A call for submissions issued on September 14 by the China Golden Rooster and Hundred Flowers Film Festival listed requirements that differ from traditional film applications.

Beyond the finished film, poster, and synopsis, applicants must also submit model toolchains, core prompts, AI generation workflows, and engineering screenshots for technical verification.

AI-generated footage must account for no less than 50% of the submitted work, with feature-length films running at least 60 minutes. These criteria come from the first standalone AI film festival brand introduced within the Golden Rooster system.

The Beijing and Shanghai International Film Festivals had already established AIGC-related sections. This year, the Hundred Flowers Awards added an AIGC recommendation category for the first time, and the Golden Rooster went further by writing feature-length duration, generation ratio, and production records into its public submission rules.

Meanwhile, distribution channels are starting to shift as well.

Two weeks ago, Journey to the West: The Later Story entered the prime-time slot on Hunan Satellite TV, marking the first time AIGC long-form content has reached this traditional television broadcasting scenario.

While film festivals and award systems carve out dedicated entry points and verification rules for AI works, AIGC long-form content is simultaneously moving into more conventional distribution channels. AI filmmaking is transitioning from experiments within tech companies and creator communities into more formal industry processes.

To understand how this new production method has reached this point, one must return to the most traditional production site of a film: the set.

When the Set Disappears

Filmmaking is fundamentally collective labor that depends on manpower, funding, and physical space.

Actors must arrive on time, while cinematography, lighting, art direction, costumes, and props require simultaneous coordination. A house that does not exist must be built, and a rain scene lasting only a few minutes demands water trucks, lighting equipment, and dozens of crew members. Once the machines start rolling, every day generates expenses, and if a scene is not completed, the subsequent schedule shifts accordingly.

Liu Chi, a commercial director with over a decade of experience who has worked on advertising, web films, and the virtual human short drama Liu Yexi, describes the changes brought by AIGC as "earth-shattering."

Earlier this year, her six-person team completed the 45-minute AI animated film Seven Days of Floating Life, which won Best Feature at the AIGC Film Unit of the 16th Beijing International Film Festival. The entire process from script to post-production took about a month and a half, with core video generation and post-production taking approximately three weeks.

For the theatrical project she is currently advancing, around a dozen people handle the AI generation segment. Even including scriptwriting, directing, storyboarding, concept design, and post-production, the full team still only requires twenty to thirty people.

"In the past, a standard film production might require a crew of over a hundred people. Now, with AI, fewer than ten are enough, and no shooting is needed at all. All processes happen on computers, making it more like a post-production company."

The core of this shift is not merely a reduction in crew size but a reshaping of the middle stage of film production workflows.

Director Chen Youxue describes the new process as a workflow situated between live-action filmmaking, animation, and post-production.

Directors can use text commands to manipulate 3D models in Blender, generating white-model animatics to confirm character positions, camera placement, and movement in advance. After actors complete real performances, their actions and expressions are transferred onto AI characters. Work previously handled by set design, lighting, and on-site direction is thus converted into digital asset production, shot breakdown, and model control.

Theoretically, by increasing headcount and computing power, a generation team can simultaneously produce large volumes of material and compress production timelines once character assets mature.

However, when actually making a film, Liu Chi still chooses to "polish it scene by scene." The team repeatedly checks for character drift, whether emotions flow coherently across scenes, and whether visuals serve the narrative, rather than generating a large batch of random shots and then searching for usable clips. Machines boost visual output but cannot replace aesthetic and narrative judgment.

This is why creators with backgrounds in art, animation, and concept design are entering AIGC filmmaking earlier.

Long Yitong, a student at the Central Academy of Fine Arts, began experimenting with AI imagery in 2023 and founded Yuemo Studio.

In 2025, the AI feature Tai Sui, an urban supernatural tale produced with Yuemo's participation, premiered on Tencent Video, becoming one of the early AI films to land on a major long-form video platform in China. Through this project, the team completed its first full run of the AI feature production pipeline for the first time.

