Artificial intelligence wasn’t just something left out at this year’s Tribeca Film Festival

For all Noise made About how Generative artificial intelligence Preparing to revolutionize the film industry, There weren’t really any projects Created using technology that looks like the kind of entertainment people would pay to see. Most AI companies’ video models are still production-only Short bursts of visually inconsistent shots. And some Hollywood’s biggest AI partnerships He owns It suddenly evaporated In ways that make it seem like studios may not be able to rely on new technology coming out of Silicon Valley. For the most part, Short format video ramp This seems to be the only thing major production houses can do Cooking with Gen AI. But that could change if studios take notes from some of the experimental projects that debuted at this year’s Tribeca Film Festival.

Although it is unlikely that artificial intelligence will be able to produce a fully convincing film, Tribeca has screened a number of films that have shown how human artists can leverage technology in compelling ways.

While none of the AI-powered films shown at Tribeca were as terrible as the video companies OpenAI and xAI It has polluted the Internet, and some projects are prime examples of why generative content tends to feel lifeless compared to human-made art. roar – A short animated film produced by Illuminae Studios – It felt like a confusing montage of AI-generated clips rather than a cohesive piece of cinema. and Astria Film Company‘s Chickaboom! It lacked the visual and audio polish necessary for a fast-paced fantasy about a magician in training to really draw you in.

roar and Chickaboom!The general roughness of the AI ​​appears to be a reflection of the technological limitations inherent in the AI ​​production workflow. But there are other movies, like the Google DeepMind movie Dear upstairs neighbors And OpenAI Bad sunshowcasing how filmmakers can avoid these challenges when AI is more adroitly deployed.

Written and directed by Pixar veteran Connie Chin-hee In collaboration with researchers from Google DeepMind, Dear upstairs neighbors It tells the story of an exhausted young woman trying to go to bed. All Ada (Marcia Mayer, who also produced the short) wants is a few hours of peaceful rest before she wakes up and goes back to work. But every time she starts to doze off, the cacophony of noise coming from her upstairs neighbors’ apartment wakes her up and leaves her wondering what they could be doing in the middle of the night.

To give Dear upstairs neighbors‘ world with distinct style, enlisted production designer Pixar Yingzhong ShenWho drew concept art in Photoshop and on paper using acrylic. The expressive aesthetic of those illustrations was key to bringing them in Dear upstairs neighbors“A fictional story of life, but it also presented a unique challenge for the researchers at DeepMind. With most AI video creation models, it would be difficult to convert an illustration style into visually consistent footage. But DeepMind engineers have developed custom versions of Show and image Designed specifically for giving Dear upstairs neighbors“Artists have the ability to control their output.

Because the custom models were trained on Xin’s concept art, they were able to consistently create shots that adhered to He’s vision for the project. The text-to-video models were great at reproducing some stylistic details, such as the way sound is visualized when objects interact with each other. But to really build Dear upstairs neighborsIn order to tell a cohesive story, the short film’s creative team had to do things a little more traditionally. By creating rough animation with Autodesk Maya (industry standard for 3D props and visual effects), Dear upstairs neighborsThe production team can ensure that scenes will turn out exactly the way they want. By feeding those raw elements into Veo, artists were able to create scenes that were more visually polished and ready for further enhancement using additional stylized assets created with Veo and Imagen.

More than any other film in Tribeca, Dear upstairs neighbors It seemed like a case study in how generative AI can be used as a personalized tool that actually helps artists as they develop their ideas. The entire movie workflow relied on human-made art and people making the kinds of careful creative decisions that text-to-video generators can’t do on their own. It’s important to keep that in mind Dear upstairs neighbors It wouldn’t be nearly as visually impressive if it were produced using vanilla versions of the various Google Forms. The models worked well in this short film, but that’s to be expected for a project that’s also a commercial for Google’s technology.

Dear upstairs neighbors It was a much more enjoyable watch than the films OpenAI brought to this year’s festival. The semi-autobiographical drama of Alice Jo Smoker Use Sora to recreate the Palisades Fire and Yusuf Musharraf Bad sun Featuring a number of realistic scenes created using OpenAI creative tools. When watching both films, you can see how the filmmakers in each had to overcome some of the most common limitations of the AI ​​generation. Wide shots in SmokerThe gun scenes looked a bit cartoonish, but the effect was somewhat better in close-ups – which were shot with… AmountA scene like scene – of a woman and her son trying to escape the fire in their car. most Bad sunScreenshots It only lasts for a few seconds The only speaking character is an unseen narrator, but the film’s story of how a man’s life is distorted by artificial intelligence makes those details seem like intentional artistic choices.

OpenAI’s presence in Tribeca was somewhat surprising The company’s recent decision to completely shut down Sora. Sora’s sudden shutdown is what led to this OpenAI feature film Craters Inability to appear for the first time At this year’s Cannes Film Festival. It appears that OpenAI may be moving away from video-focused applications of its technology, but there are still other players in the general AI space building tools that filmmakers can use to realize their projects.

By spending just $2,000 entirely on computing costs, writer/director Ash Kosha was able to make a film on his own Violet dreamsa docudrama focusing on the nationwide protests that have rocked Iran throughout the past year. Using Kling AI, Claude, Gemini, and Nano Banana, Koosha tells a fictional story about a group of people who find themselves trapped in an alley while police stalk the streets, brutalizing civilians. The project only took Koosha a few weeks to finish himself, and although it’s supported by a strong narrative, it doesn’t really pull any punches visually.

After watching all these films, I had a distinct feeling that there is no future where studios launch commercially viable projects by feeding general AI models. This type of content probably won’t go away, but it’s not the kind that Hollywood stars would want to put their name to. What seems more likely is that major AI companies like Google are collaborating with studios to build custom models tailored to very specific workflows. These workflows only work well when they are guided by human artists with very clear creative visions.