AI technology is penetrating into all industries at an amazing speed, and the field of film and television production is no exception. In this wave of integration of technology and art, the rise of AI has brought unprecedented opportunities for change in the film and television industry. Especially in the field of short drama production, it has long faced many challenges such as high production cost, long production cycle and creative bottleneck. Joly Dimension, a company specializing in AI film and TV technology, is opening up a whole new world for short drama production with its self-developed "Saiwa" model. From the 3D AI film and TV production route to a series of key technologies such as AI actor generation, AI performance and AI action generation, Juridic Dimension's "Saiwa" model has deeply empowered every aspect of short drama production, making it more efficient, sophisticated and full of creativity.
Metaphysical Fantasy: From Playing Erhu: A First Look at Industrial-Scale Applications of AI Short Plays
On February 14, the red fruit platform went online with the first industrial-grade AI short drama, "Metaphysical Illusion: Starting from Playing Erhu". This short drama was jointly created by Poly Dimension, Impossible Technology and Zhongguancun Science Fiction Industry Innovation Center. This technical team used code as needle and thread to embroider a work full of technological aesthetics in the field of digital film and television, which is a bold and innovative attempt. The release of this drama not only marks the breakthrough progress of AI technology in the field of film and television production, but also signifies a solid step forward in the industrial-level application of AI technology.
At present, AI technology has been widely used in automated manufacturing, intelligent customer service, voice recognition, image generation and many other fields, almost ubiquitous. However, in the film and television industry, although there have been some experimental projects based on AI technology, there are very few cases that have really reached the industrial level of application. Most of the AI applications still remain at the level of entertainment experience on the web page, such as text generation, selfie beauty, virtual dress-up, etc. Although convenient and interesting, there is still a big gap from the actual production needs.
However, with the launch of "Saiwa", a multimodal 3D video model, AI film and TV production has finally ushered in a new era of industrial-grade applications. The "Saiwa" not only has powerful image generation capabilities, but also can efficiently handle the video content creation process, helping professional film and TV teams to produce content in a more efficient and higher quality way. Compared with traditional film and TV production modes, the application of AI technology makes the creation process more flexible and faster, significantly improving the team's work efficiency. The short drama "Metaphysical Fantasy: Starting from Playing Erhu" is a successful practice of AI technology to break through the bottleneck of traditional film and television production.
Adapted from a popular online novel, the drama takes "martial arts" elements as its theme and tells the story of a blind young man who, with his exquisite erhu skills, reverses his destiny and breaks into the mysterious world of the mysterious Jianghu. The plot is tight and fast-paced, full of adventure and fantasy.
In the episode, the strong martial arts atmosphere and the fantasy world are skillfully blended, injecting a unique charm into the story. AI technology shows great creative potential in the production process, from character modeling, scene design to action choreography, all of them can be efficiently generated in a very short period of time, and present a stunning visual effect. Especially in the presentation of special effects and action scenes, AI makes these scenes even more grandiose and shocking, bringing a refreshing movie-going experience to the audience.
"Saiwa" Big Models: Breaking the Boundaries of AI Creation
The core business model of PolyVision is to empower the traditional film and television industry based on AI technology, especially in animation production. The core of its technology system is the "Nuwa" model, which is integrated into "CyberDirector", a full-process 3D AI film and TV production platform independently developed by PolyVision. The "CyberDirector" platform is able to independently generate objects in the video in a 3D structured way, ensuring their high consistency, editability and controllability in different shots and motion states. The platform can also realize real-life "old actor" level AI movie and TV performances, and support content creation of any length. Through this innovative platform, PolyVision hopes to create a new era of AI film and TV production by giving everyone the opportunity to have their own AI crew and realize the goal of producing 10 minutes of high-quality film and TV content a day.
CyberDirector demo image
"One of the highlights of the CyberDirector platform is that the creator does not have to be physically involved in the performance. The "CyberDirector" model prioritizes the creator's core need - unique object generation, i.e., a 3D digital human model with bindings. The model supports multimodal input, including random generation, text description, image reference, and other methods, as well as fine-grained editing.
CyberDirector demo image
Through AI actor generation technology, Poly Dimension creates virtual actors who can perform and convey emotions. These AI actors are not only highly realistic in appearance, but can also show rich emotional changes according to the requirements of the script. With the iterative upgrading of technology, the expressive power of AI actors is expected to be comparable to that of real actors in the future. The introduction of AI actors effectively solves the problems of actor shortages or tight schedules that are common in short drama production. AI actors are extremely flexible and can be "on line" at any time to perform roles of different styles and emotions, and can even maintain a stable level of acting skill in complex plot design. What's more, AI actors are not restricted by physical conditions and can "shoot" in a variety of extreme environments, whether it's going up to the sky or diving in the deep sea, AI can easily handle it.
