Artificial intelligence could write a script

AI and the cinema of the future

Deep Tech is radically changing the world of film: algorithms suggest films on Netflix based on user behavior and even write scripts. How does it look?

Bringing the right films together with the right audience - what streaming platforms like Netflix have long mastered with flying colors must also be possible in the cinema. At least that's what Martina Berger, Jannis Funk, Tobias Lohse, Paulo Ruhrländer and Jean-David Herld thought. Artificial intelligence, as used in the tried and tested digital recommendation and targeting models, has "great potential to get to know cinema-goers better and to make them personal film recommendations," the film and tech experts told Media Tech Hub. In April 2017 the five founded the start-up "Cinuru Research GmbH". With the help of investors such as the Seed Fund of the Hasso Plattner Institute, a business angel from the film industry, the FFA Filmförderungsanstalt and the Brandenburg State Investment Bank, they launched an innovative customer loyalty system for independent cinemas with the app of the same name. "Our KinoApp Cinuru rewards loyal visitors with rewards and personalized film recommendations," said the entrepreneurs. In addition, cinemas and the film industry can "optimize their marketing and program planning": The more data Cinuru Research can evaluate, the more distributors learn about their target groups and can take appropriate measures. Cinuru, which has been run by Paulo Ruhrländer since June 2020, was able to convince around 20 cinemas of its intelligent software. Further distributors are to join and use the collected data in the first test projects.

Where film and tech world meet

As self-proclaimed film enthusiasts and tech nerds, the Cinuru founders are in good company in the capital region: More than 4,000 film and TV companies are based in and around Berlin. The excellent production infrastructure, appropriate funding conditions and the creative atmosphere attract producers, film distributors, streaming portals and VFX companies from all over the world. This can also be confirmed by the Cinuru team, which has been at home in the MediaTechLab on the Babelsberg film site since the beginning of 2019: "For us, Potsdam is the place where the world of film and tech collide and where things get very creative," they rave about the network.

Promoting the exchange between the film industry and technology start-ups is also the concern of Erwin M. Schmidt, founder of the Berlin innovation studio “Cinemathon International”. As the initiator of the Film Tech MeetUp and the international Propeller Film Tech Hub, the managing director of the Association of German Producers has been dealing with innovation technologies and strategies for the industry for years: “We are constantly checking blockchain technologies and big data, artificial intelligence and data-driven Marketing, new forms of storytelling, sales distribution and data piracy, "he says," these topics are essential for the future of the film industry. "

KI named Benjamin writes scripts on his own

A collection of comical sentences and stage directions, everything is confusing - the 2016 science fiction short film "Sunspring" is not an award-winning one. Nevertheless, it went down in film history: The script was written by an artificial intelligence named Benjamin. "The approach that AI writes scripts has not manifested itself further," emphasizes Erwin M. Schmidt. The machine would only recognize patterns, "the creative gene" is missing. Benjamin was programmed with dozens of scripts from science fiction films, figured out which sequences of letters were common, recognized the structure and learned to imitate words, sentences, paragraphs and finally the writing of a script.

This pattern recognition may be a hindrance when writing an - artistically appealing - script. When evaluating a film script, on the other hand, it pays off: “A blockbuster works according to certain patterns,” the Berlin producer knows, “the distribution of dialogues, the appearance of characters, the dramaturgical arc - the AI ​​can compare hundreds of parameters and with a probability predict how successful a film will be. ”Corresponding deep-tech programs from providers such as Largo Films, the American start-ups Cinelytics and StoryFit or Scriptbook are already in use in Hollywood. In this country, the industry is still skeptical: "There is a fear that we will see more of the same," says Schmidt, "Outliers like the indie hit 'Systemsprenger' would fall through the cracks."

Personalized content with AI

Erwin M. Schmidt does not see the potential of deep tech and artificial intelligence in content creation anyway. In addition to targeting models for marketing à la Cinuru, he can imagine personalizing the content. "We have to move away from a final product towards stories that not only have different narrative leaps, but are constantly being changed and adapted to the context of the user." A British broadcaster would have made such an attempt in the laboratory: A story was shot in such a way that that it could be played in 50,000 variations - depending on the music preference, environment or availability of the viewer. In addition, its biological functions such as pulse, heart rate or skin tension were measured and a product was tailored based on these factors. “No new work is created, but something that already exists is adapted and personalized,” says Schmidt, summarizing the principle.

Another possible application of artificial intelligence sounds just as futuristic: When shooting a film, a lot of raw material is created that has to be sorted out for the editing process according to numerous criteria, such as technical quality. "Here you have to sift through a lot of material and pick out the scenes that work best," says Erwin M. Schmidt, "that is a physical process. The AI ​​could automatically take over this filtering out and only make the scenes available to the editor that are technically in order. ”In daily soaps with thousands of episodes that are structured according to a certain schematic pattern, the AI ​​could even produce a first raw version for editing and thus enable more efficient work processes.

Rethink post-production and automate it with AI

“To automate complex processes, which up to now were almost like a manufacture and mostly manually”, that is also the ambition of Holger Lehmann of the Berlin production company “Rotor Film”. In particular, the finalization of international productions such as “The Square” or “Iron Sky 2” has been challenging so far: huge amounts of data have to be transmitted, sound processes and color spaces have to be coordinated, and last-minute changes are necessary for final acceptance. Rotor Film would like to simplify these time-consuming and material-intensive processes with the research alliance DWerft at the Babelsberg site. The so-called metadata play an important role: Evaluation information, content or technical details are collected in production and distribution and anchored in the film. The more metadata there is, the easier it is to find and use the film. "In this way, the metadata becomes the fuel for artificial intelligence and a necessary prerequisite," Lehmann describes to MediaTech Hub Potsdam. “The question is: How can we think further about the use of metadata? I could save information about second or third use in other countries or analyze the optimal narrative thread of a romantic comedy. Special dialogue points, additional information on locations or music rights and language versions can be categorized in a targeted manner. ”Manual post-production sequences, such as noise reduction, could also be automated with the help of artificial intelligence in the future. It would be just one of the many application possibilities with which Deep Tech is changing the film world - from Berlin.