No Qing palace dramas in AI-generated short plays? AI: This humble concubine cannot possibly do that, Your Majesty!
In AI short dramas, some characters travel back to the Song Dynasty to work as litigation masters, some are reborn in the early Ming Dynasty to stir up the imperial court, some fall in love in Chang'an of the Tang Dynasty, yet the gate of the Forbidden City remains tightly shut. Looking at the hit rankings, there are barely any Qing Dynasty palace dramas.
From War and Beauty to Empresses in the Palace and Story of Yanxi Palace, a Manchu headdress swaying, a court bead necklace hung, a single line of "this palace" or "this dowager" could prop up an entire episode. The "palace-themed memories" that once dominated the screens have suddenly disappeared collectively. What seemed like a ready-made traffic secret formula has surprisingly failed to work in the field of AI short dramas.
Lao Wang, who has been producing AI short dramas for more than two years, puts it bluntly: "Qing Dynasty palace dramas were the viewership guarantee in the era of live-action dramas, but now they can't even get a sip of the soup in the AI short drama sector."
1. Manchu headdresses askew, imperial robes in disorder: Can AI not produce qualified Qing Dynasty palace dramas?
Lao Wang took a Qing Dynasty time-travel script last year, thinking that the palace-fighting tropes were ready-made, AI could generate images quickly, and he could make a profit safely by producing a whole drama with only a few thousand yuan. However, as soon as the footage was rendered, Lao Wang's face turned pale. AI generation is fast, but the more detailed the drama is, the more various and absurd the defects become. One day the imperial concubine's Manchu headdress is tilted, the next day the patterns on the emperor's dragon robe are mixed and mismatched, and the day after that, even the traditional Manchu platform shoes turn into sports shoes. Staring at the screen, he once suspected that AI had treated the Qing Dynasty palace drama as a hodgepodge of random ancient costume dramas. Every single item looked passable on its own, but when put on the same character, the whole look seemed like it came from a completely wrong set.
The Qing Dynasty themed content is an over-the-syllabus problem for AI. For ancient costume dramas set in the Tang, Song and Ming dynasties, AI can generate a general charm, and it is hard to find obvious flaws even if the clothes are slightly looser or tighter, or the hair buns are slightly higher or lower. Qing Dynasty palace dramas have been filmed by live-action film and television productions for decades, and generations of audiences are very clear about how to wear the Manchu headdress, how many strings of court beads to wear, whether the rank badge is embroidered with a crane or a golden pheasant, and what etiquette the maids should perform when meeting concubines. If AI makes even one mistake, the bullet comments will blow up immediately, and the comment section will turn into a public trial.
When AI makes mistakes, Lao Wang can only invite professionals to proofread. What seemed like a cost-saving tool ended up costing all the money on hiring proofreaders. For a drama that was supposed to cost a few thousand yuan, he ended up spending tens of thousands more. "The proofreading fee is higher than the production cost. This business is impossible to run."
What is even more fatal is the production model. Lao Wang did the math. A live-action film crew spends years polishing one drama, and can hire history consultants to tweak the details slowly. The production cycle of an AI short drama is calculated in days, while the proofreading cycle for a Qing Dynasty palace drama may take weeks. But the very foundation of AI short dramas is the assembly line model that emphasizes fast turnaround. The "asset-heavy" Qing Dynasty palace dramas that require frame-by-frame quality control have an extremely low input-output ratio by comparison. As for the hit dramas of the same period, modern overbearing president dramas use algorithms to pile up luxury cars, and fictional historical dramas can be written however the creators want. With the same computing power, feeding AI ancient paintings can produce artistic conception, but feeding it Qing Dynasty palace drama materials will most likely lead to errors. On the same track, others are enjoying fast food, while it needs special custom-made meals.
2. Audiences cast their votes of no with their thumbs
The technical accounts don't add up, and the audience's preference is the real judge of traffic.
The general narrative logic of Qing Dynasty palace dramas is hierarchy, suppression, and slow climbing up the social ladder under rigid rules. Zhen Huan spent years working her way from a low-rank attendant all the way to the empress dowager, and Ruyi went from deep love to cutting off her own hair; the sense of refreshment is delayed, and the pleasure is stifled. Audiences have to accompany the protagonist for dozens of episodes before they can get a breath of relief. The logic of short dramas is completely different: a twist every three seconds, a face-slapping scene every ten seconds, and the protagonist becomes almighty and defeats the villain in just thirty episodes. The narrative logics of Qing Dynasty palace dramas and short dramas seem to be inherently conflicting.
