{
 "sources": [
  {
   "org": "a16z",
   "title": "State of Generative Media（基于 fal 的推理数据）",
   "author": "Jennifer Li、Justine Moore",
   "date": "2026-02",
   "url": "https://a16z.com/",
   "used_for": "框 V：模型、推理与编排、应用三层，按行业用途（广告 56%、娱乐 43%）"
  },
  {
   "org": "a16z",
   "title": "There is no God Tier video model",
   "author": "Justine Moore",
   "date": "2025-10",
   "url": "https://a16z.com/",
   "used_for": "框 V 模型层：视频模型按用途分化，没有一家通吃"
  },
  {
   "org": "招商证券",
   "title": "AI 影视行业深度报告：AI 影视 AI 应用的扛旗者",
   "author": "顾佳、谢笑妍、刘晓珊",
   "date": "2026-11-01",
   "url": "http://stockfinance.sina.cn/stock/go.php/paper/reportid/846849258986/index.phtml",
   "used_for": "框 D：AI 视频渗透剧本分镜、画面生成、后期特效全环节，短剧与漫剧率先跑通商业化"
  },
  {
   "org": "本图新增",
   "title": "原骨架没有的层",
   "author": "",
   "date": "",
   "url": null,
   "used_for": "世界模型与实时视频：a16z 2026 年预测里的「可以走进去的视频」"
  }
 ],
 "about": "骨架：两个大框。框 V（AI 视频生成）照 a16z × fal《State of Generative Media》；框 D（AI 短剧与漫剧工具）只取招商证券《AI 影视行业深度报告》的生产工具环节。只收工具，短剧 App 和内容出品方不收。",
 "layers": [
  {
   "code": "v1",
   "name": "视频基础模型（大公司）",
   "ai_era": "拍摄加后期 → 图生视频、文生视频，大厂各出一条模型线",
   "box": "AI 视频生成 · 模型",
   "box_zh": "一句话、一张图 → 带声音的视频",
   "source": {
    "org": "a16z",
    "title": "There is no God Tier video model",
    "author": "Justine Moore",
    "date": "2025-10",
    "url": "https://a16z.com/"
   }
  },
  {
   "code": "v2",
   "name": "视频基础模型（独立公司）",
   "ai_era": "同上；独立公司靠单项能力（运镜、角色一致、实拍改造）和迭代速度活下来",
   "box": "AI 视频生成 · 模型",
   "box_zh": "一句话、一张图 → 带声音的视频",
   "source": {
    "org": "a16z",
    "title": "There is no God Tier video model",
    "author": "Justine Moore",
    "date": "2025-10",
    "url": "https://a16z.com/"
   }
  },
  {
   "code": "v7",
   "name": "世界模型与实时视频",
   "ai_era": "预先渲染的视频 → 可以走进去、实时响应的生成画面",
   "box": "AI 视频生成 · 模型",
   "box_zh": "一句话、一张图 → 带声音的视频",
   "source": {
    "org": "本图新增",
    "title": "原骨架没有的层",
    "author": "",
    "date": "",
    "url": null
   }
  },
  {
   "code": "v3",
   "name": "推理与 API 服务",
   "ai_era": "每家自己部署 GPU → 一个 API 调几百个模型，按秒计费",
   "box": "AI 视频生成 · 接入",
   "box_zh": "开发者怎么用上这些模型",
   "source": {
    "org": "a16z",
    "title": "State of Generative Media（基于 fal 的推理数据）",
    "author": "Jennifer Li、Justine Moore",
    "date": "2026-02",
    "url": "https://a16z.com/"
   }
  },
  {
   "code": "v4",
   "name": "聚合创作工具 / 视频 Agent",
   "ai_era": "一个模型一个网站 → 一个工作台接所有模型，做模板、预设、无限画布",
   "box": "AI 视频生成 · 创作工具",
   "box_zh": "做视频的人每天打开的东西",
   "source": {
    "org": "a16z",
    "title": "State of Generative Media（基于 fal 的推理数据）",
    "author": "Jennifer Li、Justine Moore",
    "date": "2026-02",
    "url": "https://a16z.com/"
   }
  },
  {
   "code": "v5",
   "name": "AI 剪辑与后期",
   "ai_era": "剪辑师逐帧剪 → 自动剪、字幕、补镜头、换脸换景、超分",
