top of page

AI technology is giving new life to the lost scenes of The Magnificent Ambersons

  • Sep 5, 2025
  • 3 min read

5 September 2025

Dolores Costello and Tim Holt in The Magnificent Ambersons, 1942. Photograph: Donaldson Collection/Getty Images
Dolores Costello and Tim Holt in The Magnificent Ambersons, 1942. Photograph: Donaldson Collection/Getty Images

For decades, Orson Welles’s follow-up to Citizen Kane, The Magnificent Ambersons, has been a haunting legend in film lore. Completed in 1942, Welles’s version was intended to clock in at about 131 minutes, only to be radically altered by RKO studio executives. They slashed the runtime to 88 minutes and destroyed the excised footage an act Welles later lamented as a betrayal of his creative vision. Now, nearly 83 years later, an AI-driven restoration effort seeks to resurrect those missing scenes and offer a glimpse of the film as Welles intended.


The company leading the charge is Showrunner, a platform powered by AI and backed by Amazon under the stewardship of Fable Studio. Their mission is to recreate the lost 43 minutes of The Magnificent Ambersons not for commercial distribution, but as a cultural experiment in the possibilities of AI and cinematic memory. “We’re starting with Orson Welles because he is the greatest storyteller of the last 200 years,” says Edward Saatchi, CEO of Fable. He frames the project as a demonstration of positive, story-driven uses of AI in film.


Key to the effort is filmmaker Brian Rose, who since 2019 has painstakingly spent years reconstructing roughly 30,000 missing frames using traditional tools: 3D modeling of the original sets, alignment of camera movements with surviving script pages, production stills, and extensive archival materials. As he puts it, one sequence a single four-minute take tracking across a ballroom with multiple characters entering and exiting stands as a testament to Welles’s early mastery, and its loss remains a cinematic tragedy.


Showrunner will layer AI into this foundation, using its advanced generative model to recreate key frames and spatial settings. Live actors will be filmed in reconstructed scenes, with AI and VFX used to transplant the original actors' likenesses onto them. Tom Clive, a visual effects and face-swapping specialist, joins the project to oversee that delicate transformation.


This is not the first attempt to resurrect the Ambersons footage years ago a still-photo reconstruction premiered at the Locarno festival, and as recently as 2020 Turner Classic Movies funded a search for a surviving print in Brazil. None have succeeded in uncovering the original footage, leading to a near-mythic status for Welles’s lost vision.


Still, Showrunner treads a fraught path. The reconstructed footage remains unofficial and non-commercial; the company does not hold rights from Warner Bros. Discovery, which controls the film. Saatchi hints that if the finished project proves valuable, it might be offered to rights holders later, but for now it is destined for scholarly or demo contexts rather than theaters or streamers.


The initiative taps into a larger debate around AI’s role in film restoration. Purists balk at the idea of using machine learning to simulate lost scenes, fearing it compromises authenticity and undercuts the artistry inherent in physical film. Others, like Rose and Saatchi, argue that when a film is irretrievably lost, an informed reconstruction may be the only way to approach its original greatness.


The stakes are not just technological but personal. Welles, who had no final cut and watched his version dismantled while he was abroad, never got the opportunity to finish The Magnificent Ambersons as he intended. The studio’s re-edited ending and the destruction of the negative left him devastated. With recent labor strife around AI usage in Hollywood fresh in public memory, this project may become a flashpoint in how technology intersects with legacy and ethics.


In the end, this is a story not just about technology or history, but about grief, reclamation, and the yearning for what might have been. The fragments of Welles’s original film linger as a specter in film history. If the reconstruction succeeds, even as approximation, it could offer a poetic kind of closure an echo of ambition restored, a silhouette of a masterpiece finally meeting its reflection.

Comments


bottom of page