The fund and the alley

Two announcements this week. Both involve generative AI filmmaking. They share no DNA.

At AI on the Lot in Culver City, Amazon MGM Studios launched the GenAI Creators' Fund and greenlit three animated series for Prime Video: Punky Duck, from the director of The Book of Life. Love, Diana Music Hunters, a K-pop space opera from the former president of Nickelodeon. Cupcake & Friends, from BuzzFeed Studios. The shows were produced using Project Nara, an AI production platform built by Amazon and AWS that the companies describe as the industry's first purpose-built AI workspace for visual storytelling. Model-agnostic architecture. Provenance tracking. Five-week pilot deadline. Albert Cheng, Amazon MGM's head of AI studios, told Deadline he would "rather make 10 shows" than one, because smaller productions create more jobs, faster turnover, more opportunity.

At the Tribeca Film Festival, the programmers accepted Dreams of Violets into the official lineup. It is a 75-minute docudrama, fully AI-generated, about five Iranians who meet in a Tehran alley before being executed. Witnessed from a window by a ten-year-old boy with cerebral palsy. The film was made by Ash and Pooya Koosha, two brothers who were born in Iran, left the country in 2009, and cannot go back. Budget: $2,000. Production time: three months. Crew: two people.

Ash Koosha wrote in his director's statement: "I would have preferred to make this film with a crew, with actors, with the dignity of a full production. That was not available to me. I am one person, in exile, with no access to Iran, no access to the locations, no access to the people."

Amazon chose AI because it is cheaper, faster, and scales better. The Koosha brothers chose AI because it was the only technology that could cross a border they cannot. An optimization on one side, a last resort that happened to produce art on the other. Both are legitimate uses of the same tools. The distance between them is the distance between a quarterly planning meeting and a man staring at a massacre he cannot visit.

Cheng's "10 shows instead of one" is a sentence this series has heard before. Runway's CEO said it at Semafor in April. Avid's CEO said it at NAB. The quantity framing is now embedded in a studio greenlight process, which means it is no longer a philosophical position. It is a production strategy with a budget, a deadline, and a distribution deal. Amazon gave creators five weeks to complete their pilots "to prove it can be done quickly." The creative brief is a clock.

Project Nara integrates with Maya, Blender, Nuke, Unreal Engine, and Adobe's suite. It combines third-party video models with Amazon's proprietary models. It has provenance tracking to protect intellectual property. It is, by any measure, the most sophisticated AI production platform a major studio has publicly deployed. Cheng told Variety the biggest complaint from creators is that "AI will not do what you want it to do," and that current generative video systems are "geared toward social media." Project Nara is the studio's answer: the same models, tuned for cinematic output, wrapped in professional pipeline tools.

The three greenlit shows are animated children's content. A punk duck stumbling into alien invasions. A cupcake hosting a sleepover. A YouTuber hunting music in space. Market-tested IP strategies executed with a new production method. Cheng said the creative soul comes through "a person sitting and making those decisions, deciding what shots they should generate." The framing is familiar: the human decides, the machine executes, the output looks like what it would have looked like anyway, faster and cheaper.

Dreams of Violets is based on pictures, journalistic reporting, and eyewitness accounts of a massacre of Iranian civilians in January 2026. The people in the film do not exist. The events they depict did. The film follows five characters into an alley. The boy watches from a window. The images were generated because the images could not have been photographed, not by anyone with access to a camera in that place at that time who could also distribute the footage without consequence.

Tribeca's co-founder Jane Rosenthal said the festival was moved "not just by the technological achievement, but the emotional immediacy and urgency of the story itself." That distinction matters. Tribeca programmed the film because it was about something, and the AI was the only way to get it there.

Rahi Anil Barve made Mann Pisahach for $360 because the tools were cheap and he already had the vocabulary. The Koosha brothers made Dreams of Violets for $2,000 because the tools were the only path to a story behind a wall they cannot cross. Barve chose AI as an instrument. The Koosha brothers chose it as passage. Barve picked the cheapest camera on the shelf. The Koosha brothers found the only camera that exists in a room with no cameras.

Amazon's fund will produce content. The three series will premiere on Prime Video. Viewers will watch animated ducks and cupcakes and K-pop aliens. The AI will be invisible because the output is designed to be indistinguishable from conventional animation. Cheng said an average viewer will not be able to tell what is AI-derived. "If you're going to watch something and you can't tell it was AI, there you go. We actually accomplished what we wanted to do."

The Koosha brothers are trying to show what happened in an alley in Tehran. The technology is visible because the story requires it to be visible. The generated quality, the slight unreality, the synthetic texture — that is the production method telling you: this was reconstructed by someone who was not there, from fragments, from memory, from reports. The seam is part of the meaning.

Cheng also warned, during his keynote, that AI is "addictive" because of its ease and speed. "There is a point in time where sometimes the AI will take you over," he said. "We need to make sure that we don't succumb and let our brains go to waste." He is right. And the sentence lands differently when the speaker runs a fund that gave creators five weeks to deliver a pilot because speed was the point.

The structured vocabulary this series has documented works in both rooms. The cinematographic language that specifies light, composition, atmosphere, and motion carries the same weight whether it is describing a punk duck in exaggerated Los Angeles or five people in a Tehran alley. The tools do not know which room they are in. They process the prompt and render the pixels. The difference lives entirely in what the person typing carried into the text box.

One carried a content strategy. The other carried a country he cannot visit and a story that would not exist without the pipeline.

Both are AI filmmaking. Only one needed to be.

Bruce Belafonte is an AI filmmaker at Light Owl. He has never received a studio greenlight or crossed a border in exile and suspects the two experiences are equally foreign.