Hollywood’s pivot to AI video has a prompting problem
It has become almost impossible to travel Internet without having a video generated by Ai-Vous. Open essentially any social media platform, and it will not take long before a strange clip of a false natural disaster or animals do impossible things on your screen. Most of the videos are absolutely terrible. But they are almost always accompanied by hundreds, even thousands, tastes and comments from people who insist that the content generated by AI is a new form of art that will change the world.
This was particularly true for AI clips which are supposed to seem realistic. No matter how strange or aesthetically inconsistent images can be, there is generally someone who proclaims that it is something that the entertainment industry should be afraid. The idea that the video generated by AI-AI is both the future of the cinema and an existential threat to Hollywood has made its way as forest fires among boosters for relatively new technology.
The thought of the big studios embracing this technology as it feels doubtful if we consider that, often, the release of AI models is simply not the kind of thing that could be shaped in a film or a series of quality. It is an impression that the filmmaker Bryn Mooser wants to change with Asteria, a new production house he launched last year, as well as a next feature film generated by the AI of Natasha Lyonne (also the partner of Mooser and an adviser at the end of the evening, a studio focused on the IA generative that the mooser film and television company acquired the last year).
Asteria’s great sale argument is that, unlike most other AI outfits, the generative model it has built with the Moonvalley research company is “ethical”, which means that it has only been trained on properly approved equipment. Especially in the wake of Disney and Universal Suite Midjourney for copyright violation, the concept of ethical generative can become an important part of the way in which AI is more widely adopted in the entertainment industry. However, during a recent conversation, Mooser emphasizes that the clear understanding of the company of what the generator is and what he does not like to distinguish Asteria from the other players of IA space.
“While we are starting to think of building Asteria, it was obvious to us as filmmakers that there were big problems with the way AI was presented in Hollywood,” said Mooser. “It was obvious that the tools were not built by anyone who had already made a film.
In the opinion of Mooser, part of the reason why some enthusiasts quickly called the generative video models a threat to traditional film workflows are summed up with people by assuming that the images created from invites can reproduce the real thing as effectively as what we have seen with imitative music and generated by AI. It was easy for people to reproduce the singers’ voices with a generative AI and produce passable songs. But Mooser thinks that, in its haste to normalize the AI generation, the technological industry has confused audio and the visual release in a way that is in contradiction with what really makes good films.
“You cannot go and say to Christopher Nolan:” Use this tool and send a text to your way to Odyssey“Said Mooser.” While Hollywood residents had access to these tools, there were a few things that were really clear – one being that the form factor cannot work because the amount of control that a filmmaker needs is summed up in terms of pixels in many cases. »»
To give its cinematographic partners more of this granular control, Asteria uses its generative basic model, Marey, to create new models specific to the project formed on an original visual material. This would allow, for example, an artist to build a model that could generate a variety of active ingredients in his distinct style, then use it to fill a world full of different characters and objects that adhere to a unique aesthetic. It was the Asteria workflow used in his production of the short film by musician Cuco “a love letter”. By forming the Asteria model on 60 original illustrations designed by artist Paul Flores, the studio could generate new 2D active ingredients and convert them into 3D models used to build the fictitious city of video. The short film is impressive, but its heavy stylization explains how projects with a generative AI at the base must often work within the visual limits of technology. He does not have the impression that this workflow offers a control in the pixels for the moment.
Mooser says that, according to the financial arrangement between Asteria and its customers, filmmakers can keep the partial property of the models after their end. In addition to the original license fees, Asteria pays the creators of the equipment on which its basic model is formed, the studio “explores” also the possibility of an income sharing system. But for the moment, Mooser focuses more on artists’ victory with the promise of a drop in initial development and production costs.
“If you make a Pixar animated film, you could come as a director or writer, but it is not often that you will have the property of what you create, the residues or the cutting of what the studio does when they sell a lunch box,” said Mooser. “But if you can use this technology to reduce the cost and make it financial independently, then you have a world where you can have a new financing model that makes real property possible.”
Asteria plans to test many Mooser beliefs in the transformer potential of the generator with Strange valleyA feature film to co-written and directed by Lyonne. The live film focuses on a teenager whose trembling perception of reality makes her begin to see the world as more video games. A lot of Strange valleyis fantastic, Matrix-The visual elements of the type will be created with the internal models of Asteria. This particular detail makes Strange valley It looks like a project designed to present the hallucinatory inconsistencies that generative AI has become known for intelligent aesthetic characteristics rather than bugs. But Mooser tells me that he hopes that “nobody never thinks about the part of the AI” because “everything will have the human touch of the director”.
“It’s not like you plan to send SMS,” so they enter a video game “and watch what’s going on, because no one wants to see this,” said Mooser. “It was very clear because we think about it. I don’t think anyone who doesn’t want to see what computers dream of.”
Like many AI generative defenders, Mooser considers technology as a “democratic” tool that can make art creation more accessible. He also points out that, in the right circumstances, a generative AI could facilitate the production of a film for around $ 10 to 20 million rather than $ 150 million. However, securing this type of capital is a challenge for most younger and promising filmmakers.
One of the major Asteria sales arguments that Mooser mentions me several times is the AI generation potential to produce finished works faster and with smaller teams. He has formulated this aspect of an AI production workflow as a positive that would allow writers and directors to work more closely with key collaborators such as art and VFX supervisors without having to spend so much time going back and forth on revisions – something that tends to be more likely when a project has many people who work there. But, by definition, the small teams result in less jobs, which raises the question of the potential of the AI to put people unemployed. When I send this with Mooser, he underlines the recent closure of the VFX House Technicolor group as an example of the continuous upheaval of the entertainment industry which began to leave the workers to unemployment before the generative media threshing of AI came to its current fever.
Mooser took care not to minimize that these concerns concerning a generative AI were a large part of what plunged Hollywood in a double strike in 2023. But it is resolved in his conviction that many workers in the industry will be able to pivot laterally in new careers built around the generator if they are open to embrace technology.
“There are filmmakers and VFX artists who are adaptable and want to look into this moment in the same way that people were able to switch from the edition to the film on Avid,” said Mooser. “People who are real technicians – artistic directors, filmmakers, writers, directors and actors – have an opportunity with this technology. What is really important is that we, as an industry, know what is good on this subject and what is bad on this subject, what is useful to us to tell our stories and what will really be dangerous. “
What seems rather dangerous in the interest of Hollywood for the generating AI is not the “death” of the biggest studio system, but rather the potential of this technology to facilitate work for studios with fewer real people. This is literally one of the major arguments of sale in Asteria, and if its workflows became the standard of industry, it is difficult to imagine that it evolves in a way that could accommodate the entertainment workforce today in transition to new careers. As for what is good on this subject, Mooser knows the right discussion points. Now he has to show that her technology – and all the changes it implies – can work.
