iQIYI’s new AI film studio starts with actors on a real set
The Shanghai facility connects live-action filming to NadouPro’s AI post-production tools. The company’s account leaves cost and quality gains unproven.
Editorial visualizationiQIYI’s new AI Film Studio in Shanghai begins with people performing in front of cameras. The company says shooting on its inaugural project, the internet feature Cop in Trouble, started on October 28.
An English release dated November 1 in the United States—November 2 in Asia—describes set footage flowing into NadouPro, with performances preserved as visual surroundings change.
The distinction is consequential for how to understand an AI production facility. An actor’s timing, interaction with another performer and position in the frame can give a production something concrete to build around. Generating the surrounding world then becomes a task guided by recorded material, rather than starting every scene from an empty prompt.
A platform gets a physical production space
NadouPro predates the Shanghai studio. In a March 30 investor announcement, iQIYI said it had opened commercial testing of the system, describing a pipeline running from creative ideas to video generation. That was a software-development milestone, separate from the new filming facility.
The company’s September 10 platform update helps explain the production logic. It described tools for arranging 3D models and cameras on a timeline before final generation, plus panorama and multi-angle image tools for building reference libraries. Those are ways to establish composition, movement and consistency before asking a model to produce a shot.
The same September release distinguished available updates from a planned Canvas Agent and skills library. An announced roadmap should not be treated as a list of features already working in the Shanghai production. Nor does a model’s availability in a platform establish which tool made a particular scene.
For the new studio, iQIYI describes synchronized recording, continuity logging and footage review, connecting capture with generation.
All three announcements are company sources, not independent tests. The latest release supplies no comparative budget, production schedule or finished-film assessment establishing the Shanghai studio’s claimed advantages. Earlier platform descriptions explain the intended process; they cannot validate the new facility’s results.
The meaningful test will be visible in the finished work: whether generated environments remain coherent around performances across successive shots. A convincing individual image is only one part of a scene. The actors’ relationships, movement and dramatic rhythm still have to survive the edit.
Publication history
— First published.
No published corrections or earlier versions.
How this was made
UpThat uses AI-assisted research and writing. Source checking and publication follow our editorial review workflow. See the original documents above to assess the reporting.
How we use AI ↗

