Netflix has acquired InterPositive, the artificial-intelligence company co-founded by Ben Affleck to work on material that has already been filmed. The distinction matters: the stated goal is not to ask a model to invent a film from nothing, but to give directors and productions tools capable of modifying, completing and refining existing images. The center of the operation is therefore post-production—the point at which a shot can be recomposed, corrected or expanded without necessarily returning to the set.
According to Netflix, InterPositive will operate as an internal structure serving the streamer’s creative partners. The applications cited include removing unwanted elements, reconstructing parts of a frame, harmonizing lighting and creating missing imagery from the context of a production. Reuters described the acquisition as moving an experimental technology into the industrial scale of the platform; Variety later reported a value of $587 million for the transaction, based on company documents.
The essential point is not only the figure. Netflix did not buy a new button for generating video, but expertise built close to the language of the set. InterPositive starts from the materials of a single production and uses them as reference: characters, costumes, environments, lenses, light and visual continuity. In theory, this reduces the risk of producing images that are generically ‘cinematic’ but foreign to the film. The model has to learn the identity of that specific work rather than impose an average derived from thousands of different works.
For a director, the potential advantage is concrete. A scene that would otherwise require reshooting can be recovered; a safety wire can disappear without manually reconstructing dozens of frames; an incomplete background can be extended; a lighting variation can be made coherent with the rest of the sequence. These are tasks that VFX and compositing have performed for years. AI promises to accelerate some of them, making interventions possible that a smaller production might previously have removed from the film for reasons of time or budget.
The deeper issue begins here: acceleration does not mean eliminating work. Every correction still has to be checked frame by frame, integrated with color, verified in motion and approved narratively. A model can propose a plausible reconstruction; a VFX supervisor must understand whether it respects the physics of the scene, continuity and the director’s intention. Professional value shifts from the execution of repetitive steps alone toward control, selection, traceability and direction of the result.
A question about data remains. If the system is trained or adapted on the footage of a production, someone must establish who can authorize that use, how long the material is retained and whether faces, voices, performances or set designs may be reused elsewhere. A license to complete a single sequence is not the same as the right to create new performances. Contracts, consent and version records therefore need to enter the same technical workflow that manages the files.
Affleck’s public position also makes the case significant. The actor and director has repeatedly argued that AI can imitate forms and produce variations, but cannot replace the judgment that turns a possibility into an artistic choice. InterPositive translates that idea into a product: the machine is not asked to become the author of the film; it is inserted into a chain in which real people define the problem, compare alternatives and decide what deserves to remain in the frame.
Netflix’s scale makes this integration an industrial precedent. The company has said that generative tools have already touched hundreds of titles, especially in production and post-production stages. Bringing the technology in-house means connecting it to pipelines, material security and common standards. But it also means concentrating tools and decisions inside a platform that controls production, distribution and viewing data. Efficiency will need to be accompanied by transparency around credits, accountability and employment impact.
INVISIBLE POST describes the paradox well: the better these tools work, the less the public notices their intervention. The point will not be identifying every modified frame, but knowing that behind that image there are permissions, professionals and documented decisions. The acquisition of InterPositive does not announce a cinema without sets. It shows instead that the set continues inside the machine, and that AI post-production is becoming a stable part of audiovisual infrastructure.