A six-person team of career changers has pushed an AI-generated historical drama through ten serialized seasons and a reported four billion views across platforms. [1][2] The drama, "Chi Xie Ru Song" (An Armed Squad in the Song Dynasty), runs on Hongguo, ByteDance's free short-drama app, and stands as the clearest public test yet of what video-generation toolchains can carry for a small amateur team. The result supports a narrow claim. The tools have crossed the production line, because keeping characters, battle scenes and period sets consistent season after season is now demonstrably within reach of people who taught themselves the workflow starting in 2025. [1] What the hit does not settle is the part that made it stick, and that part ran on a 1.7-million-character source novel, historical research and deliberate character operation rather than on model quality alone. [2][3]
Dingduan News reported cumulative views above four billion across platforms when the eleventh season opened for pre-orders on November 3. [2] Viral posts crowned the drama No. 1 on a board they called the "AI recommendation list" and carried the story that its makers were laid-off fund analysts. [1] The ranking claim cannot be checked against any board the platform publishes. On the AI-drama hot chart Hongguo does run, the first season sat at No. 8 on November 5 with 55.61 million heat, an 8.7 rating, 3.86 million favorites and 4.89 million likes, heat being the platform's own popularity metric. [4] The chart sits inside a dedicated AI-drama category the app runs beside its live-action and comics-drama boards. [4]
The six people behind it never worked in film
Hongxing News interviewed the chief planner, Zhang Hansheng, on November 3 and published his account on November 4, and it redraws the picture the rumors painted. The team holds more than thirty members across production cells, none of them former fund analysts, a claim he called complete nonsense. [1] The cell that built this drama has six members, two women and four men, a former rideshare driver, a delivery-van driver, a wholesale trader, a mother who juggles school pickups with meetings, and two ordinary office employees. [1] They live across Guangdong, Anhui, Hubei, Jiangxi and Fujian, while Zhang lives in Shijiazhuang in Hebei, and the whole group coordinates online. [1]
Zhang himself is in his fifties, has never worked in the film industry, and had cycled through failed ventures before the idea arrived. [1] In 2025 he noticed generative video and described the moment in plain terms, saying he suddenly found that one sentence could produce an image and one sentence could turn that image into a video, after which he spent more than ten hours a day teaching himself. [1] He took client orders for AI short dramas that year and moved to independent production in 2026. [1] The six members began as part-timers and, by his account, mastered a full workflow of image generation, image-to-video and editing. [1] He said the tools handed the group a new employment opportunity. [1] The raw material is a web novel by Cha Gen'er that had reached 1.737 million characters by early November and is still updating, with a note on its Fanqie Novel page saying it has been adapted as an animated comic on Hongguo. [3] A plausible reason the series reaches a tenth season without running dry is that inventory.
The output now survives a ten-season continuity test
Ten seasons of recognizable characters say more about the toolchain than any single frame could, and the reporting records the specific capabilities behind that continuity. Zhang said the two women in the cell were not yet skilled at war-scale sequences at first, and that after steady practice they now complete such shots. [1] The group researches historical sources for rank-specific costume colors and grades of dress, and built the character Li Qingtang from well-known portraits of the Song-dynasty poet Li Qingzhao. [1]
In comparisons published in December 2024, nearly two years before this drama, the strongest domestic models still showed limb-motion glitches and physics errors. [5] Viewer comments relayed by Hongxing News make the same comparison from the audience side, describing expressions that no longer look stiff the way earlier AI dramas did, costumes and architecture that read as Song-era, and a plot that holds up. [1] Through 2025, listed companies advertised full-chain pipelines and open-sourced video models such as SkyReels, though those remain company claims rather than independent benchmarks. [6] No third-party benchmark of this drama exists either, so the capability verdict here rests on the team's account and on audience response as relayed in reporting, and the two point in the same direction. [1]
The gates that stayed human
The same week this drama trended, criticism trended with it, and the criticism points at three gates the tools did not move. As the show rose, the hashtag "short dramas make you dumber" climbed the hot-search list, and Dingduan News compressed the issue into one line, saying a high AI share does not equal content quality, and that drawing a scene convincingly still depends on the creator's historical literacy. [2] Volume is the first gate. Dingduan News reports hundreds of thousands of AI dramas launching, most of them vanishing almost as fast as they appear. [2]
Being remembered is the second gate, and the production spends human effort on it. Public materials cited by Dingduan News identify the studio behind the show as Jin Ye Theater, and an industry source quoted in the same report says the team came into short drama from another industry and does not share a track with traditional short-drama producers. [2] The team operates its lead character like a person, crediting the cast list as "Lin Zhao as Lin Zhao" and running a character account that shares its inner thoughts in comment sections like a real person. [2] Dingduan News put the line that fits this gate best in its closing stretch, that AI lowers the cost of making something but not the threshold of being remembered. [2]
Rights are the third gate, and platform enforcement around them is active. According to DoNews, Hongguo's April 7 announcement pulled 670 AI-drama works in the first quarter for misusing cartoon, AI-brand, game-character and actor likenesses, with strengthened review and accountability. [7] Xinhua's May investigation documented "face-stealing" disputes over face data, a March case at the Beijing Internet Court and an actors' committee statement on unauthorized likeness use, while the research firm DataEye counted more than 100,000 AI drama and comics-drama launches in the quarter, 55 percent of them AI-simulated-human dramas. [8] No report connects this drama to any of those violations, and the platform keeps the show listed and charted. [4]
The budget change reaches live-action drama too
The same tooling is repricing shots inside conventional productions, which is why this story matters beyond the AI-drama category. The fifth installment of the live-action family saga "The Eighteen-Year-Old Great-Grandma Arrives" carried a reported twenty billion cumulative views across its first four installments and 15.6 million pre-orders before launch. [2] Its production hands some hard-to-shoot environment effects to AI assistance and concentrates budget and crew attention on characters and plot, so that helicopters and massed armies, ultra-high-cost shots for live action, become affordable through generation. [2] Generation carries the six-person drama's visuals from image through video to edit, and inside the conventional production it is one cost line among many. [1][2]
The platform context explains the reach of both hits. QuestMobile counted 718 million monthly users for short drama as of February 2026, with Hongguo at 304 million users, up 84.6 percent year over year, and an average of 125 minutes of daily use per viewer, far ahead of the second-place app's 39 million. [9] Hongguo's comics-drama app reached 24 million users within four months of launch, with about half of its new users funneled internally from the short-drama app. [9] Douyin vice president Li Liang said in late October, according to DoNews, that content quality is the core of short drama and that he is optimistic about AI-enabled creation. [7]
For the creators and tool watchers deciding what this headline means, the verdict stays narrow. Ten seasons and four billion reported views show the toolchain now carries serialized production for a team of six. Twenty billion reported views on a live-action saga show the same tools entering conventional budgets as the affordable answer to unshootable shots. What still has to be paid in human effort is consistency upkeep, rights exposure and the odds of being remembered in a feed where most AI dramas sink fast, and on current evidence those three, not image quality, decide which dramas break out. [1][2]