AI didn’t just drop at this year’s Tribeca Film Festival

For all the sound that is made what about generative AI is poised to revolutionize the film industry, there hasn’t been any work it was created by technology that looks like entertainment that people pay to see. Many of the industry’s AI initiatives are still a work in progress a brief burst of unrelated images. And some more The biggest AI collaboration in Hollywood to be he suddenly evaporated in ways that make it seem like studios can’t rely on new technology coming out of Silicon Valley. Most of the time, short form video slop it seems like the only thing that big production houses can do cooking with gen AI. But that could change if studios take cues from some of the experimental productions that debuted at the Tribeca Film Festival.

Although it seems unlikely that gen AI will be able to whip up the entire fabric of cinema, Tribeca had several films that showed how human artists can use technology in compelling ways.

Although none of the AI-driven movies shown at Tribeca were as scary as the movie industry OpenAI and xAI polluted the internet with, other projects were excellent examples of why artificial objects are lifeless compared to human-made art. Roaring -short film made by Illuminai Studios – felt like a confusing piece of AI-generated video rather than interactive video. And Company Profile Asteria Film Co., Ltd.‘s ChikaBOOM! they didn’t have the bright sights and sounds of the fast-paced fantasy of a magician in training to draw you in.

Roaring and ChikaBOOM!All of this rigor appears to be a reflection of the technical restrictions placed on their AI-forward manufacturing process. But other films, such as Google DeepMind’s Dear Upland Neighbors and OpenAI Bad Sunshowed how it is possible for filmmakers to avoid those problems when gen AI is used in a very smart way.

Written and directed by Pixar veteran Connie Qin He in in collaboration with Google DeepMind researchers, Dear Upland Neighbors tells the story of a very tired girl who is trying to sleep. All Ada (Márcia Mayer, who also made the short) wants to rest in peace for a few hours before getting up and going back to work. But every time he starts to doze off, a loud noise from the neighbor’s upstairs apartment shakes him and leaves him wondering what he might be doing in the middle of the night.

Giving Dear Upland Neighbors‘ the world is a different style, Wrote Pixar producer Yingzong Xinwho painted artwork in Photoshop and on paper using acrylics. The beauty of those illustrations was important in the delivery Dear Upland Neighbors‘ an interesting story in life, but it also presented challenges for DeepMind researchers. With so many types of AI animations, the image quality can be difficult to convert into a good visual quality. But DeepMind’s engineers created the usual models What? and Picture which was designed to provide Dear Upland Neighbors‘ artists can change their output.

Because the customized models were trained on Xin’s skills, he was able to create designs that followed His vision for the project. The sound-to-video interface was great for recreating some of the text, such as how words appear when objects interact. But the real construction Dear Upland Neighbors‘ events in a way that would tell a coherent story, a small group of creative people should act normally. By creating animations and Autodesk Maya (industry standard for 3D rigging and VFX), Dear Upland Neighbors‘ the production team can ensure that the events will be carried out according to their wishes. And by feeding roughs into Veo, artists can create images that are polished and ready to be enhanced with additional tools created by Veo and Imagen.

More than any other film at Tribeca, Dear Upland Neighbors felt like a lesson in how artificial AI can be used as a popular tool that helps artists as they create their ideas. The entire flow of the film relied on human-made art and people making smart decisions that text-to-video producers couldn’t make on their own. It is important to remember that Dear Upland Neighbors it wouldn’t look as good if it were made with the vanilla version of Google. These models worked well for this short task, but they should wait for a project that is also a product of Google’s technology.

Dear Upland Neighbors it was a much more interesting watch than the films that OpenAI brought to this year’s festival. A semi-autobiographical play by Alice Gu Smoking used Sora to recreate Palisades Fire, by Youssef Michraf Bad Sun consists of several images created with OpenAI tools. Looking at all these videos, you can see how their filmmakers work around the limitations of the AI ​​genre. Great shot inside SmokingThe fire scenes looked like paintings, but the result was a little better – the ones that were painted using it Volume– as the setting – of a mother and her son trying to escape the fire in their car. Many a Bad Sunto shoot the end of only a few seconds and the only person speaking is an invisible narrator, but the film’s story of how human life is being disrupted by artificial intelligence makes those points seem like deliberate choices.

OpenAI’s presence at Tribeca was amazing the company’s latest decision to shut down Sora completely. Sora’s sudden shutdown was the lead OpenAI long video Critterz unable to make his appearance at this year’s Cannes Film Festival. It seems that OpenAI may be moving away from technical movies, but there are other players in the next gen AI architecture that filmmakers can use to achieve their projects.

With only $2,000 spent on computer equipment, writer/director Ash Koosha was able to produce it all by himself. Violets dreama drama about the international protests that have rocked Iran over the past year. Using Kling AI, Claude, Gemini, and Nano Banana, Koosha tells a fictional story about a group of people who are trapped in a street while the police roam the streets harassing civilians. The project took Koosha several weeks to complete on his own, and while it’s backed by a strong legend, it doesn’t break any ground.

Watching all these videos, I realized that there is no future in which studios are creating commercially profitable projects by feeding the movements of AI models. This kind of stuff may not go away, but it’s not something Hollywood heavyweights would want to put their names on. What seems likely to be even bigger is big AI companies like Google partnering with studios to create bespoke models that match real-world functionality. And that workflow only works best when it’s led by human artists with a clear creative vision.