   "box": "AI 视频生成 · 创作工具",
   "box_zh": "做视频的人每天打开的东西",
   "source": {
    "org": "a16z",
    "title": "State of Generative Media（基于 fal 的推理数据）",
    "author": "Jennifer Li、Justine Moore",
    "date": "2026-02",
    "url": "https://a16z.com/"
   }
  },
  {
   "code": "v6",
   "name": "营销与广告视频",
   "ai_era": "广告公司 2–4 周出片 → 几百个变体当天 A/B 测试",
   "box": "AI 视频生成 · 创作工具",
   "box_zh": "做视频的人每天打开的东西",
   "source": {
    "org": "a16z",
    "title": "State of Generative Media（基于 fal 的推理数据）",
    "author": "Jennifer Li、Justine Moore",
    "date": "2026-02",
    "url": "https://a16z.com/"
   }
  },
  {
   "code": "d1",
   "name": "剧本与分镜工具",
   "ai_era": "编剧写本、分镜师手画 → 小说 / 剧本自动拆集，出角色设定和分镜",
   "box": "AI 短剧与漫剧工具",
   "box_zh": "只收工具：短剧 App 和出品方不收",
   "source": {
    "org": "招商证券",
    "title": "AI 影视行业深度报告：AI 影视 AI 应用的扛旗者",
    "author": "顾佳、谢笑妍、刘晓珊",
    "date": "2026-11-01",
    "url": "http://stockfinance.sina.cn/stock/go.php/paper/reportid/846849258986/index.phtml"
   }
  },
  {
   "code": "d2",
   "name": "一站式短剧 / 漫剧生产工作台",
   "ai_era": "剧组 20–40 人拍 25–40 天 → 5–10 人 7–10 天，「剧本直达成片」",
   "box": "AI 短剧与漫剧工具",
   "box_zh": "只收工具：短剧 App 和出品方不收",
   "source": {
    "org": "招商证券",
    "title": "AI 影视行业深度报告：AI 影视 AI 应用的扛旗者",
    "author": "顾佳、谢笑妍、刘晓珊",
    "date": "2026-11-01",
    "url": "http://stockfinance.sina.cn/stock/go.php/paper/reportid/846849258986/index.phtml"
   }
  },
  {
   "code": "d3",
   "name": "译制与投放工具",
   "ai_era": "人工译制几个月 → AI 译配几小时；投手经验 → 投放模型、批量素材",
   "box": "AI 短剧与漫剧工具",
   "box_zh": "只收工具：短剧 App 和出品方不收",
   "source": {
    "org": "招商证券",
    "title": "AI 影视行业深度报告：AI 影视 AI 应用的扛旗者",
    "author": "顾佳、谢笑妍、刘晓珊",
    "date": "2026-11-01",
    "url": "http://stockfinance.sina.cn/stock/go.php/paper/reportid/846849258986/index.phtml"
   }
  }
 ],
 "who": [
  {
   "code": "W-model",
   "name": "模型",
   "desc": "自己训练视频模型的公司",
   "layers": [
    "v1",
    "v2",
    "v7"
   ],
   "section": "开发者和企业"
  },
  {
   "code": "W-api",
   "name": "API",
   "desc": "多模型推理服务",
   "layers": [
    "v3"
   ],
   "section": "开发者和企业"
  },
  {
   "code": "W-make",
   "name": "生成",
   "desc": "多模型工作台、视频 agent",
   "layers": [
    "v4"
   ],
   "section": "创作者"
  },
  {
   "code": "W-edit",
   "name": "剪辑",
   "desc": "剪、配字幕、补镜头",
   "layers": [
    "v5"
   ],
   "section": "创作者"
  },
  {
   "code": "W-ads",
   "name": "广告视频",
   "desc": "UGC 广告、批量变体",
   "layers": [
    "v6"
   ],
   "section": "营销团队"
  },
  {
   "code": "W-produce",
   "name": "写本到成片",
   "desc": "剧本、分镜、生成工作台",
   "layers": [
    "d1",
    "d2"
   ],
   "section": "短剧制作方"
  },
  {
   "code": "W-ship",
   "name": "出海和投放",
   "desc": "译制、投放素材",
   "layers": [
    "d3"
   ],
   "section": "短剧制作方"
  }
 ]
}