Office workers who have just finished work, squeezed the subway and collapsed on the sofa, why would they want to watch a group of people kneeling back and forth shouting "This slave deserves to die"? They are already tired enough from work during the day, who would like to watch people kowtow to apologize and suffer aggrievedly at night?
In fact, the audience has already cast their votes. The Micro Short Drama Creation Guidelines (Q2 2026) released by China Netcasting Services Association shows that the hit themes of AI short dramas are concentrated on tags such as "counterattack", "urban daily life" and "modern", with realistic narratives accounting for 60.5%. People who watch AI short dramas love the theme of "my fate is determined by myself not by the heavens", and few people would like to review the master-slave hierarchical relationship during their rest time. All Lao Wang's peers are saying that the completion rate of Qing Dynasty themed dramas has always been at the bottom, users just swipe past as soon as they see them, and the algorithm simply cannot push these dramas.
3. Tang, Song and Ming dynasties are AI's comfort zone, while the Qing Dynasty is AI's examination room
In AI short dramas, the Qing Dynasty has stepped back from the stage, while the Tang, Song and Ming dynasties have been pushed to the center position, which is no accident.
The Tang Dynasty has graceful and grand charm, the Song Dynasty has street market vibes and literati's refined tastes, and the Ming Dynasty is favored by young people for its complete Hanfu system and strong visual aesthetics. The Longest Day in Chang'an made Tang-style culture popular, The Story of Ming Lan made Song-style charm popular, and Brotherhood of Blades made Ming-style Hanfu popular. AI can follow the existing templates to produce content, so there is always a reference to follow. Ancient paintings, cultural relics, and Hanfu pattern drawings can be easily fed to AI, which naturally reduces the probability of making mistakes. Fictional ancient romance dramas have no historical facts to restrict them, so they also have higher creative freedom. In contrast, the Qing Dynasty carries heavy historical baggage, creators have to be careful everywhere when writing plots, and if they want to avoid all the minefields in palace dramas, they might get lost themselves first.
Whether a theme can be produced also depends on the market. The market is expanding, and the competition on the track is extremely fierce. The 2026 H1 AI Drama and Comic Drama Data Report released by DataEye shows that in the first 5 months of this year, the domestic market size of AI drama and comic drama has exceeded 220 billion yuan, and the full-year figure is expected to hit 400 billion yuan. The bigger the market, the more competitors there are. On average, only one hit can stand out from thousands of new works, and the hit rate is less than 0.1%. In this track where the winner takes all and the losers cannot even keep their basic capital, choosing the wrong theme means losing half the battle right from the start. Lao Wang said, "It's already very hard just to survive, no one has extra money to pay for their personal feelings."
When all AI short dramas are collectively avoiding the Forbidden City, news comes from the live-action short drama sector. TVB has partnered with a short drama platform to turn the classic IP War and Beauty into a vertical-screen short drama spinoff War and Beauty: Willow Folding by the Palace Wall, and Moses Chan will play Kong Wu again after many years. For the same theme, live-action crews dare to shoot it, while AI producers are discouraged, because the two sides calculate completely different accounts. Live-action short dramas have IP feelings and star appeal, and can sell at a premium by leveraging the nostalgia of remaking old classics. AI short dramas make money from fast-turnaround traffic, and cannot afford to lose money for three months, nor can they afford the cost of a full set of Qing Dynasty court costumes.
The Forbidden City has become quiet in AI short dramas, just because its splendor is too complex, too heavy, and requires too much careful treatment. When the traffic is pouring into the litigation masters of the Song Dynasty and the chief ministers of the Ming Dynasty, the emperors of the Qing Dynasty are sitting on the blurry dragon thrones generated by AI, with askew Manchu headdresses and messy dragon robes, waiting for a drama that can portray them correctly, which will never come. Lao Wang finally left a sentence: "Some dramas are better left to be performed by real actors."
This article is from the WeChat Official Account "Financial Gossip Girl Channel" (ID: baguanvpindao), written by Zhizhi, and authorized for release by 36Kr.