AI not only understands the emotions and character settings in the script, but also automatically generates actions that match the character's personality and optimizes the performance effect through algorithms. This makes it possible to realize high-quality presentation of fight scenes, emotional scenes and complex action designs in short dramas without live actors. Imagine, with the support of AI technology, the skit can easily present high-speed rotating battle scenes, between the light of swords and shadows, the character's posture changes, moving and jumping, all of these difficult actions are done by AI smoothly. Through accurate motion capture and detailed action design, AI technology not only makes the picture more visually appealing, but also significantly saves production costs and time during the creation of the skit.
Zhao Tianqi, founder of Poly Dimension, said that the current production cost of short plays is roughly between 500,000 and 1,000,000 yuan, which can produce about 100 minutes of content, which means that the cost per minute is about 5,000 yuan to 10,000 yuan. Among them, the cost of 10,000 yuan per minute of short drama is already a boutique, while 5,000 yuan per minute is the cost of the bottom line, below this standard, the quality of the content will be difficult to attract viewers.
Zhao Tianqi pointed out that if the lower limit of the cost of animation short plays cannot be controlled at 5,000 yuan per minute, but as high as 50,000 yuan per minute, then the whole industry will face the risk of shrinking. Even if the limit of cost compression, the production team to move to the city of very low labor costs, and completely sacrificed profits, the animation production cost is very difficult to be lower than 15,000 yuan per minute, which has already exceeded the upper limit of the cost of short plays. Moreover, even for an animation costing $15,000 per minute, its production quality is often not as good as that of a live-action short drama costing $5,000 per minute, and its performance is even worse in certain aspects.
Although the head of the short drama works can exceed millions or even tens of millions of revenue, but the industry average revenue share is only about 200,000 yuan, so the short drama industry is facing the challenge of high cost and high risk. In order to keep the cost down, most of the short dramas are rough, with fewer special effects, lack of animation elements, and a single form of overall content. There are mainly two existing movie and TV production methods: real shooting + small part of virtual shooting + post-production; and pure animation production. Under the background of the rapid development of AIGC technology, embracing AI has become a major trend in the film and television industry.
PolyVision has chosen the 3D route, more precisely the 3D structured generation route. Unlike 2D video models, PolyVision believes that the 3D route is more in line with the nature of movie and TV production. In terms of camera operation, lighting, consistency, etc., 3D technology can be effectively solved even without AI. In contrast, some 2D video big model companies claim that their products are able to realize camera running and provide several camera running methods for users to choose, but this requires targeted training to realize. The "3D" emphasized by Polyvision does not just mean three-dimensional visual effects, but more importantly, "structuring". Structuring is the key to achieving controllable, movable and editable content. The combination of 3D and structuring retains the advantages of 3D technology, such as the ease of mirror transport, while enabling independent control, generation, and combination of the various parts of the model. This approach breaks down a complex problem into several smaller problems, one by one, avoiding the dilemma of an uncontrollable overall model.
According to Zhao Tianqi, when discussing the future of AI film and television, the definition of "film and television" must first be clarified in order to evaluate the advantages and disadvantages of the 3D route and the high and low costs. In his opinion, film and television works should reach the production level that is generally accepted and loved by the public under the traditional professional production process. This is the definition of "film and television" from two dimensions: production process and final effect. From the technical point of view, film and television is to film and present the stories that happen in the 3D world. Therefore, it is important to clarify the definition of "film and television", which is often narrative in nature, rather than simply a record of what has been filmed.
The reason why PolyVision chose the 3D route is precisely because of its controllability, which is more in line with the nature of movie production. However, the disadvantage of the 3D route is that there are still some challenges in simulating the effect of live shooting, and it is also difficult to quickly generate rapidly changing visual effects. Therefore, the business model of Joly Dimension is based on AI visual capture technology, which empowers the traditional film and television field, especially animation production.
Zhao Tianqi also admits that for small startups, relying solely on industry data accumulation is no longer a core competency. Nowadays, it is becoming easier and easier to obtain huge amounts of free video, image and text data. He believes that the development trend of large models is "the strongest is always stronger", and every order of magnitude improvement in model parameters will form a suppressive advantage over the next order of magnitude. For example, a model that invests a billion dollars in size will overwhelm a model that invests 100 million, and a model that invests 10 billion will in turn outperform a model that invests a billion. Eventually, only a few head players may remain in the global market. Training a powerful big model requires a huge investment that can cost hundreds of billions, and there are only a handful of companies with market capitalizations in excess of hundreds of billions. If the cost of training a big model is so high and only a few winners can be accommodated in the end, the competitive landscape of the market will be very different from the past. In the past, companies could pursue opportunities in different segments, take different technology paths, focus on different application directions, adopt different business models, or offer differentiated services. But in the era of "big models," such opportunities are diminishing dramatically.
