AI-Assisted Short-Drama Team Reports More Than 20 Monthly Releases and 1 Million Yuan in Revenue
According to a report citing Dahe Daily, a digital media art student leads a team of more than 20 people using AI tools to make animated short dramas, with more than 20 works produced monthly and three to four episodes released each week. The report says the startup generated more than one million yuan in revenue within five months. It also says the founder trained 14 specialized models for tasks such as script breakdown and storyboard generation. According to the report, members review AI-generated footage, correct visual problems and direct narrative, aesthetic and stylistic decisions.
The report describes a human-in-the-loop production model in which specialized models and tools such as Kling and Jimeng handle repetitive production tasks while team members review outputs, refine visuals and make narrative and aesthetic decisions.
据大河报报道,一名数字媒体艺术专业的大学生带着二十多人的团队,靠 AI 工具杀入漫剧短剧赛道,创业五个月营收突破百万元,如今每月稳定产出二十余部作品、每周更新 3 到 4 部。耐人寻味的是,他最初还担心 AI 普及会冲击自己这个视觉创作方向——如今角色已彻底翻转:从亲手做素材的执行者,变成了项目决策者与质量把关人。
14 个自训模型流水线作业 效率的跃升写满细节。过去拆解一部完整剧本要花一周,现在大模型一天搞定;他自主训练了 14 个专用模型,分别适配剧本拆解、分镜生成等不同环节,按需自动调用。日常工作中,可灵、即梦等生成工具持续运转、任务排队处理,随身的笔记本上密密麻麻记着提示词、镜头编号和待返工的画面,攒下一整套人机配合的经验。
机器管产量,人管灵魂 这支团队并不迷信全自动。AI 吐出素材后,成员要逐一审核,修正人物表情僵硬、场景不符合时代设定等问题,反复迭代画面;通读剧本、梳理故事冲突、把控整体风格后,再把活儿分派给素材、修图、剪辑等岗位。在他们看来,AI 能承接大量重复的基础制作,却理解不了故事里细腻的情绪,也抓不准核心冲突——画面氛围感、人物情感逻辑和作品内核,仍得靠人的审美判断与叙事构思。这门五个月百万的生意,本质是把机械劳动外包给机器,把创意留给人。