The AI blogger community is now crowded with people from the Central Academy of Drama and Beijing Film Academy.
AI creators on short video platforms are facing increasingly fierce competition. One group is accelerating the frequency of testing new models and new tools, while the other group that has stepped into the AI creation field is also moving closer to the professional standards of film and television creation.
In this round of competition, professionals with formal film and television education backgrounds have a particularly strong presence.
The gold award of the Douyin AI Creation Competition, which wrapped up in September, was won by *Closure* from the team "2XLabs", whose members have more than ten years of experience in the film and television industry. The author of the silver award work *There's No Time to Live Anyway*, "Shadow That Stands Up", has previously worked as a director, cinematographer, screenwriter and lighting technician, and has also taught photography and film and television courses in colleges.
Screenshot from Douyin
The creator "DiDi_OK" on Bilibili is no exception.
He graduated with an animation major and previously worked in advertising and visual roles at WPP, the London-based advertising giant. This year, he won the first prize in the open track of Bilibili's first AI Creation Competition with his work *Sign*, and launched his new work *Candy* in September. In the same month, his other work *Website* won the "Reply AI Studios Grand Prize" and the "WeShort Audience Choice Award" at the Reply AI Film Festival held in Venice.
For AI creators, model updates are certainly important, but how far a single work can go increasingly depends on how much creative skill the person using the tools actually has. It can even be said that the AI creation circle is quickly turning into a "gathering of film and television professionals".
01
The AI circle is starting to become a "Beijing Film Academy gathering"
Nowadays, the atmosphere in the AI circle is very different from what it was a few months ago.
In the first half of this year, when works such as *All Things Grow* and *The Laid-off Girl* became hits one after another, the most exciting keyword in the industry was still "lowered threshold". One person, one computer, and several AI tools can complete the work that used to require the joint efforts of screenwriters, directors, cinematographers, art designers, actors, and post-production staff.
This vision of "one person being an entire film crew" did constitute the earliest appeal of AI videos, and *The Laid-off Girl* is a typical example of this. Director Feifeifei worked as a model for 7 years before transitioning to work as a screenwriter and director for short videos and drama accounts. She incorporated her own modeling experience into the story of Fang Taozi, the protagonist of *The Laid-off Girl*, and used a small team of less than 5 people to produce an AI drama series with hundreds of millions of views.
Still from *The Laid-off Girl*
But if we look further ahead, we will find that the competition for AI videos is entering the next stage — when everyone can shoot videos, why can some people shoot better works?
A very straightforward answer is that more and more people who have actually worked in the film and television industry are starting to participate in this field.
In Bilibili's first AI Creation Competition this year, director and screenwriter Yi Xiaoxin (Jiaoshou) teamed up with the AI creation team "Fenmo" to produce the work *The Death of Nüwa*, and finally won the third prize in the open track. Instead of blindly pursuing the high-definition and realism that AI is best at, this film chose a clay stop-motion style with a rough handcrafted feel: Nüwa creates humans by kneading clay, and the images themselves look like they are gradually shaped from clay. To serve this setting, the team advanced the whole process from script, visual style to storyboarding, generation, and editing according to the logic of a complete short film, rather than first generating a batch of beautiful images and then trying to piece them together.
Screenshot from *The Death of Nüwa*
Another group of younger creators also place the script content in a more important position than AI-generated images.
For example, "Wudi Xingcheng" from Communication University of China participated in an AI animation hackathon in May this year, and needed to complete an animation short film *Dream Catcher* of more than 200 seconds within 36 hours. But in the first few hours after getting the topic, they did not rush to "generate images", but kept debating the script direction, core of the story and character relationships, and did not officially finalize the plan until the afternoon. After that, they entered the stages of character and scene asset generation and editing. In the end, *Dream Catcher* was shortlisted for the AIGC Excellence Promotion Unit, a supporting event of the 38th Hundred Flowers Film Awards.
But in fact, this kind of professionalization does not only mean that there are more and more formally trained directors, but also means that the organization mode of AI creators themselves is getting closer and closer to a compressed version of a film crew.
Some people are better at commercial advertising and post-production, while others prefer realistic themes and take on the role of director. When encountering problems that AI cannot solve for the time being, they will not insist on "everything must be generated by AI", but directly turn back to traditional tools such as PS. This is already a little subtly different from the early narrative that "one person with one computer can make a movie".
