可惜我不抽セブンスタ

【生成内容】 日式情绪短片。7 场。\ 女主以"讲故事 + 反应"贯穿全片，自言自语，漂浮体（贴着当下时态）。\ 男主完全不出画面，只通过 2 条微信/LINE 异步语音存在（S2 + S5）。\ 基调：日式小清新，不要伤痛文学。\ 女主说话方式：INFP+天秤——不下定论、半句话、自我对话、对小东西共情。 ## 三条创作铁律 1. 不解释 2. 不点题 3. 用具体反空泛 ## 隐线 S1-S2 她还会被声音勾一下 → S3-S4 进入自己的节奏 → S5 手机震但没拿 → S6 烟火"哇"真的破 → S7 主动把手机扣下。 --- ## S1 · 洗衣店 · 夜 独白： > 这家洗衣店，\ > 不知道为什么我挺喜欢的。\ > 灯太亮了，\ > 这样反而安心。 > > 便利店的鲑鱼饭团，\ > 偶尔会咬到一颗特别咸的米。\ > 不是不好吃，\ > 就是会停一下。 > > 玻璃门外，\ > 一只橘猫慢慢走过去，\ > 看都没看我。\ > 嗯… 也对，\ > 它有它要去的地方。 > > 刚烘好的衣服抱在怀里，\ > 是热的。 > > 明天，\ > 想买点葡萄。 环境声：滚筒水声 / 荧光灯嗡鸣\ 画面信号：她坐着吃饭团 / 抱着衣服出门\ 勾连埋点：鲑鱼饭团 → S6 烟花味觉回闪；葡萄 → 后续场景找机会回扣\ 男主语音：无

【生成内容】 日式情绪短片。7 场。\ 女主以"讲故事 + 反应"贯穿全片，自言自语，漂浮体（贴着当下时态）。\ 男主完全不出画面，只通过 2 条微信/LINE 异步语音存在（S2 + S5）。\ 基调：日式小清新，不要伤痛文学。\ 女主说话方式：INFP+天秤——不下定论、半句话、自我对话、对小东西共情。 ## 三条创作铁律 1. 不解释 2. 不点题 3. 用具体反空泛 ## 隐线 S1-S2 她还会被声音勾一下 → S3-S4 进入自己的节奏 → S5 手机震但没拿 → S6 烟火"哇"真的破 → S7 主动把手机扣下。 --- ## S2 · 自动贩卖机 · 深夜 独白： > 这条街上的自动贩卖机，\ > 我比较喜欢拐角那一台。\ > 灯是暖色的，\ > 不像别的那么白。 > > 今天想喝什么呢…\ > 按钮一颗一颗看过去，\ > 最后还是按了平时那瓶。\ > 也没什么新鲜事。 \[手机震 — 男主语音\]\ \[她看一眼，按下取货键\] > 罐子掉下来的声音，\ > 比想象中响。\ > 冰得有点过分了，\ > 等一下再喝吧。 环境声：贩卖机嗡鸣 / 罐子落下 / 远处车声\ 画面信号：手指在按钮前停一下 / 手机震她看一眼放下\ 勾连埋点：这瓶饮料 → S4 冰箱里再出现\ 男主语音：「起きてる？」（睡着没？）—— 1 秒短语音

镜头1（6s） 剧情：女主在明亮的洗衣店里吃鲑鱼饭团，看着门外，享受短暂的安心感与放空状态。 景别：中景 角色动作：女主双手轻轻捧着一个吃到一半的鲑鱼饭团，缓慢咀嚼，眼神放空地看向画面外的玻璃门，随后微微收紧怀抱中刚烘好的衣物。 情绪：平静、轻微的游离与安心感 场景标签：深夜洗衣店，滚筒洗衣机，冷色荧光灯，玻璃门 光影氛围：明亮偏冷的荧光灯光均匀洒下，对比度低，营造出一种与外部黑夜隔绝的孤独但安全的大气环境。 音效：滚筒洗衣机转动的水声，荧光灯微弱的嗡鸣声 台词：（独白）这家洗衣店，不知道为什么我挺喜欢的。灯太亮了，这样反而安心。便利店的鲑鱼饭团，偶尔会咬到一颗特别咸的米。不是不好吃，就是会停一下。玻璃门外，一只橘猫慢慢走过去，看都没看我。嗯… 也对，它有它要去的地方。刚烘好的衣服抱在怀里，是热的。明天，想买点葡萄。 角色【女主】：[女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] ▼ 图片生成提示词： [画面构图：中景，平视机位，人物位于画面右三分之一处] + 镜头中 [女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] 正坐在洗衣店的长椅上，双手微微捧着一个鲑鱼饭团 + [主体与空间关系：背景是一排白色的滚筒洗衣机，左侧有反光的玻璃门透出夜色] + [微表情：眼神略微失焦，咀嚼动作缓慢而轻柔，带着一种发呆的松弛感] + [环境元素：白色的洗衣店环境，背景有几件颜色柔和的衣物在滚筒内] + [光影：顶部的冷白荧光灯无死角照明，无明显阴影] + [视觉风格：日式胶片感，低对比度，空气感] + [技术参数：35mm镜头，f/2.8光圈]. ▼ 视频动作提示词： [缓慢推镜头] + [女主轻轻咬了一小口饭团，停顿了一下，似乎咬到了那颗咸米，眼神微微闪烁，随后目光移向侧前方，双手略微抱紧腿上的衣服] + [背景洗衣机滚筒匀速旋转，带来细微的光影闪动] + [无风，极度宁静的生活流动感] + [时长: 6秒] 镜头2（5s） 剧情：女主买到了葡萄走在路上，被男主发来的语音消息短暂打断了思绪，注意力被声音勾住。 景别：近景 角色动作：女主提着装有紫葡萄的透明塑料袋，听到手机提示音后停下脚步，微微低头看向画面外的手机，思绪被打断。 情绪：思绪被牵绊，微微的迟疑 场景标签：日式街道，白云，电线杆，微风 光影氛围：清晨的自然光，阳光透过云层形成柔和的漫反射，明亮且通透。 音效：街道微风声，清脆的LINE消息提示音 台词：（LINE语音/男主）昨天那家餐厅的伞，你是不是拿错了？ （女主独白）葡萄上的白霜，洗掉的话… 会不会有点可惜。 角色【女主】：[女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] ▼ 图片生成提示词： [画面构图：近景，微微仰拍，以天空为背景] + 镜头中 [女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] 正微低着头，单手提着一串饱满的紫葡萄 + [主体与空间关系：女主占据画面中心，背后是淡蓝色的天空与模糊的日式电线杆] + [微表情：被声音打断思绪，眼眸微垂，眼神中有一丝波澜，略显迟疑] + [环境元素：透明塑料袋里的葡萄带有明显的白霜，质感清晰生动] + [光影：通透的自然日光，侧逆光勾勒出发丝的轮廓] + [视觉风格：日式小清新，高明度] + [技术参数：50mm镜头，f/1.8大光圈，背景柔和虚化]. ▼ 视频动作提示词： [手持微晃固定镜头] + [女主正看着手里的葡萄，突然听到提示音，动作微顿，微微抬眼，眼神中闪过一丝波澜，但身体没有大幅度动作，仅在原地发呆] + [微风吹过，衣角轻摆，塑料袋在风中发出细微光斑变化] + [时长: 5秒] 镜头3（4s） 剧情：女主在水槽边洗葡萄，开始进入自己的生活节奏，感受水流微小的触感。 景别：特写 角色动作：女主的双手在水流下轻轻搓洗着葡萄表皮，大拇指缓缓摩挲。 情绪：沉浸，感官的细腻放大 场景标签：水槽，水流，紫葡萄 光影氛围：午后的阳光透过窗户洒在水槽上，波光粼粼，温暖舒适。 音效：清脆细微的流水声，风吹过窗外晾衣架的微响 台词：（独白）水流过手指的时候，有一点痒。风… 好像吹了三下。 角色【女主】：[女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] ▼ 图片生成提示词： [画面构图：特写，俯拍机位] + 镜头聚焦于 [女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] 的双手，她正在不锈钢水槽内清洗紫色的葡萄 + [主体与空间关系：手部与葡萄占据画面主体，背景是模糊的水槽边缘] + [微表情：虽然未露全脸，但放松的手部姿态传递出沉浸感] + [环境元素：晶莹剔透的水珠挂在紫色的葡萄表皮上，冲刷掉部分白霜] + [光影：午后阳光斜射，水流折射出细碎的金色光斑] + [视觉风格：微距质感，日式生活流] + [技术参数：100mm微距镜头，f/2.8光圈]. ▼ 视频动作提示词： [缓慢平移镜头] + [清澈的水流冲刷着葡萄和女主的手指，她的大拇指极其轻柔地摩挲着葡萄表皮，动作缓慢且享受] + [水珠在阳光下折射出动态的光泽，水流连续不断] + [时长: 4秒] 镜头4（4s） 剧情：女主在黄昏的光线中观察空气中的灰尘，彻底进入自己的内心世界，不再被外界干扰。 景别：近景 角色动作：女主趴在木桌上，视线平视着前方空气中悬浮的灰尘，静静地一动不动。 情绪：极致的宁静，发呆的抽离感 场景标签：木桌，光束，悬浮灰尘 光影氛围：黄昏的金色逆光形成丁达尔效应，温暖而带有一丝慵懒的迟暮感。 音效：极其微弱的呼吸声，无 台词：（独白）灰尘在光里面… 只要不盯着看，它们就是静止的吧。 角色【女主】：[女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] ▼ 图片生成提示词： [画面构图：近景，侧面机位，视线平齐] + 镜头中 [女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] 安静地趴在原木质感的桌面上 + [主体与空间关系：女主在画面左侧，右侧是一道倾斜的金色光束] + [微表情：眼神跟随着空气中肉眼难辨的微小事物移动，处于极度发呆与自我沉浸的状态] + [环境元素：安静的木桌，背景模糊的室内陈设] + [光影：黄昏的强烈侧逆光照亮了空气中悬浮的细小灰尘，形成极具氛围的光束] + [视觉风格：胶片质感，温暖慵懒] + [技术参数：50mm镜头，f/1.4光圈，大虚化]. ▼ 视频动作提示词： [固定镜头] + [女主安静地趴着，几乎没有身体动作，仅眼睛微微眨动，视线随着光束里浮动的灰尘缓缓移动，呼吸均匀且轻柔] + [灰尘在金色的光线中进行缓慢无规则的布朗运动，光影细微变化] + [时长: 4秒] 镜头5（5s） 剧情：夜晚，手机震动弹出男主消息，但女主沉浸在剥葡萄的动作里，并未理会震动的手机。 景别：中近景 角色动作：女主盘腿坐在地毯上剥着葡萄皮，旁边的手机屏幕亮起并震动，她余光瞥见但手并未停下剥皮的动作。 情绪：情绪稳定，边界感的建立 场景标签：夜晚房间，暖色台灯，剥皮的葡萄，亮起的手机 光影氛围：室内仅有一盏暖黄色的台灯照明，局部光影温馨静谧，周围呈现柔和的暗调。 音效：手机在木桌上震动的低频嗡嗡声 台词：（LINE语音/男主声音发闷似背景音）周末… 要不要一起出来走走？ （女主独白）半透明的果肉，有点像… 丢掉的玻璃球。 角色【女主】：[女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] ▼ 图片生成提示词： [画面构图：中近景，略带俯视的平视角度] + 镜头中 [女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] 盘腿坐在地毯上，双手正捏着一颗剥了一半皮的葡萄，旁边木桌上放着一部亮起屏幕的智能手机 + [主体与空间关系：人物为中心，发光的手机作为前景侧面的点缀] + [微表情：余光察觉到手机亮起，但眼神没有移过去，依然专注盯着手里的半透明葡萄，面容平静] + [环境元素：夜晚温馨的房间，原木质感家具，一小碗洗好的葡萄] + [光影：台灯的暖黄色局部照明照亮她的脸颊和双手，手机屏幕散发微弱冷光形成对比] + [视觉风格：日式居家感，静谧温馨] + [技术参数：35mm镜头，f/2.0光圈]. ▼ 视频动作提示词： [缓慢拉镜头] + [桌面的手机突然亮起屏幕，女主的手指微微停顿了一瞬，但随即继续专注地剥落葡萄皮，完全没有伸手去拿手机的意图] + [室内光线静谧，手机屏幕的冷光在木桌上产生微弱的动态反光] + [时长: 5秒] 镜头6（5s） 剧情：女主在阳台上看到远处的烟火，发出真实的惊叹，感官被打通，联想到第一场的饭团。 景别：近景 角色动作：女主靠在夜色下的阳台栏杆上，突然天空中炸开微弱的烟火，她眼睛微微睁大，发出轻微的惊叹。 情绪：被真实的瞬间打动，轻盈的愉悦 场景标签：夜晚阳台，远处烟花，夏末微风 光影氛围：深蓝色的夜空背景，烟火的暖色光芒瞬间照亮女主的脸庞，产生明暗交替的戏剧感。 音效：远处烟火沉闷的砰砰声，夜风拂过衣物的声音 台词：（女主轻呼出声）哇... （独白）今天的风，怎么会有咸咸的鲑鱼味。 角色【女主】：[女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] ▼ 图片生成提示词： [画面构图：近景，正面微侧机位] + 镜头中 [女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] 双手撑在阳台栏杆上，抬头望着夜空 + [主体与空间关系：女主占画面大比例，背景是深邃的夜空与模糊的城市灯火] + [微表情：眼睛微微睁大，瞳孔中清晰反射出绚丽的烟火光芒，嘴唇微启，流露出不加掩饰的真实惊叹与极浅的微笑] + [环境元素：夜晚的金属栏杆，被微风吹动的短发] + [光影：夜空中突然亮起的烟火光芒，以暖红色与橘色为主，瞬间映照在她的脸庞与衣服上] + [视觉风格：电影感夜景，色彩反差，情绪高点] + [技术参数：50mm镜头，f/1.4大光圈，暗光捕捉]. ▼ 视频动作提示词： [缓慢推镜头特写] + [女主正看着天空，突然一道明亮的烟花光芒照亮画面，她眼睛一亮，不自觉地微微张开嘴发出“哇”的口型，随后露出一个真诚而松弛的浅笑] + [烟火的橘色光影在她脸上产生动态的明暗交替，夜风吹动她的空气刘海] + [时长: 5秒] 镜头7（4s） 剧情：女主回到桌前，主动将亮着屏幕的手机扣在桌面上，完成了情绪的释怀与状态切断。 景别：特写 角色动作：女主伸出手指，干脆而轻柔地将桌面上依然亮着屏幕的手机翻转，屏幕朝下扣在木桌上。 情绪：释然，彻底放下 场景标签：木桌，手机，手部特写 光影氛围：暖黄色的台灯光，营造出一种告别后的安全与放松氛围。 音效：手机接触木桌面的轻微‘嗒’声 台词：（独白）明天… 换个口味吧。 角色【女主】：[女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] ▼ 图片生成提示词： [画面构图：特写，俯拍机位] + 镜头中 [女主: 20岁出头的年轻女孩，日式空气刘海，齐肩短发，穿着宽松的棉麻质感米色长袖衬衫与浅卡其色长裙，神态柔和，带有INFP的内敛与文艺气质，眼神清澈且经常发呆，给人一种轻盈、透明的呼吸感] 的一只手进入画面，正捏住桌面上亮着屏幕的手机边缘，将其向下翻转 + [主体与空间关系：手部与手机是绝对视觉中心，背景是木质桌面纹理] + [微表情：画面仅露出下半部分脸颊与嘴角，能看到一抹轻松释然的微小弧度] + [环境元素：旁边放着装有几颗剩下葡萄的小碗，充满生活气息] + [光影：暖黄色的室内台灯光在手背上打出柔和的高光，手机反扣后冷光消失] + [视觉风格：日式静物特写，克制且余韵悠长] + [技术参数：85mm镜头，f/2.8光圈]. ▼ 视频动作提示词： [固定镜头] + [女主的手轻柔但干脆地捏住手机边缘，将其屏幕朝下扣在桌面上，随后手指松开并缓缓离开画面边缘] + [随着手机被扣下，画面中那一丝微弱的屏幕冷光彻底熄灭，只留下温暖安静的台灯光芒] + [时长: 4秒]

