creative ways in which A.I and human are able to interact and converge with their own specificity. For this, I will introduce ‘Deep.Photoplay,’ a research project, which I am co-working with my colleagues, to develop an AI machine that is able to produce musical accompaniments for silent films, create sound effects and co-work with human composers to compose original silent music scores.