It can be said that in the past, the most prominent label of an AI creator was "whether they know how to use AI", but now this is rapidly being reduced to a basic threshold. What really sets people apart is starting to become another set of older standards — whether the story is logical, whether the characters are memorable, why the camera cuts in that way, whether the visual style can serve the content, and whether the audience is willing to keep watching after ten minutes.
02
The stronger AI gets, the more valuable "professional expertise" becomes
In May this year, Yiyu Guancha (ID: yiyuguancha) sorted out a group of AI creators in the article *Are the Most Popular AI Creators on Douyin, Kuaishou, Xiaohongshu and Bilibili Making Money?*.
At that time, the concept of AI creator was still very broad. Some people do tool testing, some make AI short dramas, some operate virtual humans, and others make money by selling courses, self-developed products, and platform incentives. At that time, the industry had already begun to move from technical dividends to "in-depth content cultivation", but a large number of creators could still gain attention by relying on new tools, new gameplay, and new art styles.
Just a few months later, this logic has moved a step further. Behind this is the fact that AI tools are rapidly becoming more accessible to all.
When a model is updated today, a large number of tutorials will appear tomorrow. A new camera movement, transition or character consistency gameplay may become a standard action that all creators can replicate in a few days. The "technical barriers" that used to take a long time to research are increasingly easy to be eliminated in the next round of model upgrades.
This is why, it is increasingly difficult to create hit works by simply relying on the "AI aesthetic". The award-winning works of the Douyin AI Creation Competition have verified this to some extent.
The director of the silver award work *Non-Demon* in the competition, "Bu Chi Zhacai", used to be engaged in photography and editing. To determine the ancient Chinese supernatural temperament of the work, he specifically studied the old versions of *Strange Stories from a Chinese Studio* and *The Legend of the Eight Immortals*, and even looked at the craftsmanship and carvings of ancient furniture. When the storyboards generated by AI failed to meet the requirements, he split the shots by himself, determined the shot sizes, and deliberately slowed down the camera movement to keep the audience in the emotional context of the story.
This set of aesthetic standards must also be supported by expressive themes. *Non-Demon* uses supernatural stories to talk about realistic issues such as online violence and human trafficking, and the characters' situations are the entry point for the audience's empathy. Everyone can generate an ancient mansion, but why let the characters meet here and how long the camera should stay are still decisions that must be made by the creator.
The cases on Xiaohongshu also explain this professional ability more specifically.
Yiyu Guancha (ID: yiyuguancha) introduced the director Shan Yin in the article *New Generation Directors Are Growing Up from Xiaohongshu*. He also won the Xiaohongshu Short Film Rising Star Award. He incorporated his anxiety from switching between multiple identities into the AI short film *A Ming's Day*, in which people can obtain different personalities from an automatic vending machine for personalities, but A Ming is always an ordinary person who is exhausted from coping with external demands. The absurd setting corresponds to the familiar emotions in real life, and AI gives this idea a visible form.
Shan Yin mentioned in the interview that when expressing sadness, you can't just give AI an emotion label, you have to break it down into specific expressions and movements. Effective prompt writing often comes from the creator's prior understanding of how to present emotions. This basic skill of screenwriting, which was originally used to communicate with directors and actors, still works even when using different tools.
All these make the meaning of "professional" clear. It includes not only narrative and audio-visual expression, but also the continuous development of characters. Formal training and industry experience can provide accumulation, and non-professional creators can also learn through practice. The final judgment standard falls on the works. The more links the model completes for you, the more you need to know which results are worth retaining and which must be overturned and recreated.
Looking back at the past six months, from *All Things Grow* and *The Laid-off Girl* to *Closure* and *Candy*, AI videos can hardly be simply classified as a kind of "technical experiment".
It has truly begun to enter the evaluation system of the content industry. The audience will not automatically watch five more minutes just because a work uses a more advanced model. The platform will not give a character more vitality just because the prompt is written more complex. A series of videos will not grow into an IP just because it has enough episodes.
AI once promised that "everyone can make movies". Now, this promise has been half fulfilled to some extent. Cameras, special effects studios, and huge film crews are no longer insurmountable thresholds.
When everyone has a camera, who actually knows how to make a movie? Just like after smartphones all support the camera function, the people who can take amazing photos are often still those photography bloggers who have a solid foundation in photography skills.
This article is from the WeChat official account "Yiyu Guancha" (ID: yiyuguancha), written by Lu Ke, and authorized for release by 36Kr.