镜头1（3s） 剧情：深夜空旷的街道，拐角处的自动贩卖机亮着暖光 景别：远景 (Long Shot) 角色动作：无角色出现 情绪：孤独、安静 场景标签：日本街头，深夜，自动贩卖机，路灯，柏油路 光影氛围：贩卖机的琥珀色暖光与深蓝色的夜幕形成冷暖对比，空气中有微弱的夜雾 音效：贩卖机低沉的嗡鸣声，远处的微风声 ▼ 图片生成提示词： [画面构图：远景，低角度仰拍，对称偏右构图] 深夜空旷的日本街道拐角，一台孤零零的自动贩卖机散发着强烈的琥珀色暖光。[无人物] [明确的场景环境元素与前景/背景道具] 街道两侧是紧闭的木质百叶窗与暗色墙壁，湿润的柏油路面反射着贩卖机的黄色光斑，远处的街灯发出微弱的冷白光。[光影几何与大气效果] 丁达尔效应微弱，暖光在深蓝色的夜雾中晕开，形成强烈的冷暖色调对比。[视觉风格/胶片质感] 日系小清新，富士胶片低对比度色彩，情绪颗粒感极强。[技术参数：35mm广角镜头，f/2.8光圈，阿莱摄影机] ▼ 视频动作提示词： [缓慢的推镜头] 摄影机从空旷的街道中心极其缓慢地向发光的贩卖机推进。[环境物理动态：夜风吹过，路边的一根电线微微摇晃，光影在湿润的地面上产生轻微的涟漪感] [音效与音乐氛围描述] 夜晚绝对的寂静，只有贩卖机极其轻微的电子嗡鸣声，营造出一种与世隔绝的漂浮感。[时长: 3.0秒] 镜头2（4s） 剧情：小初站在贩卖机前，静静地看着 景别：中景 (Medium Shot) 角色动作：站在机器前，双手放在开衫口袋里 情绪：惬意、漫无目的 场景标签：贩卖机前，夜景，暖光 光影氛围：侧逆光，贩卖机的光照亮她的侧脸轮廓 音效：脚步声停下，轻微的衣物摩擦声 台词：这条街上的自动贩卖机，我比较喜欢拐角那一台。 角色【小初】：[小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] ▼ 图片生成提示词： [画面构图：中景，平视机位，三分法构图] [小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] 站在画面右侧，面向左侧发光的贩卖机。[人物空间与互动关系：人物占据右侧，左侧为发光的机器面板，形成视觉平衡] [极具体的微表情与面部特写] 眼神柔和地看着前方，嘴唇微闭，带着一种放松的游离感。[明确的场景环境元素与前景/背景道具] 夜晚的街道背景被完全虚化。[光影几何与大气效果] 贩卖机的暖黄色光线柔和地打在她的侧脸、鼻尖和几缕垂下的发丝上，边缘泛着金光，暗部呈现深青色。[视觉风格/胶片质感] 极浅景深，电影感散景，复古胶片感。[技术参数：50mm镜头，f/1.4大光圈] ▼ 视频动作提示词： [固定镜头，细微的呼吸摇晃] 摄影机保持平稳，捕捉人物站定的瞬间。[主体极其具体的物理动作细节与微表情变化：小初双手插在针织衫口袋里，身体随着呼吸有非常轻微的起伏，眼神极其缓慢地在机器面板上游移] [环境物理动态：夜风轻轻吹动她耳边的几缕碎发，针织衫的边缘微微摆动] [音效与音乐氛围描述] 日系清脆的吉他泛音缓缓响起，伴随轻微的衣料摩擦声。[对话台词与语气] (内心独白，语气轻柔且漂浮) “这条街上的自动贩卖机，我比较喜欢拐角那一台。” [时长: 4.0秒] 镜头3（3s） 剧情：特写贩卖机内部的暖色灯管 景别：特写 (Close-up) 角色动作：无 情绪：温暖、细腻 场景标签：灯管，商品陈列，光晕 光影氛围：柔和扩散的暖光，高光微泛光斑 音效：贩卖机微弱的电流声 台词：灯是暖色的，不像别的那么白。 ▼ 图片生成提示词： [画面构图：特写，微距平视] 自动贩卖机内部陈列窗的局部特写。[无人物] [明确的场景环境元素与前景/背景道具] 一排排整齐排列的日系饮料罐，商品标签在光线照射下显得清晰，顶部是一根散发着柔和琥珀色光芒的复古灯管。[光影几何与大气效果] 光线在展示玻璃表面产生极其柔和的折射与光晕（Halation），画面整体充斥着一种温暖、毛茸茸的视觉触感，暗部细节丰富。[视觉风格/胶片质感] 柔焦滤镜效果，日系高调色彩，温暖治愈。[技术参数：100mm微距镜头，f/2.8光圈，局部对焦] ▼ 视频动作提示词： [极其缓慢的平摇镜头] 摄影机沿着发光的灯管和饮料罐从左向右极其缓慢地平移（Pan）。[环境物理动态：灯管的光芒产生微弱的频闪感，光晕在镜头镜片上拉出柔和的柔光效果] [音效与音乐氛围描述] 细微的电子嗡鸣声被放大，吉他泛音持续。[对话台词与语气] (内心独白，自我对话般的呢喃) “灯是暖色的，不像别的那么白。” [时长: 3.0秒] 镜头4（4s） 剧情：小初的脸部特写，视线在按钮间游移 景别：特写 (Close-up) 角色动作：眼神上下移动，思考 情绪：犹豫、日常的放空 场景标签：面部特写，暖光反射 光影氛围：脸上有着按键光斑的微微反射 音效：无 台词：今天想喝什么呢… 角色【小初】：[小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] ▼ 图片生成提示词： [画面构图：面部大特写，平视机位] [小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] 的面部大特写。[人物空间与互动关系：脸部充满整个画面，背景完全融入黑暗] [极具体的微表情与面部特写] 她的眼睛微微睁大，瞳孔中清晰地反射出贩卖机一排排按钮的发光倒影，嘴角带着一丝似有似无的慵懒笑意。[明确的场景环境元素与前景/背景道具] 无明显背景，纯粹的光影塑造面部。[光影几何与大气效果] 一排排细碎的彩色按钮光斑投射在她的脸颊和额头上，形成生动的色彩点缀，整体光照温暖且柔和。[视觉风格/胶片质感] 细腻的肌肤纹理，胶片微粒，极度唯美。[技术参数：85mm人像镜头，f/1.2极致浅景深] ▼ 视频动作提示词： [极其缓慢的推镜头] 镜头几乎不易察觉地向角色的眼眸推进。[主体极其具体的物理动作细节与微表情变化：小初的眼球有规律地从左向右、从上向下极其细微地转动，正在阅读饮料标签，睫毛轻轻眨动了一下，透出一种不着急的慵懒感] [音效与音乐氛围描述] 轻柔的吉他拨弦停顿留白。[对话台词与语气] (内心独白，轻飘飘的疑问) “今天想喝什么呢…” [时长: 4.0秒] 镜头5（3.5s） 剧情：手指悬停在发光的按钮上方 景别：特写 (Close-up) 角色动作：指尖在按钮前犹豫 情绪：选择困难（INFP特质） 场景标签：指尖，发光按钮，贩卖机面板 光影氛围：按键的红色与蓝色光交织在指尖 音效：衣服袖口的摩擦声 台词：按钮一颗一颗看过去， 角色【小初】：[小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] ▼ 图片生成提示词： [画面构图：手部微距特写，俯视角度] 画面聚焦于 [小初: 20岁出头的年轻亚洲女性，穿着宽松柔软的米色针织开衫，露出白皙纤细的手指] 的右手。[人物空间与互动关系：手指悬停在贩卖机面板上方约一厘米处，占据画面对角线] [极具体的微表情与面部特写] 无面部，重点展现指尖的姿态，手指微微蜷缩，显得犹豫不决。[明确的场景环境元素与前景/背景道具] 背景是贩卖机上一排排发着微光的透明圆形按钮，上面标着日文价格。[光影几何与大气效果] 按钮发出的红蓝光芒交替映照在她的指尖和针织衫起毛的袖口上，呈现出极具肌理感的光影质感。[视觉风格/胶片质感] 极浅景深，焦外光斑圆润。[技术参数：100mm微距镜头，f/2.8光圈] ▼ 视频动作提示词： [固定机位] 摄影机保持绝对静止，焦点死死锁定在指尖与按钮之间。[主体极其具体的物理动作细节与微表情变化：小初的食指悬停在一个按钮上方，轻微颤动了一下，随后像是在思考般，指尖极其缓慢地向右侧的另一个按钮平移，并没有立刻按下] [环境物理动态：针织衫袖口的细小绒毛在光线下微微颤动] [音效与音乐氛围描述] 细微的衣袖摩擦声。[对话台词与语气] (内心独白，慢条斯理) “按钮一颗一颗看过去，” [时长: 3.5秒] 镜头6（3s） 剧情：最终按下熟悉的那个按钮 景别：特写 (Close-up) 角色动作：按下按钮 情绪：习惯、安分 场景标签：按键亮起，手指按下 光影氛围：按下去的瞬间，按钮内部的灯光发生变化 音效：清脆的按键声“咔哒” 台词：最后还是按了平时那瓶。也没什么新鲜事。 角色【小初】：[小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] ▼ 图片生成提示词： [画面构图：手部微距特写，侧面平视角度] [小初: 20岁出头的年轻亚洲女性，手指纤细，穿着宽松柔软的米色针织开衫] 的手指坚定地按在一个透明圆形的自动贩卖机按钮上。[人物空间与互动关系：手指紧贴按钮，按压产生的物理凹陷感清晰可见] [极具体的微表情与面部特写] 仅展示手部，动作显得果断了一点。[明确的场景环境元素与前景/背景道具] 一排按钮中的一个，按钮内部有着透明的塑料质感，上面写着日语。[光影几何与大气效果] 按钮被按下的瞬间，内部的背光灯骤然变亮，强烈的光线穿透指尖边缘，形成微弱的透射光（次表面散射）。[视觉风格/胶片质感] 高清微距质感，日系胶片色彩，冷暖光对比。[技术参数：100mm微距镜头，f/2.8光圈] ▼ 视频动作提示词： [极轻微的推镜头] 镜头跟随手指的动作向前极小幅度推进。[主体极其具体的物理动作细节与微表情变化：犹豫的食指最终坚定地落在最右侧的一颗按钮上，向下按压到底，手指肌肉有轻微的发力感，按钮按下后灯光瞬间闪烁变亮] [音效与音乐氛围描述] 伴随一声清脆且极具机械质感的“咔哒”按键声。[对话台词与语气] (内心独白，带着一丝自我妥协的释然) “最后还是按了平时那瓶。也没什么新鲜事。” [时长: 3.0秒] 镜头7（3.5s） 剧情：手机在口袋或手中突然震动亮起 景别：中近景 (Medium Close-up) 角色动作：低头看向亮起的手机 情绪：轻微的意外，被打断 场景标签：口袋边缘，手机屏幕亮起，夜色 光影氛围：手机屏幕的冷白光突然打破了原本的暖色调 音效：急促的两下手机震动声“嗡嗡” 角色【小初】：[小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] ▼ 图片生成提示词： [画面构图：中近景，俯视角度，腰部特写] [小初: 20岁出头的年轻亚洲女性，日系小清新风格，穿着宽松柔软的米色针织开衫和棉质长裙] 的腰部和手部局部。[人物空间与互动关系：她的左手握着一部智能手机，手机屏幕朝上，打破了画面的宁静] [极具体的微表情与面部特写] 无面部，从身体语言（肩膀轻微的停顿）传达出注意力被转移的状态。[明确的场景环境元素与前景/背景道具] 针织衫的纹理、棉质长裙的褶皱，以及手中突然亮起的手机屏幕。[光影几何与大气效果] 手机屏幕瞬间亮起的冷白光极其刺眼，与贩卖机投射在衣服上的琥珀色暖光形成强烈的冷暖色调冲突，照亮了针织衫的绒毛。[视觉风格/胶片质感] 真实的光照物理反馈，暗部压暗，保留胶片噪点。[技术参数：50mm镜头，f/2.0光圈] ▼ 视频动作提示词： [固定镜头] 画面保持静止，记录光线突变的瞬间。[主体极其具体的物理动作细节与微表情变化：小初原本闲置的左手中，手机屏幕突然亮起并伴随强烈的物理震动，她的手因为震动本能地轻微颤了一下，大拇指悬停在屏幕边缘，没有立刻滑动解锁] [环境物理动态：手机屏幕的冷光瞬间点亮了周围原本暗沉的布料褶皱] [音效与音乐氛围描述] 寂静中突然爆发出极度清晰、干脆的手机震动声“嗡嗡”两下，打破了刚刚的宁静氛围。[时长: 3.5秒] 镜头8（3s） 剧情：手机屏幕特写，收到男主LINE语音消息 景别：特写 (Close-up) 角色动作：无 情绪：悬念，被拉回现实 场景标签：手机屏幕，LINE消息界面，语音条 光影氛围：纯粹的屏幕冷光源，周围极暗 音效：LINE的消息提示音 / 男主异步语音 台词：（男主语音）起きてる？ ▼ 图片生成提示词： [画面构图：极近特写，过肩主观视角（POV）] 屏幕亮起的智能手机界面特写。[无人物面部] [明确的场景环境元素与前景/背景道具] 手机屏幕上清晰显示着LINE应用的锁屏通知横幅，横幅上显示收到一条时长1秒的语音消息图标，虽然文字模糊但能看清UI界面特征。[光影几何与大气效果] 屏幕发出的冷白光刺透黑暗，屏幕玻璃表面有着极其细微的指纹痕迹和来自贩卖机暖光的微弱反光，背景完全陷入失焦的黑暗与散景中。[视觉风格/胶片质感] 偏冷的科技感光线与日系复古环境的碰撞，极浅景深。[技术参数：85mm微距镜头，f/1.8光圈] ▼ 视频动作提示词： [极其缓慢的推镜头] 镜头模拟人眼的专注度，极其缓慢地向亮起的手机屏幕推进。[环境物理动态：屏幕的高光在镜头前产生细微的光晕（Bloom），几秒后屏幕光线极其细微地变暗了一阶（即将息屏的预兆）] [音效与音乐氛围描述] 清脆的LINE“叮咚”提示音响起，紧接着直接切入男主从手机听筒质感传出的低沉且随意的异步语音。[对话台词与语气] (男主语音，带点深夜的沙哑和随意) “起きてる？”（睡着没？） [时长: 3.0秒] 镜头9（4s） 剧情：小初看了一眼手机，但没有回复，将手放下 景别：近景 (Close-up) 角色动作：垂下眼帘看手机，随后若无其事地移开视线 情绪：平静、不下定论、刻意留白 场景标签：人物面部，冷暖光交织 光影氛围：脸部下半部被手机冷光照亮，随后冷光消失 音效：远处偶尔驶过的汽车轮胎声 角色【小初】：[小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] ▼ 图片生成提示词： [画面构图：近景，微侧面平视角度] [小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫，眼神温柔且带有一丝游离感，五官清秀，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] 的近景。[人物空间与互动关系：人物居中，视线向下注视着画外底部的手机] [极具体的微表情与面部特写] 眼神低垂，瞳孔中映着一抹冷白光，嘴角没有表情，眼神中闪过一丝极细微的波动（被声音勾了一下），但瞬间恢复了那种游离的平静。[明确的场景环境元素与前景/背景道具] 虚化的黑夜背景。[光影几何与大气效果] 脸部下方被手机的冷光照亮，与侧面贩卖机的暖光在脸颊上形成极其唯美的冷暖光过渡（Split lighting），随后下方的冷光瞬间熄灭，脸部重新回到纯粹的暖光中。[视觉风格/胶片质感] 岩井俊二式的克制感，胶片颗粒感。[技术参数：50mm镜头，f/1.4大光圈] ▼ 视频动作提示词： [固定机位] 镜头静止，死死捕捉面部情绪的变化。[主体极其具体的物理动作细节与微表情变化：小初低垂着眼眸看了一秒，眼睫毛轻轻闪动，并没有伸手去操作，而是十分平淡地将握着手机的手垂直放下，随着手机的落下，她脸上的冷白光瞬间消失。她微微抬起头，视线重新看回正前方的贩卖机] [音效与音乐氛围描述] 一辆汽车在极远的街道驶过，留下低沉的“呼——”的轮胎摩擦声，凸显出此刻的空虚与安静。[时长: 4.0秒] 镜头10（3s） 剧情：饮料罐重重地掉入取货口 景别：特写 (Close-up) 角色动作：无 情绪：突兀、打破沉静 场景标签：取货口，饮料罐落下 光影氛围：取货口内部的阴影，反光的金属罐 音效：极响亮的“咣当”声 台词：罐子掉下来的声音，比想象中响。 ▼ 图片生成提示词： [画面构图：特写，低角度俯视取货口内部] 自动贩卖机底部深灰色的取货口内部特写。[无人物] [明确的场景环境元素与前景/背景道具] 取货口的黑色塑料挡板，一罐带有水珠的绿色日系茶饮料/咖啡罐。[光影几何与大气效果] 外部的暖光极其微弱地漏进这个相对黑暗的金属槽内，打在铝制易拉罐的金属表面上，拉出冷硬的高光反射，与外面的温暖形成质感对比。[视觉风格/胶片质感] 高对比度，冷硬的金属质感。[技术参数：100mm微距镜头，f/4.0光圈，高速快门捕捉瞬间] ▼ 视频动作提示词： [轻微的震动镜头（Camera Shake）] 镜头在罐子砸下的瞬间产生一丝符合物理规律的震颤。[环境物理动态：一罐饮料从黑暗的上方猛烈地掉落，重重地砸在金属底板上，塑料挡板被撞击得剧烈晃动，罐子表面瞬间凝结出的微小冰冷水珠在震动中飞溅出几滴] [音效与音乐氛围描述] 极其突兀、巨大且沉闷的金属碰撞声“咣当——”，在寂静的夜里回荡，吓人一跳。[对话台词与语气] (内心独白，带着一丝被打断后的回过神来) “罐子掉下来的声音，比想象中响。” [时长: 3.0秒] 镜头11（4s） 剧情：小初弯腰拿出冰冷的饮料罐 景别：中景 (Medium Shot) 角色动作：弯下腰，伸进取货口拿出饮料 情绪：身体上的冷感反馈 场景标签：弯腰，夜晚街道，自动贩卖机底部 光影氛围：低处光线较暗，背部受光 音效：塑料挡板的摩擦声，拿取罐子的声音 台词：冰得有点过分了， 角色【小初】：[小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] ▼ 图片生成提示词： [画面构图：中景，略带俯视的角度] [小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙] 弯下腰的背影与侧脸。[人物空间与互动关系：身体前倾，右手伸进画面底部的贩卖机取货口中] [极具体的微表情与面部特写] 侧脸在阴影中，手指刚触碰到罐子时，因为寒冷肩膀本能地微微瑟缩了一下。[明确的场景环境元素与前景/背景道具] 贩卖机的下半部分，湿润的地面。[光影几何与大气效果] 头顶的暖光打在她柔软的针织衫背部，形成一圈温暖的轮廓光（Rim Light），而她探入取货口的手部则处于深邃的阴影中。[视觉风格/胶片质感] 画面充满日常的真实感与生活气息，暗部略带胶片绿。[技术参数：35mm镜头，f/2.0光圈] ▼ 视频动作提示词： [缓慢的下摇镜头（Tilt Down）] 摄影机随着小初弯腰的动作缓慢向下摇动。[主体极其具体的物理动作细节与微表情变化：小初弯下腰，右手推开透明的塑料挡板，指尖刚触碰到易拉罐时，因为过低的温度动作停顿了半秒，手指肌肉微微收缩，随后将带着一层白霜的罐子紧紧握住，从阴影中拿了出来] [环境物理动态：随着她弯腰，长裙的下摆在脚踝处轻轻堆叠] [音效与音乐氛围描述] 塑料挡板被推开的“嘎吱”声，以及指尖摩擦金属罐的声音。[对话台词与语气] (内心独白，带着一点对温度的娇嗔感) “冰得有点过分了，” [时长: 4.0秒] 镜头12（4s） 剧情：双手握着冰冷的罐子，贴在脸颊旁 景别：特写 (Close-up) 角色动作：拿着罐子，轻轻贴在脸颊，看向远方 情绪：对小东西的共情，回归自我世界 场景标签：脸颊贴罐子，水汽，暖光面部 光影氛围：柔和暖光，罐子表面泛着冷霜 音效：吉他尾音缭绕，夜风 台词：等一下再喝吧。 角色【小初】：[小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫和棉质长裙，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] ▼ 图片生成提示词： [画面构图：面部与手部的双重特写，平视机位] [小初: 20岁出头的年轻亚洲女性，日系小清新风格，微卷的中长发随意披散，几缕发丝垂在脸颊，穿着宽松柔软的米色针织开衫，眼神温柔且带有一丝游离感，五官清秀，未施粉黛，表情放松且微微有些出神，整体散发出一种安静、文艺、内敛的气质] 的面部特写。[人物空间与互动关系：她双手捧着一罐绿色的日系饮料，罐子极其靠近她的左侧脸颊] [极具体的微表情与面部特写] 眼神重新变得漫无目的，望向画外的黑夜，嘴角带着一丝安分的、自我满足的微小弧度。[明确的场景环境元素与前景/背景道具] 铝制饮料罐表面布满了细密凝结的白色冰霜与水珠。[光影几何与大气效果] 贩卖机的琥珀色暖光将她的脸颊照得极其柔和透明，而手中冰冷的罐子则反射着微弱的冷光，冰与火的温度感在视觉上交融。[视觉风格/胶片质感] 极致的唯美电影感，柔焦效果，治愈系。[技术参数：85mm镜头，f/1.2光圈，完美的眼部对焦] ▼ 视频动作提示词： [极其缓慢的环绕微推镜头] 镜头以极其微小的幅度围绕角色的侧脸做环绕并缓慢推进。[主体极其具体的物理动作细节与微表情变化：小初双手握着那罐冰冷的饮料，像是对待一个小生命般，缓慢地将罐子冰凉的表面轻轻贴了一下自己的侧脸颊，她舒服地微微眯了一下眼睛，随后视线放空，完全沉浸在了自己的节奏里，将手机的震动彻底抛在脑后] [环境物理动态：易拉罐表面的水珠因为体温开始缓慢滑落一条水痕] [音效与音乐氛围描述] 音乐的吉他扫弦渐渐转为一声悠长的泛音，随后慢慢淡出在风声中。[对话台词与语气] (内心独白，轻盈、不下定论的语气) “等一下再喝吧。” [时长: 4.0秒]

参考图人物形象，和以下参考提示词结构，不用精简，帮我重新生成一段 12宫格分镜提示词。

【生成内容】 - 光：自动贩卖机冷白光柱 + 街灯钠黄低照度，写实纪录片质感（必写 naturalistic photographic light, NOT plastic/CG） - 关键节拍：第 7 镜 LINE 语音条「起きてる？」到达 → 第 11 镜没拿饮料离开（→ S4 冰箱门埋点） - 比例 16:9，silent video，白线有线耳机全程戴 - 通用 negative：sighing / parted lips / exhaling / plastic lighting / over-ear headphones / bare legs / 路灯 halo --- 镜 1 · 街角远景空镜（4s） - 景别：极远景，固定机位 - 内容：深夜街角，自动贩卖机独亮，背景便利店招牌虚化，远处一辆出租车驶过 - 起始帧：纯空镜，无人 - 节奏：0-2s 静止，2-4s 她从画面右侧远远走入，背影小到只剩剪影 - Kling 提示：极慢，camera locked，不要推拉 镜 2 · 中景背影（3s） - 景别：腰以上中景，跟拍背影 - 内容：她走到贩卖机前停下，头微低看商品 - 起始帧：背影距贩卖机 2 步 - 动作：自然走两步停住，肩线放松 - 注意：耳机白线垂在颈侧 镜 3 · 中景脸（贩卖机光打脸）（4s） - 景别：胸上中景，平视 - 内容：贩卖机冷白光从画外左打来，半边脸亮半边脸暗，她看商品 - 起始帧：脸已被光打亮，眼神专注 - 动作：极小幅度——目光从上排移到下排 - Negative 强化：no blinking-heavy, no head turn 镜 4 · 手部特写 · 投币（3s） - 景别：手部特写，俯视斜 30° - 内容：硬币投入投币口 - 起始帧：拇指食指捏一枚硬币悬停在投币口上方 - 动作：投入 → 听见硬币滚落（声音后期） - 喂图：故事板大图第三位，造型板黑吊带袖口 镜 5 · 中景 · 选项前犹豫（5s） - 景别：腰以上中景，平视 - 内容：手抬起准备按按钮，又放下，再抬起 - 起始帧：手悬停在按钮区前 5cm - 动作：抬→停→放下→再抬，节奏不规律（INFP+天秤的犹豫） - Negative：no facial expression change 镜 6 · 手指悬停按钮（3s） - 景别：手指特写 - 内容：食指悬停在一颗按钮上方 1cm，不按下 - 起始帧：指尖刚到位 - 动作：纯静止，3s 内只有极轻微呼吸带来的指尖颤 - 这是节拍点——为镜 7 LINE 到来做空白 镜 7 · 手机屏幕特写 · LINE 气泡（3s） - 景别：手机屏幕占画面 70%，斜俯视 - 内容：锁屏状态，LINE 灰色气泡浮出「起きてる？」 + 1 秒语音条波形 - 起始帧：黑屏 - 动作：0-0.5s 黑屏 → 屏幕亮起 → 气泡浮出 → 静止 - 注意：男主名字打码/模糊，时间显示 02:47 之类深夜数字 镜 8 · 脸部反应（4s） - 景别：胸上特写，平视 - 内容：她低头看了一眼手机，又抬头看回贩卖机 - 起始帧：低头中 - 动作：低头停 1s → 慢慢抬起 → 看回选项，没什么表情 - Negative 关键：no smile / no frown / no eye-roll，就是没反应——这是傲娇内核 镜 9 · 手部 · 按了另一个按钮（3s） - 景别：手指特写 - 内容：手指从原本悬停的按钮移开，按了旁边一颗 - 起始帧：手指还在原按钮上方 - 动作：横移 → 按下 → 收回 - 这是"画面信号 S2 按不同按钮"的兑现点 镜 10 · 饮料掉落取物口（3s） - 景别：取物口特写，俯视 - 内容：一瓶饮料咣当掉下来（具体是什么饮料锁死，S4 冰箱门要复用） - 起始帧：空取物口 - 动作：饮料从上方落入 → 弹一下 → 静止 - 建议饮料：罐装柠檬气泡水或牛奶咖啡，颜色辨识度高（你定） 镜 11 · 她转身离开（4s） - 景别：腰以上中景跟拍 → 推到背影 - 内容：她直起身，看了一眼取物口，转身走了，没拿 - 起始帧：刚直起腰 - 动作：低头瞥取物口 0.5s → 转身 → 慢步离开 → 走出画面 - 注意：动作要松不要决绝，是"算了"不是"不要了" 镜 12 · 取物口里那瓶饮料（4s） - 景别：取物口特写，俯视固定 - 内容：取物口里安静躺着那瓶饮料，贩卖机光柱继续亮着 - 起始帧：饮料静止 - 动作：纯静止，背景偶尔一辆车的灯光扫过 - 埋点：这瓶饮料明确是 S4 冰箱门里出现的同一款，颜色/品牌锁死

