AI simulator short drama speeds up: 3 to 4 people complete 60 episodes in a week

 AI human simulation film and television is moving from scattered experiments to stable production.In the short drama team of a company in Shanghai, the generator sends real-life images, scenes and text instructions to the creative platform. First, video clips of 5 seconds to 15 seconds are obtained, and then edited into 40 to 60 episodes of the short drama. The finished film will be launched on overseas paid platforms.A more mature team usually consists of 3 to 4 people and can complete a short play in a week, with a computing power cost of about 10,000 to 15,000 yuan.Changes in speed and cost are re-dividing the work of short play production.

The first thing to be solved is the problem that “ looks like a human, but something is always wrong.Characters generated in the early stage often have excessively smooth skin, drifting facial features, deformed fingers, and stiff movements. The audience can easily detect traces of the machine.The new model improves cross-frame stability and can also generate pores, stains, and fine wrinkles.The producers will also use different tools according to the subject matter, and then use imaging software to adjust skin color and facial details.A mole or a few freckles on a character's face can sometimes add more realism than a standardized, refined face.

The quality of the input material determines the generated results.The team will first make a big head portrait and full body portrait of the character, using a light gray background to reduce interference, and then refine the distance between the eyebrows, the fold of the face, the material of the clothing, and the direction of light.If the prompt words are too broad, the character characteristics may easily drift; if the instructions are too full, the model may focus on one thing and not the other.The most important task must be determined first for each shot: to see the change in eyes this time, to complete running or falling next time, and to break complex requirements into small executable steps.

After the face becomes realistic, acting like a human becomes a more difficult level.At this stage, AI characters are good at clear expressions such as crying, laughing, and angry, but it is easy to turn emotions into fixed actions.When a character becomes anxious, he clenches his fists and adds an exaggerated sound effect to his voice. This type of ”rubber fist“ has become a familiar routine for the audience.Truly believable performances also include hesitations, pauses, aversions, and subtle reactions in relationships that require characters to maintain coherent motivations in specific situations.

The creator therefore has to make more directorial judgments.The chief director will think about the movement, light and shadow, and lens rhythm in advance. If necessary, he will perform it himself and let the generation engineer break down the actions into clear instructions.Important shots often need to be generated repeatedly, and effective images must be selected from two or three times. Emotionally complex passages may be tried ten times.Real-person reference videos can also provide action timing and facial parameters, which can then be transferred to virtual characters.Tools have increased the speed of film production, but screening, selecting and organizing materials still takes up a lot of work.

This change continues to extend to the upstream of the film and television industry.In the past, digital technology mainly changed photography, post-production, distribution and projection. Now, generative tools have entered character design, space construction, storyboarding, performance control and sound production.Small teams have the opportunity to realize world views and action scenes that they could not afford in the past, while large teams will recombine models, automated processes, and traditional positions.Screenwriters, directors, cinematographers, artists, and editors will not suddenly disappear; the boundaries between positions will be rearranged.

When most teams can generate a clear picture, pure ” will quickly lose scarcity using AI %.Why the character makes a choice, why the camera stops here, and whether the emotions before and after can be captured still test the drama, aesthetics, and life experience.The next step for the AI ​​simulator short drama depends on whether it can break out of template expressions and make the audience willing to follow the character to complete a story.Technology has lowered the production threshold, and content competition has just entered a truly difficult stage.