Subsequently, they collaborated with a professional directing team on the 30-minute film Black Bird, handling all AI content production. The film later won the Best Visual Creativity Award at the Aranya Theater Festival.

In Long's view, once AI lowers the operational threshold, lighting, color, composition, and artistic style become even more important. The new workflow demands what she calls a "hybrid talent": someone who understands filmmaking, art, and models, a profile that does not map neatly onto any single role in a traditional crew.

Traditional companies are also beginning to create positions for this kind of hybrid work.

Zhang Zefan, a former film industry professional and film school instructor, told All-Weather Technology that leading production companies such as Daylight Entertainment and Yaoke Media have begun recruiting for "AI producer" roles.

In his view, this new role currently has no clear boundaries. But unlike traditional producers who manage budgets, accounts, people, and operations, the AI producer more closely resembles a pipeline producer in animation: breaking down creative intent into shots and digital assets, coordinating generation, art, and post-production, while managing personnel capacity, computing costs, and final delivery quality.

The new production system is already operating, and industry practices are quietly evolving.

Big-Picture Thinking, Small Stakes

Treating cost reduction as the sole value of AIGC would be reductive.

In discussions between multiple creators and All-Weather Technology, "flexibility" appeared more frequently than "cheapness."

Previously, a film typically progressed along a linear path of script development, assembling the core team, shooting, and post-production. Funding continued to flow after principal photography began, and investors often had to wait until substantial costs were incurred before judging from sample footage whether the characters, story, and visual style held up.

This uncertainty is also one of the reasons film investment has long relied on well-known IP, stars, and experienced directorial teams. These factors do not guarantee success, but they provide investors with a degree of recognizable credit before a finished film exists.

AIGC offers a different way to place bets.

Chen Youxue first has his team invest modest computing power and effort to build the framework of a feature, allowing the creative team and investors to see the characters, scenes, pacing, and story direction.

"Previously, you had to wait for a sample reel to see how the film would look. Now a feature can take shape within a month," Chen told All-Weather Technology. Directors can then judge which segments deserve more attention and selectively add live-action performances or raise production standards.

He estimates that roughly 30% of early generated material still makes it into the final cut. Teams can also bring demos to investors to discuss whether better resources should be allocated or whether to halt the project before more capital is committed.

For investors, the same upfront budget can be used to test more projects simultaneously, then filter based on rough cuts and platform feedback, leveraging more predictable, higher-quality projects. Chen believes that within the film industry's investment framework, this form of risk control matters more than endlessly compressing costs. Once a project demonstrates commercial potential, the next question is not how much more can be saved but what additional resources will amplify returns.

Earlier decision-making does not mean the production process can be endlessly streamlined. Certain stages that determine final quality still cannot be bypassed.

Liu Chi believes script development cannot be accelerated indefinitely by adding more generation engineers. "A good script can take two years, ten years, or two months to write, but it cannot be produced in two days."

Post-production also does not disappear with the set.

Liu Chi lists editing, color grading, dubbing, visual effects, music, and mixing as all still present: "All the traditional post-production stages remain." AI-generated footage also requires correcting character drift, shot consistency, and visual continuity, which sometimes adds corrective work.

AIGC's rapid turnaround nature also introduces new commercial risks.

Commercial-grade projects require repeated generation, filtering, and correction, and model credits are not a negligible expense. AIGC is indeed cheaper than large-scale live-action productions, but independent creators still face significant costs to complete a commercial work.

More problematic is technological depreciation. Chen Xing, founder of Beijing Tianjiao Xingchen Culture Media, told All-Weather Technology that AIGC projects ideally should not take longer than two months from initiation to completion.

If the timeline stretches, new models may render early materials obsolete, and production efficiency could suddenly increase fivefold or tenfold. "By the time the project actually launches, audiences may already feel your image quality is no longer impressive."

After Production Rights Decentralize: Decentralization and Hidden Barriers

The concentrated release of AI-driven project supply is approaching.

