Darren Aronofsky and Google DeepMind Debut AI-Driven Short Film 'Ancestra' at Tribeca Festival
June 13, 2025
Darren Aronofsky has teamed up with Google DeepMind to produce a series of short films utilizing generative AI technology, with the first short, 'Ancestra,' premiering at the Tribeca Festival on June 13, 2025.
Directed by Eliza McNitt, 'Ancestra' is an 8-minute film that merges live-action filmmaking with generative AI, showcasing the collaborative efforts of over 200 artists and technologists.
The film draws from McNitt's personal experience of a life-threatening birth, highlighting pregnancy challenges that are often overlooked in cinematic storytelling.
To create a narrative inspired by her own biography, McNitt utilized personal family archives, including baby pictures and photographs taken by her late father, to train the AI models.
Innovative visual sequences in the film, including depictions of a woman's womb, were made possible through the integration of generative AI technology.
The production involved a large crew of 200, including artists and animators, who collaborated with AI to ensure emotional continuity and artistic integrity.
Aronofsky views AI as a natural evolution in filmmaking technology, likening it to past innovations introduced by pioneers like the Lumière Brothers and Thomas Edison.
'Ancestra' marks the inaugural project from Aronofsky's new venture, Primordial Soup, which aims to explore the intersection of technology and storytelling in filmmaking.
Primordial Soup is also focused on mentoring emerging filmmakers while refining AI tools through real-world applications, with two additional short films currently in development.
Google DeepMind provided early access to its generative AI tools, Veo 3 and Flow, to assist in the production of 'Ancestra'.
The film visually explores themes of motherhood and family, transforming personal archives and emotions into an intimate cinematic experience.
While the next two shorts from the Primordial Soup project have not been announced, they are expected to utilize Google DeepMind's video generation model, Veo.
Summary based on 4 sources