Therefore, from the perspective of data accumulation, the multimodal 3D video large model "Saiwa" may be a key opportunity for PolyVision to break through in the AI film and TV field. Because there is a relative lack of high-quality large-scale datasets in the 3D field, while PolyVision has already grasped a certain degree of data advantage after a long period of accumulation, which constitutes its important "moat" and is difficult for other competitors to reach.
Currently, the "CyberDirector" platform and "Cywa" big model technology have won a number of governmental honors, including the project of accelerating the development of high-growth enterprises in "Several Measures to Support the Development of Internet 3.0 Innovation in Chaoyang District", the project of supporting excellent vertical models in "Several Measures to Promote the Development of General Purpose Artificial Intelligence Innovation and Application in Chaoyang District", and other honors. Several Measures to Promote the Development of General Artificial Intelligence Innovation and Application in Chaoyang District" to support excellent vertical modeling projects.
PolyVision has already entered into short drama production cooperation with a number of head platforms, and the first batch of 20 AI short dramas will be released one after another, with plans to increase production capacity to 10 or even 1,000 dramas per month in the future. The company is actively forming a seed team of AI movie and TV production elites, and plans to cooperate with the government to build an AI short drama production base, train students to use the CyberDirector platform to produce AI short dramas, and cultivate a new generation of AI movie and TV creators. The entire production process will be strictly controlled by PolyVision. In the future, Juridic Dimension plans to share revenue with the government, universities and students through a contractual model to jointly promote the development of the AI short drama industry.
AI Short Drama Base: Double Wheel Drive of Commercialization and Industrialization
As a pioneer in the AI short drama industry, Joly Dimension's goal is not only limited to technical research and development, but also to build a professional AI short drama production base to comprehensively promote the commercialization and industrialization of short drama production. The rise of AI short dramas is not by chance, but based on Joly Dimension's deep insight into the industry's development trend and forward-looking layout. With the technical support of the "Saiwa" model, Juridic Dimension has successfully pushed AI short drama creation to a new level, and will further promote the large-scale development of the industry in the future. With the support of Beijing Municipal Science and Technology Commission and Zhongguancun Administrative Committee, Joly Dimension and Zhongguancun Science Fiction and Fantasy Industry Innovation Center are planning to build a demonstration center for the production of AI short dramas in Beijing. As an important platform for the development of science fiction industry in Beijing, the Zhongguancun Science Fiction Industry Innovation Center is committed to gathering all the elemental resources of the science fiction industry. The center focuses on incubating and accelerating the core technology of science fiction industry, plays the role of a bridge between the three major elements of key technology, original talents and scene construction of science fiction industry, enhances the independent innovation ability of science fiction industry, and builds a comprehensive industrial service chain of "research institute + incubator + gas pedal + industrial cluster".
In 2023, Zhongguancun Science Fiction Industry Innovation Center successively won a number of honorary qualifications, such as New Audiovisual Space, Beijing New Audiovisual Art Park, and Beijing-Tianjin-Hebei Audiovisual Industry Base. At the beginning of 2024, the digital live broadcasting base built by the Innovation Center in cooperation with Polyvision was officially unveiled in Beijing, laying a solid platform foundation for AI technology to empower the audiovisual field. The establishment of the AI Short Drama Production Demonstration Center is not only a deep fusion of technology and creativity, but also a key milestone for the AI short drama industry to mature. In the future, Gathering Dimension plans to join forces with universities, the government, and enterprises upstream and downstream of the industry chain, and extensively lay out AI short drama production bases across the country. Through in-depth research on relevant topics, we will continue to improve the AI short drama production system, expand the scale of the industry, and comprehensively promote the landing and popularization of AI short drama production nationwide.
Metaphysical Fantasy: Starting from Playing Erhu is not only a model of successful application of AI technology, but also a bold innovation for the entire film and television industry. It fully demonstrates how AI technology plays a central role in the creation of short dramas, and makes people look forward to the future development of AI short dramas. Looking ahead, with the continuous progress of AI technology and the acceleration of industrialization, more high-quality AI film and TV content will spring up. Every creator will be able to realize his or her creative dream with the help of AI as a powerful tool.
The rise of AI short dramas marks a brand new era of film and television creation. In this new era, AI is not only a tool to improve efficiency, but also a source to stimulate creativity, and a powerful driving force to push the boundaries of film and television creation to expand continuously. Looking into the future, we have reason to expect AI technology to be more widely used in the field of film and television, and AI will bring us more surprises from short dramas to movies, and from TV dramas to animations. The in-depth cooperation between Joly Dimension and Zhongguancun Science Fiction Industry Innovation Center is a sign that the AI short drama industry is about to flourish. With the maturity of the technology and the continuous expansion of the industry scale, AI short drama will no longer be the "pavilion in the air" of a few people, but will become a reality within the reach of every creator. Let's look forward to the near future, AI technology will bring us an unprecedented new movie and TV experience.
English 
简体中文 
日本語 
Deutsch (Sie) 
Português do Brasil