When AI creates the movie trailer



It’s the old cliché: if there’s one thing artificial intelligence isn’t, it’s that it’s creative. For example, computers can’t come up with stories worth reading, they say. But after the triumphant advance of machine learning (ML) and text generators based on large language models based on it, it is becoming increasingly evident that AI systems can become authors as well. – although this is only successful because they now perfectly imitate human writers. and the texts can thus “foresee”.

A team from the University of Edinburgh have now trained an AI system to create movie trailers. Such systems already exist, but their performance still needs to be improved. That’s why the Scots are using a new approach that combines old models with improved data sources. The end result was so good that human evaluators at the Amazon Turk crowdworking service preferred variants of the computer whose models had been trained in a totally unsupervised (“unsupervised ML”) fashion, i.e. – say from the new algorithm.

More from MIT Technology Review

More from MIT Technology Review

More from MIT Technology Review

The team of Pinelopi Papalampidi, Frank Keller and Mirella Lapata from the Institute for Language, Cognition and Computation at the School of Computer Science at the University of Edinburgh shows in its pre-printed paper their approach based on neural networks. To analyze an existing movie, both its script and the video are used; A graphic model is then created from the film. The system can identify the narrative structure and mood of individual sequences as well as important segments (‘turning points’) in the storyline according to film theory – such as setbacks for the lead actor, the ‘call to action’. action ”, story highlights or the grand finale.

By identifying the mood of areas of the film, the computer trailer can conform to certain narrative concepts of how trailers should generally be constructed. This includes, for example, that movie previews often begin with a medium intensity scene. However, so far the system has only been able to identify basic moods, but not the characters’ finer feelings – such as fear, joy, or sadness. A total of 41 trailers have been developed by the AI ​​system.

The Scottish Researchers system is not yet so good that it could replace human creatives – but it seems to beat previous systems that were used to generate trailers, including CCANet for Trailers and the Supervised and Trained variant. of the Graphtrailer model. “We model the movies as graphs in which the nodes are individual parameters and the edge areas represent the semantic relationships between them,” the researchers said. The system could one day help at least speed up trailer production.


Source of the article

Disclaimer: This article is generated from the feed and not edited by our team.



Comments are closed.