参考图人物形象，和以下参考提示词结构，不用精简，帮我重新生成一段 9宫格分镜图

**正向：** > 海边小站台的极远景空镜。单条铁轨贴海岸延伸到远端，海平线在画面后景，几缕薄云缓慢漂动。海面有微光反射，水波极轻晃动。整体偏冷蓝灰白。镜头极慢推近 2%，节奏非常静。 **负向（追加）：** no train entering, no people, no birds flock, no camera shake

**正向：** > 一列绿色江之电从画面深处沿铁轨缓慢朝镜头方向行驶，速度均匀不加快。车头反光带一抹高光，车窗一格一格闪过。背景海面冷蓝，云缓慢移动。镜头完全固定，不跟拍。整体偏冷调。 **负向（追加）：** no train stopping, no doors opening, no passengers visible, no zoom, no pan

**正向：** > 一双白色帆布鞋踩在木质站台地板上，左脚先踏，右脚跟上，再停一拍。木板纹理清晰，鞋底擦过木纹。背景虚焦是站台栏杆和海。镜头固定，地面散射光，鞋面带一抹暖反光。 **负向（追加）：** no running, no jumping, no stumbling, no full body shown, no face shown

**正向：** > 木长椅上一只灰白虎斑猫蜷成一团在睡觉，毛被海风轻轻吹动一下，没有醒。猫保持闭眼睡姿。背景虚焦是站台栏杆和海平线，海面冷蓝。猫身上有一抹暖白光。镜头完全固定。 **负向（追加）：** no orange cat, no ginger cat, no cat waking up, no cat standing, no cat moving away, no person entering frame

**正向：** > 中景背影。她从画面左侧入画，沿站台木板缓慢走向长椅，到长椅旁停下一拍，肩膀微微放松，视线落在长椅上的猫的方向。猫继续睡，没有反应。镜头极慢跟拍 1 步幅后停住。背光，人物轮廓亮，正面在阴影里。 **负向（追加）：** no fast walking, no waving, no talking, no cat reaction, no head turning sharply, no eye movement

**正向：** > 中景侧面 3/4。她在长椅左端缓慢坐下，离睡着的猫留一段距离。坐下后肩膀松下来，视线朝海的方向。猫保持闭眼睡姿不动。镜头完全固定。侧逆光勾肩颈线，海面在画面深处虚焦冷蓝。 **负向（追加）：** no leg crossing animation jitter, no cat moving, no head turning, no looking at camera, no eye movement

**正向：** > 脸近景侧面 3/4。海风吹起她的鬓发和衣领一角，发丝缓慢飘动，鼻尖和唇有极轻微的呼吸感。她半垂眼保持静止凝视，方向不变。海面反光打在脸的一侧，另一侧自然阴影。镜头极轻微推近 2%。 **负向（追加）：** no eye movement, no blinking sequence, no head turning, no smile, no mouth opening, no looking at camera

**正向：** > 手部+长椅木板的近景。她的手从画面下方升起来，掌心握着一部手机，**手机屏幕始终朝下**，手缓慢稳定地把手机扣在长椅木板上，扣下后手指在手机上停一拍，再缓慢离开画面。整个动作一气呵成，不停顿不分段。木板纹理清晰，手背一抹暖光。镜头完全固定。 **负向（追加）：** no screen showing, no phone face up, no phone glowing, no notification light, no flipping phone, no shaking, no multiple steps motion, no hand returning, no face entering frame

**正向：** > 极远景全景。木长椅在画面中段，她坐在长椅左端，猫在右端，铁轨贴海岸延伸，海平线和云在远端。一列绿色江之电从远端缓慢驶离画面深处。她和猫保持静止。镜头极慢拉远 6%，整体冷蓝调加海面散射反光。 **负向（追加）：** no fade out, no black screen, no zoom snap, no people walking past, no camera shake, no train stopping, no head turning

Positive： Medium shot, young Japanese woman in a black camisole top and wide-leg blue jeans standing in front of a lit vending machine at night, white wired earphones in both ears. She has been facing one product button, then her hand drifts laterally a short distance to the right and presses a different button beside it — the button she did not originally aim at. The press is light, almost casual, fingertip soft. Camera holds locked, no push or pan. Her face stays still throughout, expression unchanged, eyes following her own hand briefly then settling forward. Cold white vending machine panel light spills across her arm and the row of buttons, the pressed button glows faintly brighter as it activates. Sodium street lamp warm rim from behind. Shallow depth of field f1.8, cinematic film grain, 16:9, silent video, 5 seconds. Negative： no smile, no frown, no eye-roll, smirking, satisfied expression, decisive jab, hard press, slamming button, hesitation tremor, hand shaking, fingers morphing, extra fingers, deformed hands, multiple button presses, pressing original button, looking at camera, head turning sharply, body twisting, plastic lighting, over-ear headphones, bare legs, jitter, flicker, watermark, text overlay

Positive： Medium shot, young Japanese woman standing in front of a lit vending machine at night, white wired earphones in both ears, cable trailing down toward her pocket. She has been facing one product button, then her hand drifts laterally a short distance to the right and presses a different button beside it with a soft fingertip — the button she did not originally aim at. The press is light, almost casual. Camera locked, no push or pan. Her face stays still, expression unchanged, eyes following her own hand briefly then settling forward. Cold white vending machine panel light spills across her arm and the row of buttons, the pressed button glows faintly brighter as it activates. Sodium street lamp warm rim from behind. Shallow depth of field f1.8, cinematic film grain, 16:9, silent video, 5 seconds. Negative： no smile, no frown, no eye-roll, smirking, satisfied expression, decisive jab, hard press, slamming button, hesitation tremor, hand shaking, fingers morphing, extra fingers, deformed hands, multiple button presses, pressing original button, looking at camera, head turning sharply, body twisting, outfit change, clothing morphing, color shift on clothing, fabric distortion, plastic lighting, over-ear headphones, jitter, flicker, watermark, text overlay

Positive： Medium shot, young Japanese woman in a black camisole top and wide-leg blue jeans standing in front of a lit vending machine on a quiet city street at night. Long straight black hair with blunt bangs, white wired earphones in both ears, cable trailing into her pocket. She is mid-hesitation in front of the coin slot, one hand resting near the machine, the other holding a phone loosely at her side. Her face stays completely still, expression locked, eyes fixed steadily on the machine, no micro-expression change. Only subtle natural breathing and faint hair sway from night air. Cold white light from the vending machine front panel, sodium street lamp warm fill from behind, shallow depth of field f1.4, cinematic film grain, 16:9, silent video, 5 seconds. Negative： no facial expression change, eyes moving, eyeballs rolling, gaze shifting, pupils drifting, smiling, frowning, parted lips, sighing, exhaling, head tilting, face morphing, identity drift, extra fingers, deformed hands, blurry face, plastic lighting, over-ear headphones, bare legs, jitter, flicker, watermark, text overlay

【生成内容】 全场基调 - 比例 16:9 cinematic - 主光：白天海边自然光，整体偏冷蓝灰 + 一点点暖 - 镜头语言：以静镜+空镜为主，节奏比 S6 更慢，留白更多 - 造型：按已定稿（不写进 prompt，交给 ref 图锁） - silent video（音轨后期：海浪 / 电车进站 / 风 / 木板那一声） - 大光圈电影质感保留，但比夜场柔，眩光控制在弱 --- ## 镜 1 · 远景定场 - **景别**：极远景（站台+铁轨+海平线） - **时长**：4–5s - **画面**：江之电小站台空着，单条铁轨贴海岸延伸，远端是海平线和云。无人。 - **镜头运动**：固定（或极慢推近 2%） - **光线**：午后偏冷的白光，海面反光散点 - **备注**：开场空镜，节奏一定要慢，让 S6 烟花余韵彻底沉下来 ## 镜 2 · 电车驶过 - **景别**：中远景（站台侧面 + 电车） - **时长**：4s - **画面**：一列绿色江之电从画面右侧入画驶过站台，车窗反光闪一下，再缓缓出画。站台仍空着没人下车。 - **镜头运动**：固定（电车自带横向运动） - **光线**：电车反光带一抹高光，背景海面冷调 - **备注**：呼应独白"以前坐过一次，没下车"——这镜是过去那次的回响，不是她下车的镜 ## 镜 3 · 站台脚步 - **景别**：脚部+木板特写 - **时长**：3s - **画面**：白色帆布鞋踩上站台木板，一步，再一步。木板纹理清晰，鞋底擦过木纹。 - **镜头运动**：固定 - **光线**：地面散射光，鞋面有一抹暖反光 - **备注**：不给脸，先给"她到了"的物质感。木板的质感要清楚，为镜 8 扣手机的"那一声"铺触感 ## 镜 4 · 长椅上的猫 - **景别**：中近景（长椅+睡着的三花/灰猫） - **时长**：4s - **画面**：木长椅上一只猫蜷成一团在睡，毛随海风轻轻动一下，没有醒。背景虚焦是站台栏杆和海。 - **镜头运动**：固定 - **光线**：猫身上一抹暖白光，背景冷蓝海 - **备注**：回扣 S1 橘猫（\[\[project_2026_05_24_mv2_progress\]\] 勾连物件）。注意：这只猫不是橘猫，毛色换一只——独\ 白说"跟那只一点都不像" ## 镜 5 · 她走近停步 - **景别**：中景（背影+长椅+猫） - **时长**：3–4s - **画面**：她从画面左侧入画，走到长椅旁停一拍，看了一眼猫，没打扰。 - **镜头运动**：极慢跟拍（1 步幅内停住） - **光线**：背光，人物轮廓亮，正面在阴影里 - **备注**：脚步要轻，"不会理我"那种默契感靠停步那一下传 ## 镜 6 · 坐下 - **景别**：中景（侧面 3/4，她+猫同框） - **时长**：3s - **画面**：她在长椅另一端坐下，离猫留一段距离，肩膀松下来。猫没动。 - **镜头运动**：固定 - **光线**：侧逆光勾肩颈线，海面在画面深处虚焦 - **备注**：人和猫一同框就是关系——"都是一样的，不会理我" ## 镜 7 · 风从海那边 - **景别**：脸近景（侧面 3/4） - **时长**：4s - **画面**：风吹起鬓发和衣领一角，她半垂眼，没看任何具体的东西。鼻尖、唇有一点点动。 - **镜头运动**：固定（或极轻微推近 2%） - **光线**：海面反光打在脸的一侧，另一侧自然阴影 - **备注**：对应"风从海那边过来，有一点点咸"。眼神不写动（\[\[feedback_kling_no_eye_movement\]\]），靠起始帧锁住"半\ 垂"那个状态 ## 镜 8 · 扣手机（核心镜） - **景别**：手部+长椅木板的近景 - **时长**：4–5s - **画面**：她的手从兜里掏出手机，手机屏朝下，往长椅木板上扣下去——一气呵成，不停顿。手指离开手机后停一拍。 - **镜头运动**：固定 - **光线**：手背一抹暖光，木板纹理清晰 - **备注**：全片隐线终点。一定一气呵成，不要写成"掏出→翻过来→放下"三段（\[\[feedback_kling_i2v_advanced_pitfalls\]\ \] 时序别分步）。i2v prompt 里要锁"屏幕朝下"这一点，否则 Kling 容易给你拍成正面屏 ## 镜 9 · 终镜拉远 - **景别**：极远景（站台全景） - **时长**：5s - **画面**：长椅、她、猫、铁轨、海平线全在画里，她占画面的一小块。一列电车从远端入画驶来或驶离（任选其一，看哪\ 条更松）。 - **镜头运动**：极慢拉远（pull out 5–8%） - **光线**：全景冷调 + 海面反光 - **备注**：收全片。不要用 fade out，让画面自己留住——独白"听不到震动了"已经收过尾，画面再煽就毁了

参考图人物形象，和以下参考提示词结构，不用精简，帮我重新生成一段 12宫格分镜图

Cinematic 16:9, locked symmetrical wide shot down the center aisle of a moving late-night Japanese commuter train. Pink upholstered bench seats line both sides. On the left bench sits a young woman with short black bob hair and blunt fringe, wearing white over-ear headphones, a black leather jacket, a blue striped collared shirt, a short black skirt with bare legs, black ankle boots, a black shoulder bag beside her. She is absorbed in looking down at the seat in front of her, eyes soft. A man in a dark suit and hat sits on the right bench in the middle distance. A couple of other figures further down the carriage. Overhead hand straps hang in a row. Warm-white LED ceiling lights glow steadily. Photoreal, quiet, late-night atmosphere, shallow grain.

Prompt： ▎ Create a single starting frame: an intimate tight side profile chest-up close-up of a young Japanese woman, framed from her left side. Her face is in clean true profile — pure side view, looking softly forward and slightly downward at something off-frame in front of her (the ema plaque she stands before). Eyes softly cast down, lids gentle, lashes long and clear in profile silhouette. Lips closed, relaxed, the corner of her mouth neither up nor down — pure quiet, no expression — neither sad nor happy, just present and still. ▎ ▎ Her facial features must exactly match reference 1 in profile — same nose line and tip, same lip line, same chin shape, same single-eyelid eye in profile, same youthful Japanese face. Short black bob with blunt straight bangs, fine wisps of hair at her temple and near her ear, the line from her ear down her jaw to her chin clean and visible. ▎ ▎ She wears the outfit from reference 2: black thin-strap sleeveless tank top, the strap visible on the near shoulder, the white wired earphone cable trailing down from her ear into the lower frame. ▎ ▎ Lighting is the soul of this shot: warm afternoon sunlight from the front-right at about 30 degrees (from in front of her, the direction she is looking), softly tracing her bangs at the forehead, the bridge of her nose, the curve of her upper lip, the tip of her chin, and the outer rim of her ear with delicate warm light. The back of her cheek and her jawline fall into gentle warm shade — no harsh edges, no raccoon eye shadow. ▎ ▎ Background heavily soft-focused into a warm dreamy pale gold bokeh — the ema wall behind her dissolved into out-of-focus warm wood tones and soft warm light spots, almost abstract. The atmosphere matches reference 3. ▎ ▎ Daytime mid-afternoon, warm soft sunlight. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained closing tone, soft contrast, slightly desaturated except the warm bokeh, shallow depth of field — face crisp, background dreamy gold. ▎ ▎ 16:9 cinematic aspect ratio. gative： face toward camera, three-quarter angle, fa, eye contact, looking up, head tilt up,smiling, half smile, parted lips, sighing, raised eyebrow, frowning, contemplative dramatic, lip bite, mouth opening, surprised, hand to face, hair in tucking gesture, harsh side rim light,raccoon eye shadow, blown highlights on forehead, ring light look, flat backlight, sharp focus background, no bokeh, dramatic black background, modern indoor , multiple people in frame, modified outfit,missing earphone cable, oversaturated colors, plastic CG lighting, glow halo, lens flare, anime style, illustration, ▎ painterly, vintage filter, cartoon

Prompt： ▎ Create a single starting frame: a medium wide shot of a densely packed wall of Japanese ema (絵馬) votive wooden plaques hanging on a traditional wooden ema rack at a Japanese shrine. The frame is filled almost entirely with hundreds of small pentagonal (house-shape, 5-sided) wooden plaques densely covering a horizontal wooden frame structure, each plaque tied with a red or white cord and hanging at slightly different orientations — some facing forward showing the back-side wood, others rotated showing the front-side with faded handwritten ink wishes and small painted illustrations. The plaques overlap each other in dense layers. ▎ ▎ The wood plaques have varied warm aged wood colors — light tan, weathered honey-brown, faded oak — with worn natural texture. Some plaques have small painted images (zodiac animals, simple shrine motifs). Handwritten ink calligraphy is partly visible but not legible / not readable in detail. Wooden frame beams partially visible at top and bottom of frame. ▎ ▎ Behind the rack, the soft-focused background reveals a hint of an old dark brown weathered shrine wooden building wall and a glimpse of cedar foliage edge. Side-top warm afternoon sunlight comes in from the upper-right at about 30 degrees, lighting the plaques with a warm dappled glow and casting soft criss-crossing shadows of the wooden frame across them. ▎ ▎ No people in this frame. ▎ ▎ Daytime mid-afternoon. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated warm wood tones, shallow depth of field with the front plaques sharp. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ people appearing, person in frame, hand reaching, hand touching plaque, modern plastic plaque, glossy fresh plaque, brand new wood, smooth perfect wood, English text on plaques, modern font, neon lighting, fluorescent lighting, flat lighting, frontal flat light, oversaturat, painterly, vintage filter, cartoon, sparseplaques, empty rack, mostly empty wall

Prompt： ▎ Create a single starting frame: a close-up shot of three or four Japanese ema wooden plaques hanging from a traditional wooden frame, framing fills most of the composition. The plaques are pentagonal house-shape, tied with thin red or white cords to a horizontal wooden beam at the top of frame. The plaques are caught mid-sway, hanging at slightly varied angles as if just stirred by a gentle breeze — one tilted a few degrees right, another slightly left, the cords showing minimal slack. ▎ ▎ Each plaque shows warm aged wood texture — weathered honey-brown, light tan, faded oak — with traces of partly visible handwritten ink wishes on their fronts (not legible in detail). One plaque shows a small simple painted illustration (a shrine motif). The wooden cord knots at the top are tidy and traditional. ▎ ▎ Behind the plaques in soft-focus background: more dense plaques receding into bokeh, hints of warm wood frame structure. ▎ ▎ Side-top warm afternoon sunlight from the upper-right at about 30 degrees, lighting the wooden surfaces with a soft warm glow, casting gentle moving shadows from the cords down across the plaques. Atmosphere matches reference 1 in cinematic feel. ▎ ▎ No people in this frame. No hand visible. ▎ ▎ Daytime mid-afternoon. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained intimate tone, soft contrast, slightly desaturated warm wood tones, shallow depth of field on the three foreground plaques. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ people appearing, hand visible, hand touch swinging, dramatic wind, blown plaques,plaques flying off, broken cords, fallen plaques, modern plastic plaque, glossy fresh plaque, brand new wood, smooth perfect wood, English text, modern font, lat lighting, frontal flat light, sharp focus background, no bokeh, oversaturated, anime style, illustration, painterly, vintage filter, cartoon, completely ▎ still rigid plaques

Prompt： ▎ Create a single starting frame: a chest-up frontal medium shot of a young Japanese woman, eye-level framing, slightly off-center to the right of the composition. She is not facing the camera directly — her gaze and head are gently angled to the left (her right) by about 10-15 degrees, looking softly at something off-frame to the left (the ema wall, off-camera). Her expression is completely neutral, calm, quiet — not sad, not curious, not contemplative in a dramatic way, just simply not reacting. Eyes softly open, lips closed but relaxed. ▎ ▎ Her facial features must exactly match reference 1 — same single-eyelid eyes, same nose, same lip line, same youthful Japanese face. Short black bob with blunt straight bangs sitting cleanly on her brow. Skin texture natural, smooth, slightly luminous in afternoon light. ▎ ▎ She wears the outfit from reference 2: black thin-strap sleeveless tank top, light blue denim crossbody bag strap crossing her shoulder visible, white wired earphone cable trailing down from her ear along her neck on both sides. ▎ ▎ Lighting: soft warm afternoon sunlight comes from the front-left at about 30 degrees, softly washing her face with even, gentle, beautiful light — no harsh shadows, no raccoon eye shadow under her brow, no overexposed forehead. A subtle warm rim-light from the back highlights the edge of her hair and ear faintly. ▎ ▎ Background is heavily soft-focused — warm out-of-focus blurred wooden shapes suggesting the ema wall (绘馬) with hints of brown wooden plaques and a faint hint of green cedar foliage further behind. Background atmosphere matches reference 3 in color tone. ▎ ▎ Daytime mid-afternoon, soft warm sunlight. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, shallow depth of field on her face — face crisp, background dreamy soft. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ looking at camera, eye contact, full frontera, smiling, half smile, parted lips,sighing, raised eyebrow, frowning, sad expression, contemplative dramatic, lip bite, mouth opening, surprised, hand to face, hair in face, ear-tucking gesture, three-quarter back, harsh front sunlight,raccoon eye shadow, blown highlights on forehead, ring light look, flat backlight, sharp focus background, no bokeh, modern indoor background, fluorescent, mu modified outfit, missing earphone cable,oversaturated colors, plastic CG lighting, glow halo, lens flare, anime style, illustration, painterly, vintage filter, cartoon

