Generative AI for production centre

The accelerated advancement of artificial intelligence, in particular generative artificial intelligence (genAI), has raised extensive discussions in the media and entertainment industry. The Hollywood screenwriters’ strikes of 2023 highlight a growing concern about the role AI could take in content creation, raising debates about its potential to equal, or even replace human ingenuity. GenAI, through machine learning algorithms, enables the autonomous creation of original text, images, video and audio from simple prompts, opening new perspectives for the automation of creative processes. In this context, this thesis explores the integration of genAI technologies within a traditional production workflow for the realisation of an animated pilot based on Jules Verne’s “Around the World in 80 Days”, revisited in a futuristic key. The project, developed in collaboration with CRITS RAI, as part of a broader initiative presented at IBC 2024, aims to experiment with the application of genAI at all stages of audiovisual production, from concept art development to animation. This project involves a comprehensive analysis of the current generative AI applications available on the market and in the research field, to identify both their strengths and limitations. The workflow combines different tools for scriptwriting, image creation, animation, and sound design, including technologies like Dramatron, ChatGPT, Firefly, Stable Diffusion, Runway, Reespecher and others. Human intervention is crucial at every step to guide the AI and correct its imperfections, such as hallucinations and biases derived from the data on which the generative models were trained. The findings demonstrate that although AI AI has the potential to automate many parts of the production process, repeated iterations between humans and AI are essential to achieve the desired quality and consistency, with ongoing refinement of prompts. The latest multimodal models offer more control over the output, allowing for improved consistency issues typical of AI generations. The main goal of this research is to find a balance between automation and human input, shaping the possible future of the media and broadcasting industry.