Chen Youxue told All-Weather Technology that several platforms he knows of are producing mid-length dramas, long-form series, and films, and many AI creators around him have entered projects, though their works have not yet launched. He predicts, "By the first quarter of next year, a batch of content should be out."

This aligns with signals from iQiyi, Youku, Tencent Video, and Mango TV.

In April, Sun Zhonghuai, Vice President of Tencent and Chairman of Tencent Online Video, revealed that Tencent is producing a series of over ten episodes and a 90-minute film using a fully AI-driven process.

At the end of August, Journey to the West: The Later Story entered Hunan Satellite TV's prime-time slot and Mango TV, with the first five episodes of the "Huaguo Mountain" arc airing initially. The remaining 25 episodes are planned for winter break, adopting a season-based model of simultaneous production, review, and broadcast.

Youku launched an AIGC web film incentive program in June, offering rewards based on effective viewership after release, with single-exclusive films eligible for up to 1.35 million RMB.

Wang Ze, a graduate of the Shanghai Vancouver Film School, directed Tide Erosion, which was selected for the "Peter Pau iQiyi Film Theater" this summer. This marks iQiyi's attempt this year to push from AI short films toward commercial features.

Compared with the earlier "Peter Pau iQiyi AI Theater," which primarily produced works over 15 minutes, the new round raises the threshold to over 60 minutes, ultimately selecting ten AIGC Chinese-language commercial films for full production.

He told All-Weather Technology that the project completed final selection at the end of July and is currently in script refinement, character design, and asset production, with an ideal target of completing production by year-end.

Feature-length development once measured in years is now producing new samples measured in months.

Once barriers drop, increased supply is almost inevitable.

In March, iQiyi CEO Gong Yu proposed that as AI reduces production costs and expands the creator base, video platforms would shift from centralized content production to a more decentralized creator ecosystem.

iQiyi subsequently established an AIGC Creative Center, expanding partnerships to individuals, studios, production teams, and university groups, while opening directions in theatrical films, web stories, long-form dramas, and mid-length series.

Project sourcing has thus expanded. Individuals and small teams no longer need to first become established production companies to bring scripts, demos, and production plans before platforms.

iQiyi also opens some of its IP, computing resources, and investment financing, offering different cooperation paths such as platform investment, copyright acquisition, and revenue sharing on finished films.

However, the more costs platforms bear, the deeper their involvement in script revisions, production schedules, and final delivery becomes.

For example, add-on investment sees iQiyi contribute 10%-30% of funding with review opinions serving more as references. Lead investment and full investment see the platform cover 50%-60% and 100% respectively, with creators expected to follow platform review feedback.

Long Yitong told All-Weather Technology that she is not particularly concerned about competitive pressure from rapidly increasing AIGC supply. In her view, teams with stable creative capabilities capable of consistently producing mid-to-long-form content remain scarce.

She describes mass-produced AI comic dramas as "AI Foxconn": text enters an assembly line, characters and visuals are generated according to fixed templates, aesthetic and narrative judgment is delegated to models as much as possible, and teams amplify output by adding computing power and operators.

Once this workflow is established, low-quality supply can indeed expand rapidly. But premium long-form content cannot easily follow this logic. "It's like after the camera was invented, everyone could press the shutter, but only a few become photographers."

This does not mean the film industry has fully opened its doors to AIGC practitioners. Long has already felt this hidden threshold.

She relays a judgment from a platform executive: "Most AI creators have not yet reached the professional standards of the film industry. Many have simply enjoyed a wave of technological dividends. Professional work should still be done by professionals."

In her view, independent AI studios without traditional film team backing have low success rates when directly pitching feature projects to platforms.

Therefore, Yuemo has not positioned itself as an AI team capable of independently producing features. Instead, it actively strengthens cooperation with professional directors and traditional film teams. "We respect the professional boundaries of traditional filmmaking, and each side plays to its strengths."

Reordering the Ranks

Technological revolution does not merely replace a set of production tools; it redistributes industry positions.