Prompt： ▎ Create a single starting frame: a medium-long shot of a young Japanese woman standing in front of an ema (絵馬) wall at a Japanese shrine, seen from a clean three-quarter back angle, the camera positioned behind-right of her. She stands relaxed and quiet, her right arm gently lifted, fingertips of her right hand softly touching the lower edge of one of the ema plaques in front of her — not gripping, not lifting, just barely brushing the wood. Her left arm hangs relaxed at her side. Posture soft, shoulders slightly forward, head slightly inclined down looking at the plaque she touches. ▎ ▎ She wears the outfit from reference 2: black thin-strap sleeveless tank top exposing her shoulders, upper back, and neck; loose baggy washed light blue jeans low on her hips; the light blue denim crossbody bag strap crossing diagonally from her right shoulder down to her left hip; white wired earphone cable trailing down from her ears along the back of her neck and both shoulders. ▎ ▎ Her hair matches reference 3 — short black bob with blunt straight bangs, the back of her hair cleanly tapered, fine wisps at the nape of her neck. Skin texture natural and smooth. ▎ ▎ Lighting is critical: warm afternoon sunlight comes from the upper-right at about 30 degrees, rim-lighting the curve of her neck, the line of her shoulder, the side of her waist, and tracing a soft warm edge along her arm raised toward the plaques. The light catches her exposed nape and the wisps of hair there, and the side of her cheek (partially visible due to the back angle). The shadow side of her body falls in soft gentle shade — no harsh edges. ▎ ▎ Behind/around her, the ema wall fills the soft-focused background — many pentagonal wooden plaques in warm honey-brown and tan, slightly out of focus. The atmosphere matches reference 1's warm dappled afternoon feel. ▎ ▎ Daytime mid-afternoon, warm side-back sunlight. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained tone, soft contrast, slightly desaturated, shallow depth of field on her body silhouette. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ face fully visible, full frontal, looking era, eye contact, smiling, plaque being taken down, plaque grabbed firmly, plaque pulled, hand holding plaque, two hands, hands clasped, hands on hips, dramatpose, ear-tucking gesture, hair covering nirt, modified outfit, missing earphone cable, missing crossbody bag, harsh front light, flat lighting, raccoon eyes, blown highlights, ring light look, sharp focus background, no bokeh, sparse plaquesodern apartment, fluorescent, multiplepeople, tourists, oversaturated colors, plastic CG lighting, glow halo, lens flare, anime style, illustration, ▎ painterly, vintage filter, cartoon

Prompt： ▎ Create a single starting frame: a low-angle upward-looking shot of a young Japanese woman standing under a bright vermilion red torii gate at a Japanese shrine. The torii's vermilion horizontal top crossbeam (笠木 and 島木) dominates the top third of the frame, framing a slice of cedar canopy and slivers of bright daytime sky above. She stands in the lower portion of the frame, tilting her head up to look at the torii crossbeam, gaze gentle and curious, expression completely neutral and quiet. ▎ ▎ Her facial features must exactly match reference 1 — same single-eyelid eyes, same nose, same lip line, same youthful Japanese face. Short black bob with blunt straight bangs. From the low upward camera angle, her under-jawline, throat, and lashes are emphasized. Sunlight coming from above the torii beam rims the upper edge of her face, top of her cheekbones and tip of her nose with warm light, while the lower half of her face falls into gentle shade. ▎ ▎ She wears the outfit from reference 2: black thin-strap sleeveless tank top, loose baggy light blue jeans visible at lower frame, white wired earphone cable trailing down from her ears alongside her neck. The light blue denim crossbody strap crosses her shoulder. ▎ ▎ The torii vermilion color must match reference 3 — vivid bright vermilion red, NOT faded, NOT dull. Lush green cedar leaves visible at the very edges of frame around the torii beam. ▎ ▎ Daytime mid-afternoon, dappled bright sky above. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained tone, soft contrast, slightly desaturated, hero face framing with shallow depth of field on the face. Negative： ▎ high angle, eye-level, looking down, top-down, face fully flat-lit, harsh flat lighting, smiling, parted lips, sighing, surprised look, mouth opening, eyes wide, eye contact with camera, face away from torii, neck strain, awkward neck angle, multiple people, tourists, monks, priests, faded torii, dull torii, modern lighting, fluorescent, ring light look, oversaturated colors, plastic CG lighting, glow halo, lens flare, anime style, illustration, painterly, vintage filter, overexposed face, raccoon eyes shadow, missing earphone cable, modified outfit

Prompt： ▎ Create a single starting frame: a medium-long shot of an elderly Japanese woman sweeping fallen leaves on a stone step pathway at a Japanese shrine. She is roughly 70-80 years old, wears traditional loose dark navy or grey work clothes — a simple cotton work jacket (作務衣 style) and dark pants, a thin headscarf or no head covering, white split-toe tabi-style canvas shoes. Hair is short and grey, tied back simply. She holds a long traditional bamboo broom (竹箒) and sweeps with calm small unhurried strokes, head slightly bowed, her face mostly hidden in profile or three-quarter back. Posture quiet, slow, patient — the way someone who has done this for decades. ▎ ▎ She is the only person in frame. Scattered yellow ginkgo leaves and a few red maple leaves rest on the stone steps around her — some clustered into a small pile she has swept together, others freshly fallen and still scattered. The shrine setting matches the atmosphere of reference 1 — stone step pathway, lush green cedar forest framing the path, dappled mid-afternoon sunlight filtering through the canopy onto the leaves and stone. Optional: a faded wooden shrine structure or stone lantern blurred in the soft-focus background behind her. ▎ ▎ No torii gates visible in this composition (this shot is the side path area of the shrine grounds, away from the main torii tunnel). The atmosphere — color, lighting, forest density, stone texture — must match the cinematic feel of reference 1. ▎ ▎ Daytime mid-afternoon, soft dappled sunlight. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational documentary tone, soft contrast, slightly desaturated greens and warm earth tones. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ young woman, main character, tourist, multiple people, group of people, monks, priests, modern broom, plastic broom, vacuum cleaner, smiling face, face fully toward camera, dramatic pose, action pose, fast motion blur, autumn forest fully red, snow, rain, overcast sky, fluorescent lighting, harsh midday sun, oversaturated colors, plastic CG lighting, glow halo, lens flare, anime style, illustration, painterly, vintage filter, cartoon, low quality

Prompt： ▎ Create a single starting frame: a medium close-up still life shot of a few yellow ginkgo leaves caught in mid-air, drifting and slowly rotating downward at a Japanese shrine grounds. Two to three bright golden-yellow ginkgo leaves are suspended at slightly varied heights and rotations in the upper-middle of the frame, rendered in sharp focus, each leaf showing its fan-shaped silhouette clearly. ▎ ▎ The background is heavily soft-focused — blurred shapes of an old dark wooden shrine structure (faded brown wood walls and a hint of a curved tile roof edge) layered with blurred lush green cedar foliage. Dappled mid-afternoon sunlight filters through, creating a soft warm bokeh of light spots across the soft-focus background. The cinematic atmosphere — color, lighting, forest density — must match the feel of reference 1. ▎ ▎ No people, no main character, no broom visible in this composition — pure still life of falling leaves against blurred shrine background. ▎ ▎ Daytime mid-afternoon, soft dappled warm sunlight catching the falling leaves from behind, making them glow translucent golden. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated greens against vivid translucent gold. ▎ ▎ 16:9 cinematic aspect ratio, shallow depth of field on the falling leaves, foreground sharp / background blurred. Negative： ▎ people appearing, person in frame, hand catching leaves, broom in frame, animal, bird, autumn forest fully red, all leaves on ground, no leaves in air, leaves on ground only, fast motion blur on leaves, leaves blowing horizontally, strong wind, leaves swirling chaotically, sharp focus background, no bokeh, flat lighting, overcast sky, fluorescent light, oversaturated, plastic CG lighting, anime style, illustration, painterly, vintage filter, cartoon, low quality

Prompt： ▎ Create a single starting frame: a medium shot of a young Japanese woman standing on the side edge of a shrine stone pathway, her body and face turned at a clean three-quarter side angle toward the left side of the frame (toward something off-frame), waist-up framing. She stands relaxed, weight on one leg, hands resting at her sides, head held softly straight — no head tilt. ▎ ▎ Her facial features must exactly match reference 1 — same single-eyelid eyes, same nose, same lip line. Short black bob with blunt straight bangs, wisps of hair behind her ear catching light from the side-back. Expression completely neutral and quiet, looking calmly at something off-frame, eyes neither sad nor curious. ▎ ▎ She wears the outfit from reference 2 exactly: black thin-strap sleeveless tank top, loose baggy light blue jeans visible at lower frame, light blue denim crossbody bag strap crossing her shoulder visible, white wired earphone cable trailing down from her ear along her neck. ▎ ▎ Lighting is the hero of this shot: warm afternoon sunlight comes from the side-back at about 45 degrees behind her, rim-lighting the edge of her ear, the tip of her bangs, the bridge of her nose, the curve of her cheekbone, and the strands of hair behind her ear with warm golden glow. The shadow side of her face (toward camera-front) is soft and gentle, not harsh. Skin texture natural, smooth, slightly luminous. ▎ ▎ Background heavily soft-focused: blurred shrine grounds — dappled green cedar forest and a hint of stone pathway, perhaps a blurred suggestion of the old grandma figure far in the background (very small and out of focus, only a hint, not detailed). The atmosphere matches reference 3. ▎ ▎ Daytime mid-afternoon, warm golden side-back sunlight. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained tone, soft contrast, slightly desaturated, shallow depth of field on her face. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ face fully toward camera, full frontal, looking at camera, eye contact, smiling, parted lips, sighing, raised eyebrow, head tilt, dramatic pose, hand toering ear, flat front lighting, harsh shadow, raccoon eyes, overexposed face, blown highlights, ring light look, fluorescent light, modern apartment background, sharp focus background, no bokeh, multiplewd, modified outfit, missing earphone cable,oversaturated colors, plastic CG lighting, glow halo, lens flare, anime style, illustration, painterly, vintage ▎ filter, cartoon

Prompt： ▎ Create a single starting frame: an extreme close-up of the right front pocket area of a young Japanese woman's loose baggy washed light blue jeans, framed at hip-level from the side-front. The composition focuses on the jeans pocket area, with a subtle phone-shaped rectangular bulge inside the pocket clearly visible against the loose denim fabric. ▎ ▎ Visible at the upper edge of the frame is the lower edge of her black thin-strap sleeveless tank top, the bottom hem grazing the denim waistband. Her right hand rests still and relaxed at her side, fingertips lightly visible at the edge of the frame near her hip, not touching the pocket. A small light blue denim crossbody bag strap dangles softly into the side of frame. ▎ ▎ The denim fabric must match reference 1 exactly — same washed light blue color, loose drape, slightly faded texture. ▎ ▎ Background is heavily soft-focused shrine grounds — blurred dappled green forest and warm stone pathway, the atmosphere matching reference 2. Faint warm afternoon sunlight catches the denim and her hand softly from the right side. ▎ ▎ Daytime mid-afternoon. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, shallow depth of field on the pocket and hand. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ face visible, full body, wide shot, hand inside pocket, hand reaching for pocket, hand holding phone, phone visible, phone screen visible, screen glowing, notg pocket, dramatic gesture, tight skinnyjeans, dark jeans, sharp focus background, flat lighting, modern interior, multiple people, oversaturated colors, plastic CG lighting, glow halo, lens flarepainterly, vintage filter, cartoon

Prompt： ▎ Create a single starting frame: a camera-behind-the-character mid-tunnel point of view inside the bright vermilion red torii tunnel from reference 1, framed as a waist-up medium-close shot of a young Japanese woman from directly behind, about to step under the next torii gate. Two massive bright vermilion red torii pillars dominate the left and right edges of the frame like a doorframe, the receding tunnel of more vermilion torii visible ahead of her down the stone step path. Dappled mid-afternoon sunlight filters through the cedar canopy above, scattering soft light spots across her shoulders, back, and the stone path ahead. ▎ ▎ She matches the outfit from reference 2 exactly: black thin-strap sleeveless tank top, loose baggy washed light blue jeans low-rise, white cotton ankle socks, white sneakers, a small light blue denim crossbody bag slung over her shoulder visible at her side, a white wired earphone cable trailing down from her ears along the back of her neck and behind her right shoulder. Hair matches reference 3: short black bob with blunt straight bangs, the back of her hair clean and slightly tapered. Posture relaxed, shoulders soft, mid-stride. ▎ ▎ The torii style and color must match reference 1 — vivid bright vermilion red, NOT faded, NOT dull. Lush green cedar forest barely visible past the torii pillars on either side. ▎ ▎ Daytime mid-afternoon, dappled sunlight on her back and shoulders. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast. ▎ ▎ 16:9 cinematic aspect ratio.

【生成内容】 # MV2 · S3 · 末班电车 · 夜 · 12 镜骨架 ## 段落定位 S3-S4 隐线起点：她进入自己节奏。全段不出现手机/语音。 ## 核心动作锚点 - 镜 7：傲娇核——瞥一眼上班族就移开（对位 S2 镜 8） - 镜 10-11：剧眼——摘耳机两连，MV2 第一次主动关掉声音 - 镜 12：桥到 S4，下车背影 --- ## 镜 1 · 末班车进站 - 内容：空站台，列车从右滑入，灯泛白 - 镜头：固定中远景 - 节拍：建场，无人 ## 镜 2 · 她上车 - 内容：女主从车门进入，背影/半侧 - 镜头：固定，地面+车门框 - 节拍：主角入场 ## 镜 3 · 空车厢全景 - 内容：长车厢稀疏几人，暖白 LED - 镜头：走廊中线对称固定镜 - 节拍：建立"少人" ## 镜 4 · 她坐下侧脸 - 内容：女主坐长椅，耳机在耳，视线放空 - 镜头：侧 45° 半身中景 - 节拍：静态情绪基调 ## 镜 5 · 对面上班族 - 内容：对面西装男靠窗睡着，头微歪 - 镜头：女主 POV 方向，中景 - 节拍：prop 出现 ## 镜 6 · 书包带松特写 - 内容：男人脚边书包，肩带没扣 - 镜头：低角度特写 - 节拍：细节放大 ## 镜 7 · 她瞥一眼 - 内容：女主看一秒，移开视线 - 镜头：侧脸特写，眼神微动一次 - 节拍：傲娇核——关心一秒就收 ## 镜 8 · 车窗外房屋 POV - 内容：透过车窗，住宅区一格格滑过 - 镜头：窗外横移空镜 - 节拍：外景节奏 ## 镜 9 · 远处一户亮灯 - 内容：黑楼里某一户窗户还亮着 - 镜头：窗外定焦，亮窗滑过画面 - 节拍：情绪点（"今天也很晚"） ## 镜 10 · 拔耳机的手 - 内容：到站，女主手指捻起耳机线 - 镜头：手部+耳侧近景 - 节拍：关键动作起手 ## 镜 11 · 耳机离耳特写 - 内容：耳机在手心，线垂下 - 镜头：手心+耳机微距 - 节拍：关键动作落点（"想听空气"） ## 镜 12 · 她下车背影 - 内容：车门打开，女主踏出站台 - 镜头：背影中景，站台风过 - 节拍：桥到 S4，隐线兑现

【生成内容】 - 光：午后斑驳日光透过古杉投下，神社朱红鸟居 + 杉木深绿 + 暖木建筑 + 绘马墙暖金，写实纪录片质感（必写 naturalistic photographic light, NOT plastic/CG） - 关键节拍：镜 7 口袋手机震 → 镜 8 她不拿（这场情绪锚点·必给）/ 镜 12 侧脸特写收尾（克制留白） - 比例 16:9，silent video，造型 Voice Diarist · night-walk solitude（黑细吊带+宽松浅蓝牛仔+白短袜+蓝牛仔斜挎小包+白线有线耳机全程戴） - 通用 negative：sighing / parted lips / exhaling / plastic lighting / eye movement / blinking-heavy / smile / head turn / look at camera / tourists / monks / priests / over-ear headphones / faded torii / dull torii / anime style --- 镜 1 · 鸟居远景建立（3s） - 景别：极远景，固定机位，朱红鸟居 tunnel 纵深 - 内容：朱红鸟居一座接一座深入森林，参道石板路居中，两侧古杉，夕照斑驳。她从画面深处石板路上走来，远端剪影感 - 起始帧：她在巷子最深处，约 1/8 帧高 - 节奏：0-1s 静止，1-3s 她朝镜头方向走两到三步，scale 不能变大 - 光：bright vivid vermilion red + 深绿杉林 + 斑驳午后日光 - 喂图：场景图第一位（锁构图），造型图第二位（锁服饰），脸图第三位 - Kling 提示：camera locked，零运动；ADD 模式 - Negative 强化：torii color fading / dull torii / walking backwards / reversed playback / figure becoming large quickly - 注意：鸟居红色必须复述三遍防 \*\*\*\*\*\*\*\*\*\* 自作主张褪色 镜 2 · 穿过鸟居·中近景跟拍（3s） - 景别：腰上中近景，从她正后方跟拍 - 内容：两根朱红鸟居粗柱在画面左右两侧充当门框，她正穿过第一座鸟居，光斑落在肩头和参道 - 起始帧：她已在两根柱中间，背对镜头，步入第二座 - 动作：自然爬步穿过，肩松，白线耳机在颈侧后晃 - 光：朱红柱填充左右 + 上方斑驳日光打在肩头和参道 - 喂图：场景图第一位（锁构图），造型图第二位，脸图第三位 - Kling 提示：背后极慢推进，幅度 5-10% - Negative 强化：face visible / looking backlling out - 注意：耳机白线必带 + 鸟居柱完全不能动 镜 3 · 鸟居下仰拍·抬头看（2s） - 景别：低角度仰拍，胸上构图 - 内容：鸟居顶横木（笠木）占据画面顶部三分之 +睫毛+脖颈线被光勾出 - 起始帧：头已经抬起，光打到下巴鼻梁鼻尖 - 动作：极小幅度——头停 1s 然后微微回正（回正 - 光：从鸟居顶上方打下来 → 半边鼻梁/颧骨镀暖金，下半脸柔阴 - 喂图：脸图第一位（锁脸·这镜脸是主角），造 锁鸟居红色和绿森氛围） - Kling 提示：camera locked - Negative 强化：face fully flat-lit / neck smiling / mouth opening - 注意：美点核心 = 下颌线+睫毛+脖颈，写死 镜 4 · 老奶奶扫落叶·中远景（2.5s） - 景别：中远景固定，参道侧路 - 内容：老奶奶 70-80 岁，作務衣黑灰色，竹箒 散落的银杏+枫叶在石阶上 - 起始帧：老奶奶低头扫地，背对/侧背，姿态缓慢 - 动作：竹箒前后慢扫两下，叶子轻动 - 光：午后柔光 + 暖木建筑作背景柔焦 - 喂图：单图——场景图（锁神社氛围+石板+森林） - Kling 提示：camera locked - Negative 强化：young woman / main charactebroom / smiling face / dramatic action / fast motion - 注意：这镜没主角，唯一人物锁死 镜 5 · 银杏落叶·飘下特写（2s） - 景别：中近景特写，竖向构图 - 内容：两到三片金黄银杏叶从画面上方旋转下落，前景锐焦，后景神社木结构+杉林柔焦 - 起始帧：叶子在画面上半部分悬停，高度错落 - 动作：每片叶慢慢旋转下落，整体节奏要漂浮，不能横飞 - 光：从后方逆光打过来 → 叶子半透明发金 - 喂图：单图——场景图（锁氛围） - Kling 提示：camera locked - Negative 强化：leaves swirling chaotically / strong wind / fast motion blur / leaves on ground only / no leaves in\ air - 注意：背景柔焦 + 叶子锐焦的对比 镜 6 · 她侧脸看老奶奶·中景（2.5s） - 景别：腰上中景，3/4 侧脸朝向画面左侧 - 内容：她站在参道边静静看，3/4 侧脸+耳后碎 - 起始帧：3/4 侧脸已成立，光打到耳后碎发 - 动作：极小幅度——发丝在风中轻拂，眼睑无动 - 光：从侧后 45° 打来 → 耳缘/鼻梁/颧骨/碎发被勾金边（这镜美的核心） - 喂图：脸图第一位，造型图第二位，场景图第三 - Kling 提示：camera locked - Negative 强化：face fully toward camera / covering ear / flat lighting / raccoon eyes - 注意：耳后碎发被光勾出来——这格的美点签名 镜 7 · 口袋手机震特写（1.5s） - 景别：胯位侧前极近景特写，只见牛仔口袋+手垂在身侧 - 内容：浅蓝宽松牛仔右口袋处手机形状凸起，她 微抖一瞬 - 起始帧：手机在口袋里静止，手在边 - 动作：0.3-0.6s 之间口袋极轻微一次震颤，手 - 光：侧光打在牛仔上，背景柔焦 - 喂图：造型图第一位（锁口袋牛仔细节），场景 - Kling 提示：camera locked - Negative 强化：hand inside pocket / hand rsible / phone screen visible / hand touchingpocket / dramatic gesture - 注意：手不能动——是后续镜 8 "她没反应"的因 镜 8 · 她的脸·没反应（3s · 必给 · 情绪锚点） - 景别：胸上正面中景，平视，视线偏画外左 10- - 内容：她正在看绘马（看画外左），完全没去拿手机。表情绝对中性平静，不是悲不是喜，不是冷不是淡，是"没听见"的那种平静 - 起始帧：视线已偏左，柔光在脸上 - 动作：极小幅度——胸口呼吸微抬一下，眼睑无动，绝不眨眼戏精化 - 光：前左 30° 柔光打满脸，无生硬阴影 / 无黑暖边光勾耳和发尾 - 喂图：脸图第一位（锁脸必须严），造型图第二位，场景图第三位 - Kling 提示：camera locked - Negative 强化：looking at camera / eye contact / smiling / half smile / parted lips / sighing / blinking heavy /\ surprised / contemplative dramatic / hand to - 注意：这镜出来如果脸有戏，必须重跑。这是傲娇内核——人 ping 她她不动 镜 9 · 绘马墙全景（2.5s） - 景别：中景固定，绘马墙占满画面 - 内容：上百个五角形（家屋型）木质绘马挂在木 着，背景暖木建筑墙+杉叶柔焦 - 起始帧：绘马墙密集填充画面 - 动作：极轻微——几片绘马在风中几不可察地颤一 - 光：右上 30° 暖光打在绘马上 → 木绳投下细影 - 喂图：单图——场景图（锁神社氛围） - Kling 提示：camera locked - Negative 强化：people appearing / hand reaes / modern plastic plaque / sparse plaques - 注意：绘马字必须看不清（避免文字翻车） 镜 10 · 绘马近景·风吹（2s） - 景别：近景特写，三四片绘马 - 内容：三四片绘马在微风中catch mid-sway，绳 感强） - 起始帧：绘马已经轻微偏角 - 动作：每片绘马极轻微摇 1-2 次，绝不能大幅 - 光：右上斜光打到木质表面+绳的细影 - 喂图：单图——场景图（锁氛围） - Kling 提示：camera locked - Negative 强化：plaques violently swinging es / plaques flying off / broken cords /fallen plaques / completely still rigid - 注意：风的"刚起"质感是关键——动得"几乎看不 镜 11 · 她站在绘马墙前·侧背中景（3s · 美点 · - 景别：中远景，3/4 侧背，相机在她后右方 - 内容：她侧背镜头站在绘马墙前，右手抬起指尖轻触一块绘马的下沿，左手垂身侧 - 起始帧：指尖刚到绘马下沿，姿态成立 - 动作：极小幅度——肩肉微动，发丝在风中轻动 1 下，手只是碰，不取下 - 光：右上 30° 暖光 → 颈线/肩线/腰线/抬起的 的核心） - 喂图：场景图第一位（锁绘马墙背景），造型图第二位，脸图第三位（锁后脑+发尾） - Kling 提示：camera locked - Negative 强化：face fully visible / full frontal / looking at camera / plaque grabbed firmly / plaque pulled / two\ hands / hair covering nape / harsh front lig - 注意：颈+裸肩+腰线的光勾边是这镜灵魂。手只碰不取——和镜 6/8 内核统一 镜 12 · 收尾·侧脸特写（3s · 克制美 · 全片留白） - 景别：胸上侧脸特写，纯正侧脸（不带 3/4 倾向） - 内容：眼睛微低看着面前绘马（在画外），lashsilhouette，下颌→嘴角→耳廓被光描出，背景绘马墙柔焦溶成暖金 bokeh - 起始帧：纯侧脸已经成立，光打到额发/鼻梁/嘴 - 动作：极小幅度——眼睑微微一颤（绝不眨），唇微闭无开合，胸口柔吸一下 - 光：前右 30° 暖光 → 额发/鼻梁/上唇/下巴/耳边 - 喂图：脸图第一位（锁脸必须严），造型图第二位，场景图第三位（柔焦暖金背景） - Kling 提示：camera locked - Negative 强化：face toward camera / three-quarter angle / looking up / smiling / parted lips / sighing / hair in face / ear-tucking / harsh side rim / raccoockground / dramatic black background

