As the short-form drama ("short drama") market led by China spreads to Korea and elsewhere, "AI short dramas" that drastically cut production expense and timelines are rapidly emerging in China. Analysts say a new creative method is taking shape as video AI technology is combined with the growth base of the existing short drama industry. However, some note that to produce results with the same quality as live-action short dramas, more expense and time ultimately have to be invested, which could limit the economic effect.
According to China's 21st Century Business Herald and Caixin on the 18th, China's short drama market this year is estimated to reach 100 billion yuan (about 21 trillion won). As video AI technology advances quickly, AI short dramas produced without actual filming are expected to become mainstream within the year.
According to the reports, production expense for AI short dramas is about 10% of that of conventional short dramas. The production period, which used to take 15 to 30 days, has also been sharply reduced to within five days. As production expense and timelines have been slashed, the barrier to market entry has fallen significantly. As a result, corporations equipped with technology to produce comic-style dramas and virtual human dramas using AI are said to have seen a surge in orders and have already reached saturation.
This trend is the result of structural changes in the content industry intersecting with advances in AI technology. After comic-style dramas created with AI first appeared in China last year, it took only six months for virtual human dramas—in which AI creates the actors, backgrounds, and narrative—to become reality. Recently, ByteDance Ltd.'s video generation model "Seedance 2.0," which has been a major topic in the video content industry, showcased a function that automatically generates video composition and editing within minutes when photos and a rough story are entered. It took less than six months from OpenAI's Sora 2 announcement to the release of Seedance 2.0, and major corporations such as Google and Kuaishou (快手) are preparing follow-up models.
A drama director told Caixin that "Seedance is replacing some directors' roles," adding, "The sense of scene connection and editing is built into the model, which has greatly reduced the previously repetitive postproduction process." This person said some short-form content produced with Seedance has already reached a level that can be used immediately without additional editing. Also, as AI supplements special effects that were difficult to implement at low cost, AI short dramas currently have a high share of genres with high-level visual effects such as fantasy and science fiction (SF).
However, there is also an opinion that to achieve quality sufficient for commercial success with AI, "hidden expense" must be added, reducing the price advantage. With only low-cost AI software subscription fees and computing expense, one can produce only the framework of a short drama, and to raise the quality of the creation to a level with commercial value, more personnel, longer production periods, and repeated revisions and adjustments—hidden expense—are required.
21st Century Business Herald said, "Producing a hit with AI requires very high-level research and development investment," adding, "When evaluating the cost competitiveness of AI short dramas, we should not look only at the expense that is visible on the surface, but should calculate expense on the premise of achieving content with the same quality as live-action short dramas." A founder of a comic drama production company also told Caixin, "In the end, only those with content competitiveness will survive."
Institutional risk also remains. In addition to copyright issues in the AI model training process, there are portrait rights concerns that generated virtual characters could resemble real people. In fact, Seedance 2.0's global release has been put on hold due to copyright disputes. In response, the industry is devising risk-mitigation measures such as avoiding similarity with famous figures and tracking and managing facial data.