News
iQIYI Opens an AI Film Studio in Shanghai and Starts Shooting a Feature: What the Release Says and What We Cannot Yet See
· RenderBob team
iQIYI says it has launched what it calls China's first AI film studio in Shanghai, and that Cop in Trouble began filming on 28 October. The only account we found is a company release.
iQIYI says it has launched what it calls China's first AI film studio in Shanghai, and that Cop in Trouble began filming on 28 October. The only account we found is a company release, so the details below are the company's.
What was announced
An AI Weekly digest on 2 November, summarising a PR Newswire release we could not open, says the studio sits in iQIYI's Shanghai Creative Center. It pairs live-action shooting with the company's NadouPro AI post-production pipeline. The first title is an internet feature called Cop in Trouble, which began filming on 28 October. We found no second report, no budget, no running time and no description of which steps use AI.
What NadouPro is, from earlier releases
iQIYI's April release on Kyodo News PR Wire, dated 3 April, says it launched NadouPro on 30 March as an AI agent for professional film and TV work.
- It claims end-to-end support from script development and storyboarding to final output, with specialised agents for screenwriting, art direction, storyboarding and visual effects.
- It combines iQIYI's own QiZhi multimodal models with third-party models named as Seedance, Kling, Vidu, Hailuo and Wan.
- The release calls it pre-commercial, with internal testing from late 2025 and sixteen upcoming works in a Peter Pau × iQIYI AI Theater initiative.
- It says the platform helps long-form AI stories "approach live-action quality" beyond roughly 30 seconds, and gives no figures for time or cost savings.
What the claims are worth
"China's first" is the company's phrase, and we have not checked it. We have no samples, no independent review and no detail on how much of Cop in Trouble is shot and how much is generated. A platform is putting a shooting crew and an AI pipeline under one roof. It sits beside ByteDance's short-drama studio, which we covered on 28 October, and the Sanxingdui feature, where a "first" arrived with weak box office.
Questions to ask about a hybrid studio
- Which shots are filmed and which are generated? Credits and making-of material should say.
- Which models sit behind the pipeline? The April release names several third-party ones.
- Who controls each step? Can editors change a shot without rerunning a whole stage?
- How are likeness and rights handled for performers and generated characters?
- What leaves the pipeline? Ask about formats, metadata and prompt records.
- What does the finished film show? Judge it when it is released.
What to watch
A second source or iQIYI's own announcement, the film's release plan, any breakdown of live-action and AI work, and audience and critic response.
RenderBob 3 and this story
RenderBob 3 treats a generated shot as one more lane beside filmed material. Clips sit on their own timeline lane and the model is chosen per clip, so a hybrid cut can show where the AI work begins and ends.
More from the blog
- The New York Times and Three Newspaper Publishers Sue Brave Over AI Scraping: What Is Alleged, and Why an Unnamed Crawler Matters for Opt-Outs
On 2 November the New York Times sued Brave Software in Manhattan, and three other US newspaper publishers filed a similar suit hours later. They allege Brave scrapes journalism and sells access to AI companies.
- Mischief 1 Is a Cheap Image-to-Video Model Built on Wan 2.2 With a "Neuron-Derived" Speedup: What Is Claimed and What Is Not Yet Tested
The Biological Computing Co. released Mischief 1 on 30 October, an image-to-video model built on Wan 2.2. It says the model is about five times faster at the same quality. No independent test is published.