Macro close-up on hand, slightly overhead 30-degree angle. Frame opens with thumb and index finger pinching a single coin, hovering2just above thevending machine coin slot. Slow deliberate motion: coin drops into the slot, fingers release naturally, hand lingers for half a second then withdraws gently out of frame bottom. Black halter strap and cuff edge faintly visible at upper frame. Cold white machine light reflects on the metal slot, slight specular on the coin as it falls. Skin tone neutral, subtle film grain. Cinematic 16:9, shallow depth of field. Silent video. Negative： multiple coins, dropping coin, shaking hand, trembling fingers, jewelry, rings, nail polish chips, fast motion, sighing, parted lips, exhaling, plastic lighting, camera shake, zoom, push in, pull out, pan, two hands

Cinematic 16:9, locked static shot from inside a moving late-night Japanese commuter train, looking outward through the open carriage doorway. The metal doorframe is visible at the foreground left and right edges. A young woman stands just inside the train at the doorway threshold, her back fully to camera, wearing white over-ear headphones, a black leather jacket, a short black skirt with bare legs, black ankle boots, and a black shoulder bag hanging from her shoulder. Behind her through the open doors, a quiet station platform is visible. Inside the carriage, pink upholstered bench seats line both sides of the foreground, matching the other shots in this sequence. Cool washed-out fluorescent light. Photoreal late-night atmosphere.

under the headband, the two cushioned ear cups hanging softly downward. Hand emerging from a black leather jacket cuff. White soft-padded cups and slim metal headband catch cool fluorescent light. Background is the soft out-of-focus dark interior of a late-night train carriage with faint window reflections. Photoreal cinematic still, shallow depth of field, intimate quiet mood.

Positive: Cinematic 16:9, locked static medium shot from inside the train looking through the open carriage doorway. A young woman with short black bob hair, wearing white over-ear headphones, a black leather jacket and a black shoulder bag with subtle silver hardware, stands with her back to camera at the threshold of the train door, having just stepped onto the platform. She stands still for the first second, then takes one quiet step forward away from camera, then settles still again. Her head never turns. The platform extends ahead of her under cool white fluorescent lights, the platform floor slightly damp and reflective. A faint breeze just barely lifts the ends of her hair. Photoreal, quiet, late-night arrival atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, she turns her head, she looks back, head rotation, face reveal, profile reveal, she removes headphones, she touches headphones, she pulls out phone, hand reaching into pocket, she waves, she runs, fast walking, multiple steps in quick succession, she stops abruptly, train doors closing on her, doors sliding shut, doors slamming, other passengers exiting around her, passengers walking past her, station attendant appearing, train departing immediately, train pulling away with motion blur, station signs scrolling readable text, animated signs, color shift, warm orange tone, golden hour, daylight, sound, audio, music, dialogue, voiceover, footstep sound, ambient station sound

Positive： Extreme close-up on a smartphone screen held in a young woman's hand at night. The lock screen or LINE chat list is visible, showing the time "02:47" in large clean digits at the top. A recent contact row shows a blurred/pixelated name (mosaic obscuring the kanji), small avatar reduced to a soft blob, last message preview also blurred. Screen brightness low, soft blue-white glow lighting her fingertips. Background out of focus: faint cold vending machine light spill and warm street lamp bokeh. White earphone cable drifts along the edge of frame. Phone held perfectly still, only the faintest finger contact, no scrolling, no tapping. Shallow depth of field f1.8, cinematic film grain, 16:9, silent video, 5 seconds. Negative： no facial expression change, scrolling, tapping, typing, notification popup, message arriving, screen unlocking, finger swiping, readable contact name, readable message text, sharp text on chat row, hand trembling, fingers morphing, extra fingers, deformed hands, face visible, plastic lighting, over-ear headphones, jitter, flicker, watermark, text overlay (other than 02:47 timestamp)

Cinematic 16:9, intimate close-up from a three-quarter back angle. A young woman with short black bob hair and blunt fringe lifts both hands to her right ear, gently lifting a white over-ear headphone cushion away from her ear. Side of her jaw and ear in sharp focus, her downcast eyelashes just visible. Black leather jacket sleeves and a hint of blue striped shirt collar at frame bottom. Background is the soft out-of-focus interior of a late-night Japanese commuter train. Cool fluorescent light catches the white cushion and the strands of her hair. Photoreal cinematic still, shallow depth of field.

Cinematic 16:9, locked static medium shot from a three-quarter side angle, framed from the top of her head down to mid-thigh. A young Japanese woman with short black bob hair and blunt fringe sits on a pink upholstered train bench, wearing white over-ear headphones, a black leather jacket open over a blue and white striped collared shirt, a short black skirt, bare legs visible from mid-thigh down. Her face is in sharp focus, expression completely neutral, gaze cast softly downward into the empty space ahead, no smile, no frown. Hands rest gently in her lap. The dark train window behind her is rendered as a soft out-of-focus motion blur of the exterior night. The pink bench edge softly visible at the bottom of frame. Warm-cool fluorescent carriage light falls evenly across her face and her legs. Photoreal cinematic still, shallow depth of field, contemplative late-night quiet mood.

strands behind her ear lift gently in the breeze — only the smallest motion. Eyes do not move. Lips stay closed. Naturalistic photographic light. Silent video. 16:9. Negative: face fully toward camera, full frontal, head turning, looking at camera, hair covering ear, flat front lighting, raccoon eyes, harsh shadow, smiling, parted lips, eye movement, anime style.

Positive: Cinematic 16:9, locked static tight close-up of a young woman's face, three-quarter angle. She wears white over-ear headphones, a black leather jacket collar and a blue striped shirt collar visible at the bottom of frame. Her gaze is held softly off to her right side, eyes settled, expression completely neutral — no smile, no frown, no eye-roll. She holds this look throughout the shot with absolute stillness. Only the faintest natural breath, an almost imperceptible side-to-side body sway from the train's motion, and one slow natural blink. Her gaze direction does not change. Hair stays settled around her face. Background is the softly out-of-focus interior of the train carriage with a slight motion blur of the window beyond. Cool fluorescent carriage light evenly on her skin. Photoreal, contemplative, quiet atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, eyes darting, eye glance shift, gaze moving, looking around, looking back to center, looking down, looking up, pupil moving, eyeballs rolling, head turning, head rotation, chin lifting, chin lowering, she smiles, she frowns, she smirks, lip movement, lips parting, mouth opening, eyebrow raising, eyebrows knitting, expression change, sudden emotion, she nods, she shakes head, hair flying, hand entering frame, hand touching face, hand adjusting headphones, headphones shifting, person walking behind, station appearing in background, color shift, warm orange tone, golden hour, daylight, sound, audio, music, dialogue,

Positive： ▎ Extreme wide shot of a long tunnel of bright vermilion red torii gates receding into the depth of a Japanese forest shrine path. A dark stone step pathway runs down the center, lush cedar trees on both sides, a small wooden shrine structure visible on the right edge. Dappled mid-afternoon sunlight filters through the canopy onto the torii and stone steps. ▎ ▎ At the far end of the tunnel deep in the background, a young Japanese woman with short black bob and bangs walks forward toward the camera along the stone path — wearing a black thin-strap sleeveless tank top, loose baggy light blue jeans, white sneakers, a light blue denim crossbody bag, white wired in-ear earphones with cable trailing down. She walks at a natural everyday pace, two to three purposeful steady steps forward, the way someone walks on a quiet afternoon stroll. Her scale stays small, she remains a distant figure across the entire clip. ▎ ▎ Subtle micro motion only: cedar tree leaves and shadows on the stone path shimmer very gently in a light breeze. The vermilion torii remain perfectly still and bright. No other people appear. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Bright vivid vermilion red torii, deep green cedar forest, dappled warm daylight, dark stone steps. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated. Negative： ▎ walking backwards, retreating, going backward, reverse direction, reversed playback, walking away from camera, figure receding into depth, slow motion, slow-mo, sluggish walking, dragging feet, zombie walking, floating gait, hovering, lethargic, dreamy slow motion, frozen mid-step, hesitation, ghost-like movement, gliding, sliding instead of walking, figure becoming large quickly, running, sprinting, head turn, look at camera, face to camera, smile, parted lips, sighing, exhaling, hair flip, earphones falling out, bag swinging dramatic, fast camera push, zoom in, zoom out, pan, tilt, dolly, parallax shift, perspective warp, torii bending, stone steps warping, torii color fading, torii becoming dull, color shift, sudden brightness change, tourists appearing, monks appearing, priests appearing, other people walking, plastic CG lighting, glow halo, lens flare, oversaturated colors, painterly stylization, vintage filter, anime style,

rotating gently as they fall. Foreground leaves in sharp focus, background of warm wooden shrine structures and deep green cedar trees softly out of focus. Backlight from behind makes the leaves half-translucent, glowing gold. Floating drifting tempo — never sideways flying, never fast. Silent video. 16:9. Negative: leaves swirling chaotically, strong wind, fast motion blur, leaves only on ground, no leaves in air, leaves falling straight stiffly, snow, rain, camera movement, anime style.

Close-up of three to four wooden ema plaques caught mid-sway in light breeze, ropes visible with fine shadow detail, warm afternoon sunlight from upper-right grazing wooden surfaces. Each plaque sways very slightly 1-2 times, almost imperceptible — the "wind just starting" feel. Camera locked, zero movement. Background softly blurred warm wood and shrine forest tones. 16:9, silent video, naturalistic photographic light. Negative: plaques violently swinging, plaques flying off, broken cords, fallen plaques, completely still rigid, hands appearing, readable text, fast wind, dramatic motion

hair / nose bridge / upper lip / chin / ear edge softly rim-lit. Background ema wall melted into warm gold bokeh. Camera locked, zero movement. Eyelids tremble micro-once but do NOT blink. Lips stay closed, no parting. Chest softly inhales once. Black strap on shoulder visible, white earphone wire down neck. 16:9, silent video, naturalistic photographic light. Negative: face toward camera, three-quarter angle, looking up, smiling, parted lips, sighing, hair in face, ear-tucking, harsh side rim, raccoon eye shadow, dramatic black background, blinking, eye movement, plastic CG light, anime style

Positive: Cinematic 16:9, locked static low-angle close shot on the floor of a moving late-night train. A black business briefcase sits upright on the train floor next to a pair of polished black men's leather shoes, with one strap dangling loose and resting on the floor. The carriage rocks gently from side to side, causing the loose strap to sway softly with a slow natural rhythm and the briefcase to shift only the slightest amount on the floor. The shoes stay perfectly still. The light is cool fluorescent, casting subtle reflections on the polished shoe leather. Photoreal, intimate detail, late-night quiet atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, hand reaching into frame, hand picking up bag, hand adjusting strap, someone reaching for the bag, foot moving, shoes shifting position, shoes lifting, owner standing up, owner getting up, briefcase tipping over, briefcase sliding away, strap whipping violently, strap flying, exaggerated swing, jerky train shake, sudden lurch, train stopping, doors opening, station floor appearing, station signs visible, other objects entering frame, another passenger's feet appearing, ceiling lights flickering, color shift, warm orange tone, daylight, sound, audio, music, dialogue

Positive: Cinematic 16:9, locked static POV shot through a moving train carriage window at night. The black metal window frame is visible at the left edge. Through the glass, a dense Japanese residential block of apartment buildings slowly drifts past in one continuous horizontal direction, revealing scattered warm-lit windows and the silhouettes of balconies. A small bright moon sits in the dark sky in the upper portion of frame. The buildings move at a steady, even, gentle pace — neither too fast nor stop-and-go — conveying the constant motion of a late-night train. Window lights stay perfectly steady without flicker. Faint reflections of the interior carriage ghost subtly across the glass. Photoreal, melancholic late-night urban atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, train decelerating, train accelerating, train stopping, scenery reversing direction, scenery jerking, choppy motion, parallax breaking, buildings teleporting, sudden zoom into a window, station platform appearing, station signs appearing in window, signage with readable text, scrolling text, advertisements, train tunnel appearing, scenery turning to pitch black, weather change, rain on window, raindrops, fog appearing, building lights flickering, lights turning on or off, neon signs animating, animated billboards, moon moving across sky, moon changing size, a human figure appearing in a lit window, silhouette appearing, person appearing on a balcony, ghostly reflections of faces, color shift, warm orange tone, daylight, blue daytime sky, sound, audio, music, dialogue, voiceover

Positive: Cinematic 16:9, locked static medium shot through a moving train carriage window at night. The frame is filled by a single distant Japanese apartment building façade. One window in the middle of the building glows softly with warm interior light — visible curtains and the suggestion of a quiet room behind glass — while every other window around it remains dark or dim. The entire building drifts slowly and steadily in one horizontal direction across the frame, conveying the train's continuous motion. The lit window stays consistently glowing without flicker. No silhouettes appear behind the curtains. Subtle ghost reflection of the train interior on the glass. Photoreal, melancholic, late-night urban quiet. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, train decelerating, train accelerating, train stopping, scenery reversing, scenery jerking, choppy motion, sudden zoom into the lit window, station appearing, station signs, signage with readable text, scrolling text, animated billboards, neighboring windows turning on, lights turning off, lit window flickering, lit window changing brightness, color of lit window shifting, a human silhouette appearing behind the curtain, a figure appearing in the window, balcony figure, weather change, rain on window, fog, building changing shape, building deforming, multiple buildings stitched together, color shift, warm orange flood, daylight, blue daytime sky, sound, audio, music, dialogue, voiceover

Locked medium shot of a Japanese shrine ema wall fully filling the frame, hundreds of pentagonal house-shaped wooden ema plaques hanging in dense rows on wooden beams, warm wood shrine wall and softly blurred cedar leaves in background. Warm right-upper 30° afternoon light hits the plaques, fine rope shadows cast on wood. A few plaques in the wind tremble almost imperceptibly, 1-2 micro sways only. Camera locked, zero movement. 16:9, silent video, naturalistic photographic light. Negative: people appearing, hands reaching in, modern plastic plaques, sparse plaques, plaques violently swinging, plaques flying, readable text, anime style

3/4 rear-side medium long shot of young woman standing in front of dense wooden ema plaque wall, camera positioned to her rear-right. Black thin strap top exposing nape and shoulder line, loose light blue denim, blue denim crossbody bag, white wired earphones. Right hand lifts, fingertip just touches the lower edge of one ema plaque — only touching, not taking. Left hand rests at side. Warm right-upper 30° light traces nape, shoulder line, waist line, raised arm — golden rim on skin. Camera locked, zero movement. Shoulder muscle micro-shifts once, hair tips drift in breeze once. Hand stays at touch, no pull. 16:9, silent video, naturalistic photographic light. Negative: face fully visible, full frontal, looking at camera, head turn, plaque grabbed firmly, plaque pulled down, two hands, hair covering nape, harsh front light, plastic CG light, anime style, dramatic motion

Positive: Cinematic 16:9, locked-off static wide shot of a deserted late-night train platform. The last train slowly enters frame from the right, headlights gently flaring on the wet polished platform floor. Train moves at a slow steady pace, gradually decelerating without coming to a full stop. Cool white fluorescent platform lights cast a flat, slightly washed-out glow. Faint reflections on damp ground. Distant ambient haze. Subtle motion blur only on the moving train carriages. Everything else perfectly still. Photoreal Japanese suburban station, late night, empty, quiet atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, handheld shake, any camera movement, people appearing on platform, pedestrians, passengers walking, station staff, conductor, train doors opening, doors sliding, train coming to abrupt stop, train accelerating, fast motion, blinking signage, animated station signs, scrolling text, lens flare bursts, overexposed highlights, weather change, rain falling, snow, leaves blowing, birds, flickering lights, color shift, warm tone, golden hour, daylight, sound, audio, music, dialogue, voiceover

Positive: Cinematic 16:9, locked static medium shot of a tired middle-aged Japanese salaryman in a dark suit and grey-striped tie, asleep against the train window. His head tilts slightly to one side, eyes peacefully closed, mouth gently shut. He sleeps soundly — only the slow soft rise and fall of his chest from natural breathing, and a barely perceptible body sway from the train's motion. His head remains tilted in the same position throughout. Behind him, the dark train window carries a soft moving blur of the night exterior, and his faint reflection on the glass moves in sync. Cool fluorescent carriage light falls evenly across his face. Photoreal, quiet, late-night commuter atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, salaryman waking up, eyes opening, eyelids fluttering, eyes flickering open, head lifting, head straightening, head turning, head sliding off window, mouth opening, mouth yawning, drool, visible snoring, lip movement, eyebrows moving, frowning, smiling, hand twitching, hand moving, arms moving, posture shift, leg crossing, leg uncrossing, jacket adjusting, tie adjusting, phone vibrating in pocket, another passenger appearing, passenger walking past behind, reflection of another person in window, station appearing in window, station signs visible, train stopping, doors opening, exaggerated train shake, jerky motion, ceiling lights flickering, color shift, warm orange tone, golden hour, daylight, sound, audio, music, dialogue, voiceover, snoring sound

Very slow gentle dolly push-in from directly behind her (5–10% movement only). Two thick bright vermilion red torii pillars frame left and right of the image. Young woman seen from behind, waist up, walking calmly through the gate into the next torii ahead. Black thin strap top, loose light blue jeans, white wired earphone cord swaying gently along the back of her neck. Dappled warm afternoon light on her shoulder and the stone path. Vermilion red stays bright vivid vermilion red. Silent video. 16:9. Negative: face visible, looking back, head turning, hair covering ears, earphones missing, over-ear headphones, fast motion, camera shake, fading torii, walking backwards, anime style.

Locked static camera. Mid-distance shot of a quiet shrine side path. An elderly woman 70–80 years old in dark grey samue work clothes stands bent over with a tall bamboo broom takebouki, slowly sweeping scattered yellow ginkgo and red maple leaves on stone steps. Seen from her back or 3/4 back, posture slow and gentle. The broom moves slowly forward and back twice, leaves shifting lightly. Soft warm afternoon light, warm wood shrine architecture softly out of focus in background. Silent video. 16:9. Negative: young woman, main character appearing, smiling face, dramatic action, fast motion, broom flying, leaves swirling chaotically, multiple people, tourists, monks, priests, anime style.

a phone inside, and her hand hanging relaxed at her side beside it. Between 0.3s and 0.6s the pocket gives a single very subtle vibration — the faintest tremor of fabric — then stillness returns. Hand stays completely still beside the pocket the entire time, never moves toward pocket. Side light on denim, background softly out of focus. Silent video. 16:9. Negative: hand inside pocket, hand reaching for pocket, hand touching pocket, phone screen visible, phone pulled out, dramatic gesture, hand moving, fast motion, multiple vibrations, camera shake, anime style.