In 2002, major Hollywood studios jointly established the DCI Specification to unify file, projection, and security standards for digital cinema. A decade later, Kodak filed for bankruptcy protection, symbolizing the decline of the old technology chain.

Meanwhile, new industry positions opened up.

Companies like Avid, Autodesk, and DaVinci Resolve gradually embedded themselves into editing, visual effects, color grading, and digital asset management workflows. They did not become film studios, but they secured the standard-setting positions for "how films are produced."

This scenario is now replaying in the AI era.

In 2025, Kling became the exclusive technology partner for the AIGC film unit at the Beijing International Film Festival, while the Shanghai International Film Festival established an AIGC unit with Jimeng.

By 2026, Tamar AI, behind TAPNOW, chose the Venice Film Festival to launch Glanze, an AI-native streaming platform, extending capabilities centered on generation and workflows toward content distribution.

Theatrical films sit at the apex of the traditional film industry. Here, the challenges are not just larger screens and higher image quality standards but more stringent commercial scrutiny.

Audiences must actively pay ticket prices for a specific work, and cinemas continuously adjust screenings based on attendance rates.

Film festivals therefore function like an early seating test. Kling, Jimeng, and others are all seeking opportunities to create an "Avatar moment" for the AIGC era.

Unlike the previous generation of digital tools that primarily organized existing material, today's AI platforms directly participate in generating characters, scenes, and shots, blurring the boundaries between tools, producers, and distribution platforms.

But possessing only generation capability is insufficient to enter the core profit distribution of film. IP, investment, censorship, distribution, and theater networks still determine whether a work can truly become a theatrical commodity.

Internet platforms that hold IP, traffic, and capital are more likely to shift from serving the film industry to directly participating in it.

AIGC's rapid turnaround objectively lowers the cost for internet platforms to sit at the table.

Currently, Fanqie Novel has approximately 700,000 contracted authors and more than 1.8 million licensed works. This content has primarily been monetized through reading, audiobooks, short dramas, and comic dramas, with only a tiny fraction ever entering film development.

In June, Fanqie Novel launched a theatrical film solicitation program requiring projects to be adapted from the platform's own IP. Selected projects receive copyright and cash investment, with the platform encouraging a cooperation model of "base compensation plus box office or revenue sharing."

Liu Chi told All-Weather Technology that her company is co-producing the AI theatrical film Son of the Moon with Fanqie Novel, adapted from a Fanqie Novel IP.

According to current plans, the film will begin production in October 2026, complete principal work by year-end, and enter the censorship process next year.

This type of cooperation resembles targeted development. Fanqie enters the project as a co-producer, providing IP while retaining rights to participate in project quality and downstream revenue.

The reason may lie in the longer development cycle of theatrical films, combined with the typically exclusive nature of adaptation rights. Granting large numbers of licenses to teams lacking funding and distribution capabilities risks long-term "land grabbing" of IP. Even if projects are completed, they must still clear censorship, distribution, and screening, and a single failure could harm the original work and subsequent development.

Traditional film companies are likewise increasing their reserves. Bona's Sanxingdui: Future Memories has received its dragon mark approval, Strange Tales of China: Rahu's Sea Market has completed filing, and Off-Plan: Guanghan Operation and Dongfeng Must Reach are in production or preparation stages.

The first two are animation projects, indicating that AI is currently easier to integrate into workflows that rely on digital assets and post-production compositing.

But passing censorship does not equal commercial validation. One interviewee summarized this as the "premium paradox": low costs bring low barriers, and audiences will pay for good stories, but they are not necessarily willing to pay more for the AI label.

Therefore, companies offering only models and computing power may remain in a supplier position.

A more likely structure is: model companies provide production capabilities, internet platforms provide IP, capital, and traffic, and traditional film companies handle professional production and theatrical delivery.

Whoever controls IP and project leadership, whoever turns tools into industry standards, and whoever controls the last mile of reaching audiences will determine how the next round of profits is distributed.