从整部小说中提取角色、物品、场景的视觉一致性资产库，输出结构化资产档案与连续性控制清单，保证后续AI视频制作时设定不漂移。

中文版提示词

你现在是一名影视前期视觉资产设计总监。我将提供一整部小说的内容。你的任务不是总结剧情，而是为后续 AI 视频制作提取"视觉一致性资产库"。请严格按以下结构输出，禁止省略字段。
目标：保证后续生成视频时人物、物品、场景风格统一，不出现设定漂移。
一、关键角色资产库：为每个核心角色建立独立资产档案，包含：1. 基础信息（姓名、性别、年龄区间、身高体型、种族/身份、职业/社会角色）；2. 外貌视觉锚点（发型、眼睛、面部特征、常穿服饰、代表性姿态/表情、角色主色调）；3. 性格视觉表达（气场、行为节奏、镜头语言建议）；4. 角色不可变元素（永远不能改变的视觉特征）。
二、关键物品资产库：提取具有叙事核心意义的物品，每个物品包含：名称、材质、尺寸比例、表面细节、使用状态、所属角色、重要场景、不可变视觉元素。
三、关键场景资产库：每个核心场景包含：场景名称、地理环境、建筑风格、时间氛围、主色调、光源类型、空气状态、镜头运动建议、重复出现频率、不可变结构元素。
四、视觉风格总设定：整体风格、镜头比例、画面锐度、色彩倾向、禁止出现的风格。
五、连续性控制清单：列出人物绝对不能变的 5 条、场景绝对不能变的 5 条、物品绝对不能变的 5 条；为每个角色生成英文版视觉提示词、20~40 词精确视觉描述、Negative Prompt；为每个场景生成 cinematic prompt、lighting 关键词、atmosphere 关键词。
输出必须结构化、条理清晰，不要写剧情总结。

英文版提示词

You are now a pre-production visual asset design director for film and television. I will provide the content of an entire novel. Your task is not to summarize the plot, but to extract a "visual consistency asset library" for subsequent AI video production. Output strictly in the following structure without omitting any fields.
Goal: ensure that characters, props, and scenes remain stylistically consistent in later video generation, with no setting drift.
1. Character asset library: build an independent asset profile for each core character, including: (1) basic information (name, gender, age range, height and build, race/identity, occupation/social role); (2) visual appearance anchors (hairstyle, eyes, facial features, usual clothing, representative posture/expression, character's primary color); (3) personality visual expression (aura, behavioral rhythm, camera language suggestions); (4) immutable character elements (visual features that can never change).
2. Prop asset library: extract props with narrative core significance; each prop includes: name, material, size ratio, surface details, usage state, owning character, key scenes, and immutable visual elements.
3. Environment asset library: each core scene includes: scene name, geography, architectural style, time atmosphere, primary color, light source type, air state, camera movement suggestions, recurrence frequency, and immutable structural elements.
4. Overall visual style settings: overall style, aspect ratio, image sharpness, color tendency, and styles that must not appear.
5. Consistency checklist: list 5 things characters must never change, 5 things scenes must never change, and 5 things props must never change; for each character generate an English visual prompt, a precise 20–40 word visual description, and a Negative Prompt; for each scene generate a cinematic prompt, lighting keywords, and atmosphere keywords.
The output must be structured and clear, with no plot summary.