A young woman in chest-up frontal medium shot, eye level, looking slightly off-screen left at an ema plaque wall, completely neutral calm expression — not sad not happy not cold, just "didn't hear it". Black thin strap top, loose light blue denim, white wired earphones in both ears. Soft front-left 30° afternoon light on face, warm rim on ear and hair tips, vermilion red torii and deep green cedar forest blurred bokeh background. Camera locked, zero movement. Only her chest rises once with a tiny breath. Eyelids do not move. Lips stay closed. 16:9, silent video, naturalistic photographic light. Negative: looking at camera, eye contact, smiling, half smile, parted lips, sighing, blinking, surprised face, contemplative dramatic, hand to pocket, head turn, eye movement, plastic CG light, anime style, faded torii, dull torii

Static locked camera, no pan no zoom no push no pull. 0-2s: empty street corner completely still, vending machine glowing cold white in the dark, convenience store sign bokeh in background, faint sodium street lamp. A taxi passes far in the deep background left to right, headlights sweeping briefly. 2-4s: a slim female figure enters from the far right edge, walking slowly toward the vending machine, only a small silhouette, back to camera, unhurried pace. Subtle ambient air movement, slight neon flicker on the vending machine. Cinematic 16:9, shallow depth, film grain. Silent video. Negative： camera shake, push in, pull out, zoom, pan, dolly, handheld, sighing, parted lips, exhaling, plastic lighting, over-ear headphones, bare legs, face visible, multiple people in foreground, fast walking, running

Static locked camera, eye-level chest-up medium shot. Cold white vending machine light spills from off-screen left, splitting her face into bright and shadow halves. She is already lit at frame start, gaze focused on the products. Extremely subtle motion only: her eyes drift slowly from upper row down to lower row, head stays still. White wired earbud cable visible at neck. Sodium street lamp cool rim from behind. Hair barely shifts with breath. Cinematic 16:9, shallow depth, film grain. Silent video. Negative： head turning, head tilt, looking around, heavy blinking, fluttering eyelids, sighing, parted lips, exhaling, smiling, frowning, plastic lighting, over-ear headphones, bare legs, camera shake, zoom, push in, pull out, pan

Slow tracking shot1from behind, camera follows at waist-up medium framing, very gentle forward drift matching her pace. She walks two natural steps toward the vending machine and stops, shoulders relaxed, head tilts slightly down to scan the products. White wired earbud cable visible draping down the side of her neck, swaying minimally with motion. Cold white light from the vending machine washes over her silhouette, sodium street lamp warm rim from behind. Hair settles after she stops. Cinematic 16:9, shallow depth, film grain. Silent video. Negative： turning around, looking back, face visible, head turning, sighing, parted lips, exhaling, plastic lighting, over-ear headphones, bare legs, fast walking, running, camera shake, zoom, push in, pull out

Positive： Low angle medium shot on the lower section of a lit vending machine at night, focused on the product drop chute. A short stubby dark coffee-brown aluminum can with cream-white top ring and silver pull-tab tumbles down from inside the machine and lands in the drop bin, rolling to a stop on its side. The interior of the chute glows cold white, illuminating the can as it falls. Outside the machine, sodium street lamp warm spill on the pavement edge. Camera locked, no shake, no push. Faint motion blur on the can mid-fall only. Shallow depth of field f2.0, cinematic film grain, 16:9, silent video, 5 seconds. Negative： hand reaching in, fingers entering frame, person picking up can, can rolling out of chute, can bouncing wildly, multiple cans, wrong color can, green can, yellow can, clear bottle, plastic bottle, readable label text, brand logo visible, motion blur on background, camera shake, push-in, zoom, plastic lighting, over-saturated colors, jitter, flicker, watermark, text overlay

Cinematic 16:9, locked static shot taken from a quiet late-night Japanese train platform, looking forward toward the open doorway of a stopped commuter train. A young woman stands on the platform with her back fully to camera, facing the train doorway, in the moment just before boarding. She wears white over-ear headphones, a black leather jacket, a short black skirt with bare legs, black ankle boots, and a black shoulder bag. The open train doors of the stopped train form a frame in the middle of the image, with sliding metal door panels on both sides retracted open. Beyond the open doorway, the interior of the train carriage is clearly visible — pink upholstered bench seats line both sides, a few overhead hand straps hang above, cool fluorescent light fills the interior. The platform floor in the foreground is slightly damp and reflective, lit by the spill from inside the carriage. Photoreal, late-night, quiet realism.

Positive: Cinematic 16:9, locked static medium shot from inside the train looking through the open carriage doorway. A young woman with short black bob hair, wearing white over-ear headphones, a black leather jacket and a black shoulder bag with subtle silver hardware, has just stepped off the train onto the platform. From the first frame to the last frame she walks slowly and steadily forward away from camera, at a calm even unhurried pace, never stopping, never hesitating, never slowing down. Her head stays still throughout, no glances back. She gradually gets smaller in frame as she walks deeper into the platform. The platform extends ahead under cool white fluorescent lights, slightly damp and reflective. A faint breeze just barely lifts the ends of her hair. Photoreal, quiet, late-night arrival atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, she stops walking, she pauses mid-step, she hesitates, she slows down then stops, she stands still, she comes to rest, she stops to look around, she turns her head, she looks back, head rotation, face reveal, she runs, she sprints, fast walking, hurried pace, irregular gait, she stumbles, she removes headphones, she touches headphones, she pulls out phone, hand reaching into pocket, she waves, train doors closing on her, doors sliding shut, other passengers exiting around her, passengers walking past her, station attendant appearing, train pulling away with motion blur, station signs scrolling readable text, color shift, warm orange tone, daylight, sound, audio, music, dialogue, voiceover, footstep sound

Positive prompt: Medium shot from behind, three-quarter back angle. A woman stands in front of a Japanese vending machine at 02:47 AM, leans down briefly, picks up a short cylindrical canned coffee — deep brown matte aluminum body, narrow milk-white ring, silver pull tab, logo area unreadable — from the retrieval slot at the lower front of the machine. She straightens up holding the can in one hand at hip level, turns, and begins walking away from the machine toward screen left, exiting frame in two unhurried steps. Her gait is steady and even, no hesitation, no looking back at the machine. The can stays clearly visible in her hand throughout the walk. Cool fluorescent light from the machine bathes her back, warm streetlamp light from the corner ahead catches her front edge. Shallow depth of field, f1.8 cinematic look. Camera fixed, no movement. 16:9 cinematic, silent video. Negative prompt: looking back at vending machine, turning head over shoulder, glancing at can, examining can, drinking from can, opening pull tab, raising can to face, hesitating, stopping mid-walk, two cans, dropping the can, can falling, can not visible in hand, hand empty, leaving can in retrieval slot, walking toward camera, walking toward machine, running, fast walk, bouncing gait, dramatic stride, head down sad posture, head up confident pose, smiling, frowning, phone in hand, light brand can, green can, white can, lemon can, soda can, tall slim can, readable logo, readable text on can, motion blur, fade out

Positive prompt: Extreme close-up of a woman's hand and index finger pressing a button on a Japanese vending machine panel at 02:47 AM. The fingertip approaches the illuminated button slowly, presses it down with a single short, decisive movement, holds for half a second, then lifts off. The button glows faintly behind the fingertip. Vending machine front panel fills the frame, with adjacent product buttons softly out of focus around the central pressed button. Cool fluorescent light from the machine washes the hand in a faint blue-white tint, with a thin warm rim from a nearby corner streetlamp. Shallow depth of field, f1.4 cinematic look, sharp focus on the fingertip and the pressed button. Camera fixed, no movement. 16:9 cinematic, silent video. Negative prompt: two hands, multiple fingers pressing, finger hesitating, finger tapping multiple times, finger sliding sideways, hand shaking, button not lighting up, full machine view, woman's face visible, body visible, coin slot, inserting coin, money, wallet, phone in hand, drink can in hand, screen text readable, brand logos readable, warm lighting, daytime, blurry fingertip, motion blur on finger, finger pulling back early, no contact with button

Positive prompt: Extreme close-up of a woman's eyes only, framed from upper cheekbone to brow line. Her gaze shifts down and slightly to the side, looking toward a phone held at chest level just out of frame. The eyes move once with a small, contained motion — no widening, no blinking burst, no emotion. The pupils settle on the off-screen phone screen, then hold still. Eyelashes faintly reflect the cool bluish light of the phone screen below. Background completely out of focus, soft warm bokeh from corner streetlamp far behind. Shallow depth of field, f1.2 cinematic look, sharp focus on iris and lash line. Camera fixed, no movement. 16:9 cinematic, silent video. Negative prompt: full face visible, mouth visible, nose visible, smiling eyes, frowning, raised eyebrow, eye-roll, rapid blinking, tears, makeup glamour shot, eye contact with camera, looking up, looking far away, eyes darting around, eyes closing, sleeping eyes, head turning, head tilting, hand visible, phone visible in frame, screen text readable, hair covering eyes, dramatic lighting, color shift on eyes, motion blur, jump cut, multiple gaze movements

Prompt： ▎ Add the woman from references 2 and 3 into the bright vermilion torii tunnel scene from reference 1. She must stand at the far end of the torii tunnel deep in the background, appearing very small as a distant silhouette, only roughly 1/8 the frame height. She faces toward the camera, walking forward along the stone step path, but is too distant to see facial detail. ▎ ▎ She wears the exact outfit from reference 2: black thin-strap sleeveless tank top, loose baggy washed light blue jeans low-rise, white cotton ankle socks, white sneakers, a small light blue denim crossbody bag slung across her body, white wired in-ear earphones with white cable trailing down. Her face matches reference 3 exactly — short black bob with blunt blunt straight bangs, single-eyelid Japanese features, calm neutral expression. ▎ ▎ The torii tunnel composition from reference 1 must remain completely unchanged — same bright vermilion red torii gates receding into the depth, same dark stone step pathway, same lush green forest on both sides, same wooden shrine structure on the right edge, same dappled daylight. Do not move, remove, or add anything else. Do not replace any existing element. Do not dim or change the torii color — keep them vivid bright red. ▎ ▎ Daytime mid-afternoon, dappled sunlight filtering through cedar trees onto the stone steps. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, slightly desaturated forest greens against vivid vermilion red. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ face close-up, large figure, figure in foreground, figure mid-tunnel, multiple people, no people, monks, tourists, priests, modified torii, faded torii, dull torii, removed torii, changed stone steps, removed trees, added lanterns, oversaturated colors, plastic CG lighting, glow halo, lens flare, painterly stylization, vintage filter, anime style, illustration, overcast lighting, rain, snow

Positive: Cinematic 16:9, locked static medium shot from a 45-degree side angle. A young woman sits on a pink train bench, wearing white over-ear headphones, a black leather jacket over a blue-and-white striped collared shirt, gaze cast slightly downward into empty space. She holds perfectly still — only the faintest natural breath in her shoulders, a barely perceptible side-to-side rock from the train's motion, and a slow natural blink once or twice. No smile, no frown, no eye-roll, eyes locked forward without darting. Hair stays settled around her face. Behind her, the dark train window shows a soft continuous blur of exterior motion, indicating the train is moving. Warm-cool fluorescent carriage light falls evenly on her face. Photoreal, quiet, contemplative late-night atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, she turns her head, she looks at camera, she looks sideways, head rotation, eye darting, eye-roll, eye glance, sudden gaze shift, she smiles, she frowns, she sighs visibly, she touches face, she adjusts headphones, hand lifts to ear, she takes off headphones, she pulls out phone, hand reaching into pocket, she shifts posture, she crosses legs differently, she leans forward, she leans back, other passengers entering frame, passenger walking past behind, reflection of another person in window, station appearing in window, station signs visible, train stopping, doors opening, exaggerated train shake, jerky motion, ceiling lights flickering, color shift, warm orange tone, daylight, sound, audio, music, dialogue, voiceover

Locked low-angle camera looking up. Chest-up framing. Top third of frame is the horizontal kasagi beam of a bright vermilion red torii gate against deep green cedar forest. Warm dappled light from above strikes her jawline, eyelashes, bridge and tip of nose, leaving lower face in soft shadow. From the start her head is already tilted slightly upward. Holds 1s, then head settles back to neutral with the smallest natural micro-motion. Eyes do not move. Lips stay closed. Naturalistic photographic light. Silent video. 16:9. Negative: face flat-lit, mouth opening, parted lips, smiling, sighing, eye movement, eyes darting, blinking heavy, head turning fully, camera movement, dramatic head tilt, plastic CG, anime style.

Positive: Cinematic 16:9, locked static shot. The camera is mounted inside the moving train carriage, looking outward through the open doorway. The metal doorframe occupies the foreground edges of frame and stays absolutely fixed in position throughout the entire shot. A woman stands just inside the train, her back fully to camera, wearing white over-ear headphones, a black leather jacket and a black shoulder bag. She remains motionless, in the exact same position relative to the camera and the doorframe from the first frame to the last frame — only the faintest natural shoulder breathing and a slight hair sway. The train is in motion: the platform visible beyond the open doors slowly drifts in one horizontal direction at a steady gentle pace, the only moving element in frame. The doorframe and the woman never move within frame. Cool washed-out fluorescent light from the platform spills onto her jacket. Photoreal, late-night, quiet atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, train pulling away from camera, train receding into distance, doorway shrinking, doorway moving, doorframe shifting position, doorframe drifting, woman drifting backward in frame, woman receding, woman shrinking, woman walking forward, woman stepping deeper into train, woman stepping off train onto platform, woman remaining on platform, woman left behind, she turns her head, she looks back, head rotation, she removes headphones, she lifts hands, doors closing, doors sliding shut, station signs scrolling readable text, color shift, warm orange tone, daylight, sound, audio, music, dialogue, voiceover

Positive: Cinematic 16:9, locked static shot. The camera is fixed; the metal doorframe of the train occupies the foreground edges of frame and stays absolutely fixed in position throughout. A woman stands just inside the train doorway, her back fully to camera, wearing white over-ear headphones, a black leather jacket and a black shoulder bag. From the first frame to the last frame she walks slowly and steadily forward away from camera, deeper into the carriage, at a calm even unhurried pace. She never turns her head, never stops, never hesitates, never slows. She gradually gets smaller in frame as she moves further inside the train. The doorframe stays locked in the same position. Cool washed-out fluorescent light inside the carriage. Photoreal, late-night, quiet atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, doorframe shifting position, doorframe drifting, she stops walking, she pauses mid-step, she hesitates, she stands still, she comes to rest, she steps backward, she steps onto platform, she leaves the train, she walks toward camera, she turns her head, she looks back, head rotation, face reveal, profile reveal, she removes headphones, she lifts hands, she pulls out phone, doors closing on her, doors slamming, doors moving, other passengers walking past her, passengers entering behind her, train pulling away in background, station signs scrolling readable text, color shift, warm orange tone, daylight, sound, audio, music, dialogue, voiceover, footstep sound

Positive: Cinematic 16:9, locked static three-quarter side close-up of a young woman with short black bob hair and blunt fringe. Her hands are gently grasping the white over-ear headphone cup on her right ear. With a slow, soft, deliberate motion she lifts the headphone cushion away from her ear and gently pulls it outward and downward over the next four seconds. Her head stays still, gaze remains soft and slightly downcast, no expression change. By the end of the shot, the cushion has cleared her ear by a few centimeters, the thin white cable visible. Background is the softly out-of-focus interior of a moving late-night train carriage, passengers behind her remain motionless. Cool fluorescent light. Photoreal, intimate, slow tempo. Silent video.

Positive: Cinematic 16:9, locked static macro close-up of an open palm cradling a pair of white over-ear headphones — ear cushions resting sideways in the palm, the slim metal headband arching above, the thin white cable curving out from the bottom. Hand emerges from a black leather jacket cuff. The hand and headphones hold almost perfectly still throughout the shot — only the subtlest natural side-to-side sway from the train's gentle motion, an almost imperceptible loosening of the fingers, and the lightweight cable swaying ever so slightly. Background is the softly out-of-focus dark interior of a late-night train, faint cool window reflections in the far distance. Cool fluorescent light catches the white cushions and the silver headband. Photoreal cinematic still mood, shallow depth of field, intimate quiet tone. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, hand tilting dramatically, hand rotating, fingers closing into a fist, hand lifting headphones up, hand putting headphones down on a surface, other hand entering frame, hand bringing headphones to ear, headphones rolling out of palm, headphones falling, headphones disappearing, headphones flickering or doubling, cable whipping, cable flying, cable knotting, ear cushions deforming, headphones changing shape, headphones changing color, station appearing in background, passenger walking past, color shift, warm orange tone, daylight, sound, audio, music, dialogue, voiceover

Positive: Cinematic 16:9, locked static symmetrical wide shot down the center aisle of a moving late-night Japanese commuter train. Pink upholstered bench seats line both sides. On the left bench, a young woman with short black bob hair, white over-ear headphones, a black leather jacket, blue striped collared shirt, short black skirt with bare legs and black ankle boots, sits with a black shoulder bag beside her, gaze cast softly downward, hands resting on her lap. She remains motionless throughout, only the faintest natural breathing. A man in a dark suit and hat sits on the right bench in the middle distance, also still. A couple of other passengers further down the carriage remain motionless. Overhead hand straps sway gently in a slow synchronized rhythm from the train's movement. The carriage rocks almost imperceptibly side to side. Warm-white LED ceiling lights glow steadily without flicker. The background window panels show a continuous blur of dark exterior passing by, indicating motion. Photoreal, quiet late-night atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, woman on left standing up, woman on left getting up, woman on left walking, woman on left turning head toward camera, woman on left looking at camera, woman on left smiling, woman on left removing headphones, woman on left pulling out phone, woman on left adjusting bag, woman on left changing posture, woman on left crossing legs, other passengers standing up, passengers walking, passengers turning heads, passengers looking at camera, new passengers entering frame, passengers exiting, doors opening, doors closing, train stopping at station, station platform appearing in window, station signs appearing, hand straps swinging violently, exaggerated train shake, jerky motion, ceiling lights flickering, color shift, warm orange tone, daylight, sound, audio, music, dialogue, voiceover, announcement

Positive： Static low-angle close-up on the product drop chute of a lit vending machine at night. A short stubby dark coffee-brown aluminum can with cream-white top ring and silver pull-tab rests on its side inside the chute, fully visible, the matte metallic surface catching cold white interior light. The can is centered, sharp, perfectly readable as a shape and color. The vending machine glow is the dominant light source, sodium street lamp warm fill spills faintly on the pavement edge in foreground bokeh. No person in frame, no hand, no movement. The shot holds completely still, only the faintest ambient air shimmer. Shallow depth of field f2.0, cinematic film grain, 16:9, silent video, 5 seconds. Negative： hand entering frame, fingers reaching, person crouching, can moving, can rolling, can being picked up, multiple cans, wrong color can, green can, yellow can, clear bottle, plastic bottle, readable brand logo, sharp readable label text, camera push-in, zoom, pan, shake, dolly, motion, plastic lighting, over-saturated colors, lens flare, jitter, flicker, watermark, text overlay

【生成内容】 镜 1 · 街对面远景空镜\ 景别：远景 / 固定 / 5 秒\ 画面：夜街，洗衣店招牌灯亮着，玻璃窗后一个小小人影坐着。路上没人，便利店的灯在更远处。\ 独白入：（无 — 留 2 秒环境声后起）"这家洗衣店，不知道为什么我挺喜欢的。"\ 声音：远处车声、风、自动门隐约的"叮" --- 镜 2 · 推门进店主观\ 景别：中景 / 缓推 / 4 秒\ 画面：从门口往里看，洗衣机一排发着冷白光，地板反光，她已经坐在椅子上。视角像观众跟着她走进来。\ 独白：（继续）"灯太亮了，这样反而安心。"\ 声音：滚筒水声起、荧光灯嗡鸣 --- 镜 3 · 她坐定中景\ 景别：中景 / 固定 / 6 秒\ 画面：她坐在洗衣机前的塑料椅上，怀里一个鲑鱼饭团。微微低头，刚撕开包装。背景滚筒在转。\ 独白：（停一下）"便利店的鲑鱼饭团，"\ 声音：塑料包装窸窣 --- 镜 4 · 饭团手部特写\ 景别：特写 / 微推 / 4 秒\ 画面：双手捧着饭团，海苔的边露出来，指尖犹豫了一下。\ 独白：（轻）"偶尔会咬到一颗特别咸的米。"\ 声音：包装定格、滚筒在背景 --- 镜 5 · 咬一口侧脸\ 景别：近景侧 / 固定 / 5 秒\ 画面：侧脸，下巴和嘴角的轮廓被荧光灯打亮，她咬了一小口，嚼。眼睛没看任何地方。\ 独白：（顿）"不是不好吃。就是会停一下。"\ 声音：嚼的轻声、滚筒\ ※ 关键镜："停一下" 的视觉化，她真的停了 --- 镜 6 · 滚筒里的衣服\ 景别：物件特写 / 固定 / 4 秒\ 画面：滚筒里花花绿绿的衣服在翻，玻璃门上有水雾，映出她模糊的轮廓。\ 独白：（无 — 留白）\ 声音：水声大起来一秒，衣服打在筒壁的"啪" --- 镜 7 · 玻璃门外橘猫走过\ 景别：中景 / 固定 / 5 秒\ 画面：透过玻璃门看出去，街上一只橘猫不紧不慢走过，从画面右进左出。前景是她坐着的虚化背影。\ 独白：（轻）"玻璃门外，一只橘猫慢慢走过去，看都没看我。"\ 声音：街道风、远处自行车铃 --- 镜 8 · 她目送（视线镜）\ 景别：近景 / 固定 / 4 秒\ 画面：她抬头看了一眼玻璃门外，眼神跟着猫，但没有起身。嘴角有极轻一丝笑，不到一秒。\ 独白：（停顿）"嗯…也对。它有它要去的地方。"\ 声音：荧光灯嗡 --- 镜 9 · 滚筒停转 / 提示音\ 景别：近景 / 固定 / 3 秒\ 画面：滚筒慢慢停下，机器顶上的灯由红转灭。她从椅子上抬眼。\ 独白：（无）\ 声音：洗衣机"叮"一声、嗡鸣消失 --- 镜 10 · 抱衣服中景\ 景别：中景 / 固定 / 5 秒\ 画面：她蹲在洗衣机前，怀里抱着一团刚烘好的衣服，脸贴着衣服蹭了一下。\ 独白：（轻）"刚烘好的衣服抱在怀里，是热的。"\ 声音：布料窸窣 --- 镜 11 · 推门出去脚步特写\ 景别：低机位脚步特写 / 跟移 / 4 秒\ 画面：白色帆布鞋踩在洗衣店瓷砖上，跨过门槛踩到外面的柏油路，画面边缘玻璃门正在合上。\ 独白：（无）\ 声音：自动门"哗"开合、外面街声涌进来 --- 镜 12 · 街上背影远景空镜\ 景别：远景 / 固定 / 6 秒\ 画面：她抱着衣服往前走，背影越来越小，路灯一盏一盏延伸。空了一两秒，结束。\ 独白：（最后一句，轻）"明天，想买点葡萄。"\ 声音：风、鞋底、远处电车

A 22-year-old East Asian girl, breathtaking ethereal beauty, goddess-tier visuals, top 0.01% face, flawless facial proportions. Long elongated almond eyes, soft monolid / subtle inner-fold, slightly upturned outer corners (subtle cat-eye lift), defined sharp eyeliner-like lash line, long thick natural lashes, soft pink-brown eyeshadow with subtle glitter, glossy lower-lash catchlight, light aegyo-sal under eyes. Small straight nose with refined high bridge and tiny tip, small full lips with sharp defined cupid's bow, plump lower lip, glossy nude-pink lip tint. Very small face, sharp clean V-line jaw, tiny pointed chin, soft high cheekbones with hint of baby fat, luminous porcelain skin (matte-to-dewy, poreless, soft inner glow, blush on cheeks and nose tip). Glossy jet black hair, shoulder-length, straight with slight inner curl at ends, blunt full bangs (hime-style straight fringe) sitting just above the eyebrows, two thin face-framing strands at the cheeks. Three-view character sheet (front / three-quarter / profile), soft daylight, neutral light backdrop, plain ivory cotton tee + soft beige cardigan, calm gentle gaze, lips softly closed, youthful refined mood. Photorealistic, 85mm portrait, f/1.4 shallow depth of field, cinematic editorial color grading, warm soft fill light, luminous skin glow, ultra sharp eye and skin detail, high-end fashion magazine cover quality, Korean-Japanese idol-actress visual style. Reference: K-pop visual member tier, Japanese gravure idol crossover, youthful refined monolid beauty, 2024 editorial portrait aesthetic. Negative: large round double eyelid eyes, droopy eyes, swollen eyelid, mature face, plain face, average looking, square jaw, wide face, large face, anime, plastic skin, heavy makeup, oily skin, low-tier idol look, AI-generated stock face.

Adjust two things, keep everything else identical: 1. EYELASHES — reduce length and density: - Natural medium-length lashes (not dramatic, not long) - Soft and feathery, not bold or doll-like - Lower lashes subtle and natural - Keep the cool composed gaze, sharp clean lash line - Goal: refined natural lashes, not enhanced 2. HAIRSTYLE — change to Japanese straight short hair: - Sleek straight black short hair, length at lower jaw / just below jawline - Hair COVERING the ears completely (ears hidden behind hair) - Ends are straight blunt cut, NO inward curl, NO outward flip - Clean precise haircut, like high-end Japanese salon work - Blunt full bangs straight across at eyebrow level - Sharp polished texture, glossy black, every strand neat - Style reference: Aoi Yu / Mitsushima Hikari / early Nikaido Fumi short hair, Japanese minimalist sleek short cut Keep identical: - Same face, same skin, same makeup base - Same ivory tee and beige cardigan - Same three-view composition (front / 3-4 / profile) - Same neutral grey backdrop and soft daylight Photorealistic, sharp detail, preserve original character identity. Negative: long lashes, dramatic lashes, fake lashes, doll lashes, inward curl bob, outward flip, wavy hair, layered hair, ears showing, ear-length cut, choppy ends, different face, different person.

A 22-year-old Japanese girl, stunning beauty, soft elegant monolid eyes with long lashes, refined delicate features, small face, high nose bridge, defined cheekbones, clean sharp jawline, V-line chin, porcelain flawless skin (matte, smooth, glowing), naturally pretty no-makeup look: softly defined brows, glossy bare pink lips, bright clear eyes with catchlight, no fatigue. Long straight black hair past shoulders, silky and glossy, with soft wispy curtain bangs (air bangs / hime side fringe), gently framing the face. Three-view character sheet (front / three-quarter / profile), soft studio daylight, light grey backdrop, plain ivory cotton tee + soft beige cardigan, calm gentle gaze, lips softly closed, faint ethereal mood. Photorealistic, 85mm portrait, shallow depth of field, cinematic soft color grading, warm fill light, gentle skin glow, sharp eye detail, refined Japanese editorial aesthetic. Reference: Komatsu Nana, Mori Nana, Nikaido Fumi (monolid version), Japanese Vogue / GINZA magazine cover beauty, youthful refined idol-actress crossover. Negative: double eyelid, droopy eyes, swollen eyelid, dark circles, tired face, plain face, average looking, mature face, oily skin, visible pores, anime, plastic skin, heavy makeup, dramatic shadows.

A 22-year-old Japanese girl, stunning cute beauty, soft elegant monolid eyes with long lashes, ──slightly─upturned─outer─corners─(subtle─cat-eye─shape),─────────────────────────────────────────────────────────────── refined delicate features, very small face, tiny─pointed─chin,─V-line─jaw,──────────────────────────────────────────────────────────────────────────────────────── soft round cheeks with youthful baby fat, small straight nose with refined bridge, porcelain flawless skin (matte, smooth, glowing), naturally pretty no-makeup look: softly defined brows, glossy bare pink lips, bright clear eyes with strong catchlight, gentle innocent expression. Long straight black hair past shoulders, silky glossy, soft wispy curtain bangs / air bangs gently framing the small face, two thin face-framing strands at the cheeks. Three-view character sheet (front / three-quarter / profile), soft studio daylight, light grey backdrop, plain ivory cotton tee + soft beige cardigan, calm gentle gaze, lips softly closed, youthful pure mood. Photorealistic, 85mm portrait, shallow depth of field, cinematic soft color grading, warm fill light, gentle skin glow, sharp eye detail, refined Japanese editorial aesthetic. Reference: Hashimoto Kanna, Hamabe Minami, Mori Nana, Komatsu Nana (cute version), Japanese youthful actress beauty, single eyelid cute girl idol-actress crossover. Negative: double eyelid, big eyes, droopy eyes, swollen eyelid, dark circles, mature face, square jaw, wide face, large face, plain face, average looking, oily skin, anime style, plastic skin, heavy makeup.

Positive: Cinematic 16:9, locked static medium shot from a three-quarter side angle, head-to-mid-thigh framing. A young woman sits on a pink upholstered train bench, wearing white over-ear headphones, a black leather jacket over a blue and white striped collared shirt, a short black skirt with bare legs visible from mid-thigh down. Gaze cast softly downward into empty space. She holds perfectly still throughout — only the faintest natural breath in her shoulders, a barely perceptible side-to-side rock from the train's motion, and a slow natural blink once or twice. No smile, no frown, no eye-roll, eyes locked forward without darting. Hair stays settled around her face. Hands rest in her lap, fingers do not move. Behind her, the dark train window shows a soft continuous blur of exterior motion, indicating the train is moving. Warm-cool fluorescent carriage light falls evenly on her face and legs. Photoreal, quiet, contemplative late-night atmosphere. Silent video. Negative: camera pan, camera tilt, camera zoom, camera dolly, any camera movement, framing change, framing tightens, framing widens, she turns her head, she looks at camera, she looks sideways, head rotation, eye darting, eye-roll, eye glance, sudden gaze shift, she smiles, she frowns, she sighs visibly, she touches face, she adjusts headphones, hand lifts to ear, she takes off headphones, she pulls out phone, hand reaching into pocket, hands lifting from lap, finger fidgeting, she shifts posture, she crosses legs differently, she uncrosses legs, legs lifting, knee bouncing, she leans forward, she leans back, skirt riding up dramatically, jacket falling off shoulder, other passengers entering frame, passenger walking past behind, reflection of another person in window, station appearing in window, station signs visible, train stopping, doors opening, exaggerated train shake, jerky motion, ceiling lights flickering, color shift, warm orange tone, daylight, sound, audio, music, dialogue, voiceover

镜头1（5s） 剧情：开场确立环境与主角的孤独状态，营造夜间街头的静谧感。 景别：远景 角色动作：静静坐在洗衣店内 情绪：孤独，平静 场景标签：夜景街道，自助洗衣店，冷白光 光影氛围：环境暗调，洗衣店散发冷白荧光，远处有微弱便利店暖光，孤寂感 音效：留2秒环境声后起远处车声、风声、自动门隐约的“叮”声 台词：“这家洗衣店，不知道为什么我挺喜欢的。” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：远景，固定机位] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]透过玻璃窗可见坐在店内的微小身影 + [主体空间关系：占据画面极小比例，洗衣店作为发光体位于画面视觉中心，周围是黑暗空旷的街道] + [静止的等待姿态] + [夜晚寂静的街道，老旧洗衣店亮着招牌，远处微弱便利店灯光] + [冷暖光影对比，大面积暗部包围高光] + [日系写实胶片，电影夜景质感] + [35mm镜头，f/2.8] ▼ 视频动作提示词： 固定镜头。环境保持静止，微风吹动街道上的细小尘埃或树叶，洗衣店内部的荧光灯有极其轻微的闪烁感，无强烈动作，营造极致的静谧氛围。时长5秒。 镜头2（4s） 剧情：视角带入，展现内部空间质感，建立观众与主角的视觉连接。 景别：中景 角色动作：安坐在椅子上 情绪：安心，平淡 场景标签：洗衣机排，反光地板，室内 光影氛围：冷白荧光灯主导，地面反光强烈，空间空旷 音效：滚筒水声起、荧光灯嗡鸣声 台词：“灯太亮了，这样反而安心。” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：中景，带透视纵深的机位] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]安静地坐在画面深处的黄色塑料椅上 + [主体空间关系：左侧是一排老旧滚筒洗衣机，右侧和前方是空旷反光的过道，视角模拟刚进门的人] + [安静放松的坐姿] + [一排散发冷白光的滚筒洗衣机，水磨石反光地板] + [顶部冷调荧光灯直射，光斑反射在地板上] + [写实生活感，纵深构图] + [50mm镜头，f/4.0] ▼ 视频动作提示词： 摄像机从门口缓慢向前平稳推进，模拟第一人称走入店内的视角。滚筒洗衣机内部的灯光微微闪亮，水磨石地板的反光随镜头推进产生自然位移。时长4秒。 镜头3（6s） 剧情：展示主角正在进行的日常小动作，增加生活实感。 景别：中景 角色动作：低头撕开饭团包装 情绪：专注，日常 场景标签：黄色塑料椅，饭团，旋转的滚筒 光影氛围：顶光照明，背景有动态光源闪烁 音效：塑料包装窸窣声 台词：“便利店的鲑鱼饭团，” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：中景侧面，固定机位] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]坐在洗衣机前破旧的黄色塑料椅上，微微低头注视双手 + [主体空间关系：人物居中，背景充满庞大的工业感洗衣机] + [双手正拿着一个刚撕开一半包装的鲑鱼饭团] + [黄色塑料椅，正在旋转的滚筒洗衣机作为动态背景] + [冷调顶光，柔和的人物面部反光] + [安静的日常感，胶片色彩] + [50mm镜头，f/2.8] ▼ 视频动作提示词： 固定镜头。角色保持轻微的呼吸感，双手慢慢撕开饭团塑料包装，背景的洗衣机滚筒在稳定缓慢地旋转，衣物在内部翻滚。时长6秒。 镜头4（4s） 剧情：放大手部的微小动作，传达内心的短暂停顿和犹豫。 景别：特写 角色动作：指尖停顿，拿着饭团 情绪：微小的犹豫 场景标签：饭团特写，海苔，双手 光影氛围：局部光影，浅景深 音效：包装定格声、背景滚筒声 台词：“偶尔会咬到一颗特别咸的米。” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：手部特写，极浅景深] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]的双手 + [主体空间关系：手部和饭团占据画面核心，背景完全虚化] + [双手捧着鲑鱼饭团，深色海苔边缘露出，指尖悬停在半空中呈现犹豫姿态] + [半开的便利店饭团包装] + [顶灯在塑料包装上打出锐利高光] + [微观写实质感] + [100mm微距镜头，f/1.8] ▼ 视频动作提示词： 摄像机极缓慢地向前微推。双手拿着饭团保持几乎静止的定格状态，指尖仅有因悬停带来的极轻微不可察觉的颤动，塑料包装上的高光点随微推缓慢滑过。时长4秒。 镜头5（5s） 剧情：视觉化台词中的‘停一下’，表现思绪的抽离。 景别：近景侧颜 角色动作：咬一口饭团，咀嚼并停顿 情绪：思绪飘远，放空 场景标签：侧脸，咀嚼 光影氛围：侧顶光勾勒轮廓，背景暗调 音效：轻微咀嚼声、背景滚筒声 台词：“不是不好吃。就是会停一下。” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：近景侧面特写] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]的侧脸轮廓 + [主体空间关系：面部特写占据右侧，左侧留白] + [她轻轻咬下一小口饭团，眼神不聚焦地看着虚空，表情呈现突然的停顿] + [隐约的洗衣机边缘作为前景遮挡] + [冷白荧光灯从斜上方打亮下巴、鼻尖和嘴角的轮廓线，形成美丽的侧光] + [电影级情绪刻画] + [85mm镜头，f/2.0] ▼ 视频动作提示词： 固定镜头。角色嘴巴微动轻咬一口饭团，嚼动两下后，动作明显且僵硬地停顿下来，眼睑微垂，眼神失去焦点，维持一种思绪飘向远方的静止状态长达三秒。时长5秒。 镜头6（4s） 剧情：利用物品状态进行情绪留白，水雾倒影增强朦胧感。 景别：特写 角色动作：无 情绪：迷茫，流转 场景标签：洗衣机内部，彩色衣服，水雾玻璃 光影氛围：水雾反光，幽暗中带有色彩 音效：水声大起来一秒，衣服打在筒壁的“啪”声 台词：无 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：物件特写，正面机位] + [无角色正面，仅有倒影] + [主体空间关系：洗衣机圆形玻璃舱门填满画面大半部分] + [花花绿绿的衣物在滚筒内堆叠纠缠，玻璃门上凝结着细密的水雾，水雾表面隐约映出小夜穿着粉色衬衫的模糊轮廓倒影] + [玻璃舱门边缘的金属质感，内部昏暗的光线] + [冷光透过水雾产生漫反射] + [高度写实的水汽质感] + [50mm镜头，f/2.8] ▼ 视频动作提示词： 固定镜头。滚筒洗衣机内部剧烈旋转，花色衣物不断翻滚并拍打在玻璃门内侧。玻璃门上的细小水珠随着震动微微往下滑落，倒映在玻璃上的模糊人影也随之轻微晃动。时长4秒。 镜头7（5s） 剧情：建立内外空间的联系，引入路过的生命体增加对比。 景别：中景 角色动作：背影静坐 情绪：旁观，孤独 场景标签：玻璃门，街道，橘猫 光影氛围：室内冷光与室外夜灯交织 音效：街道风声、远处自行车铃声 台词：“玻璃门外，一只橘猫慢慢走过去，看都没看我。” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：中景，带前景遮挡的通透构图] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]的背影作为虚化的前景 + [主体空间关系：前景是虚化的人物背影与玻璃门窗框，焦点在玻璃门外的街道上，画面右下角有一只胖橘猫] + [静坐的背影] + [夜晚街道，反光的玻璃门，路面，一只橘猫] + [室内高反差冷光与室外路灯的暖黄光在玻璃上形成交错] + [空间层次分明] + [35mm镜头，f/2.8] ▼ 视频动作提示词： 固定镜头。前景的人物背影保持静止。焦点清晰处的门外街道上，一只橘猫迈着不紧不慢的步伐，匀速从画面右侧横穿走向左侧画外，完全没有转头看向店内。时长5秒。 镜头8（4s） 剧情：展现角色极微小的情绪转变，一丝被治愈的释然。 景别：近景 角色动作：抬头视线跟随，微牵嘴角 情绪：释然，微小喜悦 场景标签：抬头，眼神跟随 光影氛围：眼神光清晰，面部光影柔和 音效：荧光灯嗡鸣声 台词：“嗯…也对。它有它要去的地方。” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：正面近景特写] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]微微抬头的面容 + [主体空间关系：面部居中偏右，视线向画面左侧延伸] + [她抬头视线平行移动，眼球转动跟随画外的猫，嘴角牵起一丝几乎难以察觉的极微弱笑意] + [柔和的室内背景] + [眼底清晰地反射出门外暖黄色路灯的光点，面部照明层次丰富] + [细腻的情感流露] + [85mm镜头，f/1.8] ▼ 视频动作提示词： 固定镜头。角色的眼球缓慢从右向左移动，视线跟随画外事物。随后面部肌肉微微松弛，嘴角浮现出一丝不到一秒钟的极淡微笑，随后又收回表情恢复平静。时长4秒。 镜头9（3s） 剧情：物理状态改变，打破沉思，宣告洗衣结束。 景别：近景 角色动作：从下方抬眼看机器 情绪：清醒 场景标签：指示灯熄灭，机器面板 光影氛围：红光转暗，冷感机械 音效：洗衣机“叮”一声、嗡鸣声消失 台词：无 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：近景俯视/平视机器面板] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]的头部从画面底部边缘露出一小部分 + [主体空间关系：机器控制面板占据主画面，人物在下方边缘抬起视线仰望] + [视线注视面板] + [老旧的滚筒洗衣机控制面板，上方有一颗红色的指示灯刚熄灭的暗红色残留] + [机械冰冷感照明] + [局部细节特写] + [50mm镜头，f/4.0] ▼ 视频动作提示词： 固定镜头。滚筒洗衣机内部的旋转渐渐减速直至完全停止平息，机器顶部的红色指示灯闪烁一下后熄灭。画面底端角色的视线随之抬起，注视着停下的机器。时长3秒。 镜头10（5s） 剧情：传递物品带来的物理温暖，形成治愈感的情绪高潮。 景别：中景 角色动作：蹲下抱衣服，脸贴上去 情绪：温暖，满足 场景标签：拥抱衣物，蹲姿 光影氛围：顶光带来的柔软阴影，温馨感 音效：布料窸窣声 台词：“刚烘好的衣服抱在怀里，是热的。” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：中景，蹲姿构图] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]蹲在打开舱门的洗衣机前 + [主体空间关系：人物蜷缩在画面正中，洗衣机大门敞开在身侧] + [她双手紧紧抱着一大团刚烘好的花色蓬松衣服，闭着眼睛将侧脸深深贴在衣服堆里蹭着，表情极为放松温柔] + [敞开的滚筒，柔软的衣物材质] + [暖调与冷调交界，侧顶光打在头发与毛茸茸的衣服上形成柔和光晕] + [温馨治愈的情绪感] + [50mm镜头，f/2.8] ▼ 视频动作提示词： 固定镜头。角色从洗衣机里抱出一大团衣物拥入怀中，然后缓慢地低下头，闭上眼睛，将脸颊温柔且依恋地蹭向衣服表面，动作充满对温暖的沉醉感。时长5秒。 镜头11（4s） 剧情：强调离开空间的过渡，从室内再次切入室外环境。 景别：特写 角色动作：走出门槛 情绪：坚定，前行 场景标签：帆布鞋，门槛，自动门 光影氛围：室内外光线交界，反差大 音效：自动门“哗”开合声、外面街头环境声涌进来 台词：无 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：低机位脚部特写] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]的双脚下半部分 + [主体空间关系：一双脚正跨过画面中央的门槛界限] + [迈步行走动作，裤脚微微飘动] + [老旧的室内水磨石瓷砖地，铝合金门槛，深灰色的室外柏油路面，边缘有正在滑动的玻璃自动门] + [室内高光与室外深色地面的强对比照明] + [充满行进感与过渡感] + [35mm镜头，f/4.0] ▼ 视频动作提示词： 低机位跟随平移。摄像机紧跟穿着白色帆布鞋的双脚，拍摄其踩过室内瓷砖、跨出门槛踩在粗糙柏油路面上的连续动作。同时，画面边缘的自动玻璃门伴随风声从两侧向中间滑动闭合。时长4秒。 镜头12（6s） 剧情：留白收尾，表现生活继续向前的余韵。 景别：远景 角色动作：抱着衣服背影远去 情绪：平静，希望 场景标签：夜街，背影，路灯 光影氛围：路灯序列延伸，夜空深邃 音效：风声、鞋底摩擦声、远处电车声 台词：“明天，想买点葡萄。” 角色【小夜】：[小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围] ▼ 图片生成提示词： [画面构图：远景空镜，中心透视] + [小夜: 年轻亚洲女性，黑色短发，穿着粉色长袖衬衫内搭白色打底，下身穿着浅蓝色牛仔背带裤，脚穿白色帆布鞋，神情平静带着一丝孤独与惬意，日常文艺氛围]抱着衣服的背影 + [主体空间关系：人物背影位于街道中央偏下，由于距离很远而显得微小，道路向画面尽头延伸] + [匀速步行的背影姿态] + [空旷无人的城市夜景街道，两排向远处延伸的暖黄色路灯，树影婆娑] + [路灯在柏油路上投下长长的斜影，冷暖光线交融于夜色] + [悠长的电影片尾留白质感] + [50mm镜头，f/5.6] ▼ 视频动作提示词： 固定镜头。人物抱着衣服的背影在笔直的街道上不紧不慢地向前走去，身形在镜头中逐渐缩小变远。街道两旁的暖黄路灯有极轻微的频闪感，微风吹动路边的树叶沙沙作响，最后两秒画面中人物极小，留出安静的空余时间。时长6秒。

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 50mm lens, natural skin pores and texture, real fabric weave. Absolutely no illustration, anime, manga, cartoon, 3D render, CGI, painted, or stylized output. Create a 16:9 character identity board. Subject: use the reference face image for facial identity, use the reference outfit image for clothing — do not invent, add, or modify any clothing details, follow the outfit reference exactly. Pure white / soft cream studio backdrop. No environment, no extra props, no logos, no watermark. Design direction: cinematic identity board, photographic editorial layout — like a fashion magazine character study spread. Asymmetric, elegant, generous whitespace, varied image ratios, intentional imbalance. No grids, no blueprint, no catalog repetition. CRITICAL LAYOUT RULES: no overlapping character images, clear separation between every view, no cropped faces, no hidden limbs, no stacked figures, no merged poses. Main composition: one large hero full-body view, slightly off-center as visual anchor. Around it, smaller supporting studies with clean spacing — neutral full body, back view, side profile, seated, leaning, crouched, top-down body angle, low-angle body shot, expressive portrait studies. Identity lock across all views: same face, same hairstyle (Japanese-style straight short bob, blunt fringe, ear-covering, layered thinned ends — Aoi Yu / Mitsushima Hikari), same outfit (from reference), same body proportions, same visual personality. Include a small silhouette study area with 2–3 simplified black character silhouettes (silhouettes only — keep the rest photographic). Include a small expression study area with subtle emotional variation. Include a small detail study area showing key visual features of face, hair, and the outfit from reference. Text design: stylish character ID block, minimal, bold, editorial. Use only: NAME: — ROLE: Voice Diarist CORE MOOD: sunlit blank / floating VISUAL SIGNATURE: Japanese short bob with blunt fringe + white wired in-ear earphones Subtle handwritten labels only where helpful. Fine editorial arrows / annotation marks, minimal. Style: photorealistic, real photography, cinematic editorial, high-end fashion magazine, natural skin texture, real fabric, real lighting, NOT illustration, NOT anime, NOT 3D render, NOT painted.

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 50mm lens, natural skin pores and texture, real fabric weave. Absolutely no illustration, anime, manga, cartoon, 3D render, CGI, painted, or stylized output. Create a 16:9 character identity board. Subject: use the reference face image for facial identity, use the reference outfit image for clothing — do not invent, add, or modify any clothing details, follow the outfit reference exactly. Pure white / soft cream studio backdrop. No environment, no extra props, no logos, no watermark. Design direction: cinematic identity board, photographic editorial layout — like a fashion magazine character study spread. Asymmetric, elegant, generous whitespace, varied image ratios, intentional imbalance. No grids, no blueprint, no catalog repetition. CRITICAL LAYOUT RULES: no overlapping character images, clear separation between every view, no cropped faces, no hidden limbs, no stacked figures, no merged poses. Main composition: one large hero full-body view, slightly off-center as visual anchor. Around it, smaller supporting studies with clean spacing — neutral full body, back view, side profile, seated, leaning, crouched, top-down body angle, low-angle body shot, expressive portrait studies. Identity lock across all views: same face, same hairstyle (Japanese-style straight short bob, blunt fringe, ear-covering, layered thinned ends — Aoi Yu / Mitsushima Hikari), same outfit (from reference), same body proportions, same visual personality. Include a small silhouette study area with 2–3 simplified black character silhouettes (silhouettes only — keep the rest photographic). Include a small expression study area with subtle emotional variation. Include a small detail study area showing key visual features of face, hair, and the outfit from reference. Text design: stylish character ID block, minimal, bold, editorial. Use only: NAME: — ROLE: Voice Diarist CORE MOOD: night-walk solitude VISUAL SIGNATURE: Japanese short bob with blunt fringe + white wired in-ear earphones Subtle handwritten labels only where helpful. Fine editorial arrows / annotation marks, minimal. Style: photorealistic, real photography, cinematic editorial, high-end fashion magazine, natural skin texture, real fabric, real lighting, NOT illustration, NOT anime, NOT 3D render, NOT painted.

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 50mm lens, natural skin pores and texture, real fabric weave. Absolutely no illustration, anime, manga, cartoon, 3D render, CGI, painted, or stylized output. ABSOLUTE FACE LOCK: the face in every single view of this board must be IDENTICAL to the reference face image — same face shape, same eye shape and spacing, same nose, same lips, same jawline, same skin tone, same eyebrow shape, same overall facial proportions. Do NOT generalize, do NOT beautify, do NOT change ethnicity, do NOT swap to a different model. Treat the reference face as a hard constraint, not a suggestion. Create a 16:9 character identity board. Subject: use the reference face image for facial identity (HIGHEST priority), use the reference outfit image for clothing — do not invent, add, or modify any clothing details, follow the outfit reference exactly. Pure white / soft cream studio backdrop. No environment, no extra props, no logos, no watermark. Design direction: cinematic identity board, photographic editorial layout — like a fashion magazine character study spread. Asymmetric, elegant, generous whitespace, varied image ratios, intentional imbalance. No grids, no blueprint, no catalog repetition. CRITICAL LAYOUT RULES: no overlapping character images, clear separation between every view, no cropped faces, no hidden limbs, no stacked figures, no merged poses. Every face visible in the board must clearly match the reference face — if a face deviates from the reference, the output is invalid. Main composition: one large hero full-body view, slightly off-center as visual anchor. Around it, smaller supporting studies with clean spacing — neutral full body, back view, side profile, seated, leaning, crouched, top-down body angle, low-angle body shot, expressive portrait studies. Identity lock across all views: SAME FACE (matching the reference image exactly), same hairstyle (Japanese-style straight short bob, blunt fringe, ear-covering, layered thinned ends — Aoi Yu / Mitsushima Hikari), same outfit (from outfit reference), same body proportions, same visual personality. Include a small silhouette study area with 2–3 simplified black character silhouettes (silhouettes only — keep the rest photographic). Include a small expression study area with subtle emotional variation, faces still matching the reference. Include a small detail study area showing key visual features of face, hair, and the outfit from reference. Text design: stylish character ID block, minimal, bold, editorial. Use only: NAME: — ROLE: Voice Diarist CORE MOOD: quiet drift VISUAL SIGNATURE: Japanese short bob with blunt fringe + white wired in-ear earphones Subtle handwritten labels only where helpful. Fine editorial arrows / annotation marks, minimal. Style: photorealistic, real photography, cinematic editorial, high-end fashion magazine, natural skin texture, real fabric, real lighting, NOT illustration, NOT anime, NOT 3D render, NOT painted.

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 50mm lens, natural skin pores and texture, real fabric weave. Absolutely no illustration, anime, manga, cartoon, 3D render, CGI, painted, or stylized output. Create a 16:9 character identity board. Subject: use the reference face image for facial identity, use the reference outfit image for clothing details. Pure white / soft cream studio backdrop. No environment, no props, no logos, no watermark. Design direction: cinematic identity board, photographic editorial layout — like a fashion magazine character study spread. Asymmetric, elegant, visually memorable. Generous whitespace, varied image ratios, intentional imbalance. No grids, no blueprint, no catalog repetition. CRITICAL LAYOUT RULES: no overlapping character images, clear separation and breathing room between every view, no cropped faces, no hidden limbs, no stacked figures, no merged poses. Main composition: one large hero full-body view, slightly off-center as visual anchor. Around it, smaller supporting studies with clean spacing — neutral full body, back view, side profile, seated, leaning, crouched, top-down body angle, low-angle body shot, expressive portrait studies. Identity lock across all views: same face, same facial proportions, same hairstyle (Japanese-style straight short bob, blunt fringe, ear-covering, layered thinned ends — Aoi Yu / Mitsushima Hikari reference), same outfit, same body proportions, same posture language, same visual personality. OUTFIT (lock across all views): black thin-strap ribbed tank top, oversized light-wash baggy wide-leg jeans sitting on the hip with visible waistband, white low-top canvas sneakers, navy denim hair scarf tied as a top-knot bow, white wired in-ear earphones with cable hanging down the chest, small slate-blue suede shoulder bag with thin strap. Slim build, low body fat, visible collarbone. Reference details: clean face shape, clean hair silhouette, clean outfit silhouette, clean body shape, clean hands, clean poses, clean expression range. Include a small silhouette study area with 2–3 simplified black character silhouettes (silhouettes only — keep the rest photographic). Include a small expression study area with subtle emotional variation (detached, faint smirk, blank, looking down). Include a small detail study area showing the denim hair scarf knot, the white earphone cable, the tank top straps, and the suede bag. Text design: stylish character ID block, minimal, bold, editorial. Use only: NAME: — ROLE: Voice Diarist CORE MOOD: sharp / detached VISUAL SIGNATURE: Japanese short bob with blunt fringe + white wired in-ear earphones Subtle handwritten labels only where helpful. Allow fine editorial arrows / annotation marks, kept minimal. Style: photorealistic, real photography, cinematic editorial, high-end fashion magazine, natural skin texture, real fabric, real lighting, NOT illustration, NOT anime, NOT 3D render, NOT painted.

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 50mm lens, natural skin pores and texture, real fabric weave. Absolutely no illustration, anime, manga, cartoon, 3D render, CGI, painted, or stylized output. Create a 16:9 character identity board. Subject: use the reference face image for facial identity, use the reference outfit image for clothing — do not invent, add, or modify any clothing details, follow the outfit reference exactly. Pure white / soft cream studio backdrop. No environment, no extra props, no logos, no watermark. Design direction: cinematic identity board, photographic editorial layout — like a fashion magazine character study spread. Asymmetric, elegant, generous whitespace, varied image ratios, intentional imbalance. No grids, no blueprint, no catalog repetition. CRITICAL LAYOUT RULES: no overlapping character images, clear separation between every view, no cropped faces, no hidden limbs, no stacked figures, no merged poses. Main composition: one large hero full-body view, slightly off-center as visual anchor. Around it, smaller supporting studies with clean spacing — neutral full body, back view, side profile, seated, leaning, crouched, top-down body angle, low-angle body shot, expressive portrait studies. Identity lock across all views: same face, same hairstyle (Japanese-style straight short bob, blunt fringe, ear-covering, layered thinned ends — Aoi Yu / Mitsushima Hikari), same outfit (from reference), same body proportions, same visual personality. Include a small silhouette study area with 2–3 simplified black character silhouettes (silhouettes only — keep the rest photographic). Include a small expression study area with subtle emotional variation. Include a small detail study area showing key visual features of face, hair, and the outfit from reference. Text design: stylish character ID block, minimal, bold, editorial. Use only: NAME: — ROLE: Voice Diarist CORE MOOD: guarded warmth VISUAL SIGNATURE: Japanese short bob with blunt fringe + white wired in-ear earphones Subtle handwritten labels only where helpful. Fine editorial arrows / annotation marks, minimal. Style: photorealistic, real photography, cinematic editorial, high-end fashion magazine, natural skin texture, real fabric, real lighting, NOT illustration, NOT anime, NOT 3D render, NOT painted.

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 50mm lens, natural skin pores and texture, real fabric weave. Absolutely no illustration, anime, manga, cartoon, 3D render, CGI, painted, or stylized output. Create a 16:9 character identity board. Subject: use the reference face image for facial identity, use the reference outfit image for clothing — do not invent, add, or modify any clothing details, follow the outfit reference exactly. Pure white / soft cream studio backdrop. No environment, no extra props, no logos, no watermark. Design direction: cinematic identity board, photographic editorial layout — like a fashion magazine character study spread. Asymmetric, elegant, generous whitespace, varied image ratios, intentional imbalance. No grids, no blueprint, no catalog repetition. CRITICAL LAYOUT RULES: no overlapping character images, clear separation between every view, no cropped faces, no hidden limbs, no stacked figures, no merged poses. Main composition: one large hero full-body view, slightly off-center as visual anchor. Around it, smaller supporting studies with clean spacing — neutral full body, back view, side profile, seated, leaning, crouched, top-down body angle, low-angle body shot, expressive portrait studies. Identity lock across all views: same face, same hairstyle (Japanese-style straight short bob, blunt fringe, ear-covering, layered thinned ends — Aoi Yu / Mitsushima Hikari), same outfit (from reference), same body proportions, same visual personality. Include a small silhouette study area with 2–3 simplified black character silhouettes (silhouettes only — keep the rest photographic). Include a small expression study area with subtle emotional variation. Include a small detail study area showing key visual features of face, hair, and the outfit from reference. Text design: stylish character ID block, minimal, bold, editorial. Use only: NAME: — ROLE: Voice Diarist CORE MOOD: soft mid-afternoon VISUAL SIGNATURE: Japanese short bob with blunt fringe + white wired in-ear earphones Subtle handwritten labels only where helpful. Fine editorial arrows / annotation marks, minimal. Style: photorealistic, real photography, cinematic editorial, high-end fashion magazine, natural skin texture, real fabric, real lighting, NOT illustration, NOT anime, NOT 3D render, NOT painted.

Use the reference image exactly for the FACE only: - Same facial features: monolid almond eyes with slight cat-eye lift, small straight nose, lip shape with defined cupid's bow, V-line jaw, beauty marks, skin tone, baby fat on cheeks - Keep the same age and beauty level (22yo, top-tier visual) CHANGE the hair: - Color: glossy jet black (NOT pink, NOT pastel) - Style: short straight bob at jawline / chin length, blunt full bangs to eyebrows (hime-style fringe), hair ends gently curling inward to frame the face, sleek polished texture with high shine Three-view character sheet: front view / three-quarter view / profile view. Plain ivory cotton tee + soft beige cardigan (NOT green tank top, NO necklace, NO earrings). Soft studio daylight, neutral light grey backdrop, plain background. Calm gentle gaze, lips softly closed, natural neutral expression. Photorealistic, 85mm portrait, f/1.4 shallow depth of field, sharp facial detail, cinematic soft color grading, warm soft fill light, luminous skin glow, high-end Japanese editorial portrait quality. Negative: pink hair, pastel hair, colored hair, light hair, different face, green clothing, jewelry, accessories, desert background, outdoor background, heavy makeup change, double eyelid change, anime style, plastic skin.

Enhance ONLY the eyes and lashes, keep everything else identical. Reference the original character's eye style: - Long elongated almond eyes with subtle cat-eye lift - LONG dense black eyelashes (upper lashes thick and sweeping outward, lower lashes visible and defined root by root) - Sharp clean lash-line creating natural eyeliner effect - Eye outer corner pulled slightly upward and outward - Bright clear iris with strong catchlight in pupil - Subtle pink-brown eyeshadow with fine shimmer on lid - Light aegyo-sal under eyes (small puffy under-eye for youthful charm) - Gaze: cool, slightly aloof, with quiet intensity (not blank, not smiling — composed and watching) Keep identical: - Same face shape, nose, lips, jawline - Same black short bob with blunt bangs - Same ivory tee and beige cardigan - Same three-view composition (front / 3-4 / profile) - Same skin tone and lighting - Same neutral grey backdrop Photorealistic, ultra sharp eye detail, high-end Japanese editorial portrait, preserve the original character identity. Negative: different face, different person, double eyelid, drastic eye shape change, fake doll-like lashes, anime eyes, heavy eyeshadow, smiling, blank stare.

A 22-year-old Japanese girl, breathtaking ethereal beauty, goddess-tier visuals, top 0.01% face, flawless facial proportions (golden ratio), extremely refined delicate features. Soft elegant monolid eyes with long thick natural lashes, slightly upturned outer corners, clean defined double-eyelid-free crease, straight high-bridged nose, sharp refined nose tip, small full lips with defined cupid's bow, glossy bare pink lip color. Very small face, sharp V-line jaw, tiny pointed chin, defined cheekbones with soft contour, slight youthful baby fat on cheeks, flawless porcelain skin with luminous inner glow (matte, poreless, dewy translucence). Naturally pretty no-makeup look, softly defined arched brows, bright clear eyes, strong catchlight, gentle elegant expression. Long straight black hair past shoulders, silky lustrous, soft wispy curtain bangs framing the face, two thin face-framing strands at the cheeks. Three-view character sheet (front / three-quarter / profile), soft studio daylight, light grey backdrop, plain ivory cotton tee + soft beige cardigan, calm gentle gaze, lips softly closed. Photorealistic, 85mm portrait, f/1.4 shallow depth of field, cinematic editorial color grading, warm soft fill light, luminous skin glow, ultra sharp eye and skin detail, high-end Japanese fashion magazine cover quality. Reference: top-tier Japanese actress visuals, Komatsu Nana + Hashimoto Kanna + Hamabe Minami genetic best mix, monolid version, VOGUE Japan / GINZA / Numero Tokyo cover beauty, idol-actress visual queen tier. Negative: double eyelid, big round eyes, droopy eyes, swollen eyelid, mature face, plain face, average looking, ordinary, square jaw, wide face, large face, asymmetric features, anime, plastic skin, heavy makeup, low-tier idol look.

Glossy jet black hair, short straight bob cutting at jawline / just below chin, blunt full bangs (hime-style straight fringe) sitting just above the eyebrows, hair ends gently curling inward to frame the face, two short strands tucked at the cheeks, sleek polished texture with high shine. 整段 prompt 拼起来—— A 22-year-old East Asian girl, breathtaking ethereal beauty, goddess-tier visuals, top 0.01% face, flawless facial proportions. Long elongated almond eyes, soft monolid / subtle inner-fold, slightly upturned outer corners (subtle cat-eye lift), defined sharp eyeliner-like lash line, long thick natural lashes, soft pink-brown eyeshadow with subtle glitter, glossy lower-lash catchlight, light aegyo-sal under eyes. Small straight nose with refined high bridge and tiny tip, small full lips with sharp defined cupid's bow, plump lower lip, glossy nude-pink lip tint. Very small face, sharp clean V-line jaw, tiny pointed chin, soft high cheekbones with hint of baby fat, luminous porcelain skin (matte-to-dewy, poreless, soft inner glow, blush on cheeks and nose tip). Glossy jet black hair, short straight bob cutting at jawline / just below chin, blunt full bangs (hime-style straight fringe) sitting just above the eyebrows, hair ends gently curling inward to frame the face, two short strands tucked at the cheeks, sleek polished texture with high shine. Three-view character sheet (front / three-quarter / profile), soft daylight, neutral light backdrop, plain ivory cotton tee + soft beige cardigan, calm gentle gaze, lips softly closed, youthful refined mood. Photorealistic, 85mm portrait, f/1.4 shallow depth of field, cinematic editorial color grading, warm soft fill light, luminous skin glow, ultra sharp eye and skin detail, high-end fashion magazine cover quality, Korean-Japanese idol-actress visual style. Reference: K-pop visual member tier, Japanese gravure idol crossover, youthful refined monolid beauty, 2024 editorial portrait aesthetic. Negative: pink hair, colored hair, long hair, medium hair, large round double eyelid eyes, droopy eyes, swollen eyelid, mature face, plain face, average looking, square jaw, wide face, large face, anime, plastic skin, heavy makeup, oily skin, low-tier idol look, AI-generated stock face.

Use the reference image's FACE exactly: - Same facial features: monolid almond eyes with cat-eye lift, small refined nose, lip shape, V-line jaw, skin tone, beauty marks, cool composed expression - Same age (early 20s) and beauty level HAIR — Japanese layered straight short cut: - Glossy jet black, completely straight (NO curl, NO flip, NO wave) - Length: chin to jaw level - Layered cut with varied lengths: surface strands slightly longer than inner strands - Hair falls straight down naturally, every strand vertical - Ends are texturized with thinning shears: feathered, wispy, piecey, slightly uneven - NOT a blunt one-length cut — ends are softly tapered - Subtle thinned texture throughout, airy lightness - Blunt bangs to eyebrows, with a few thinned wispy strands breaking the bottom line - Reference: Aoi Yu in "Hana and Alice", Mizuhara Kiko early career short hair, Japanese salon textured straight short cut Three-view character sheet: front view / three-quarter view / profile view. Plain ivory cotton tee + soft beige cardigan. Soft studio daylight, neutral light grey backdrop. Calm gentle gaze, lips softly closed, cool composed expression with quiet intensity. Photorealistic, 85mm portrait, f/1.4 shallow depth of field, sharp facial detail, cinematic soft color grading, luminous skin glow, high-end Japanese editorial quality. Negative: curly hair, wavy hair, inward curl, outward flip, bob with rounded ends, helmet hair, blunt one-length cut, long hair, colored hair, different face, desert background, double eyelid.

Use the reference image's FACE exactly: - Same facial features: monolid almond eyes with cat-eye lift, small refined nose, lip shape, V-line jaw, skin tone, beauty marks, expression character - Same age (early 20s) and beauty level (top-tier visual) CHANGE the hair to layered Japanese short cut: - Glossy jet black hair - Length: chin to jaw, with short-to-shorter layers - Layered texture: surface layers slightly longer, inner layers shorter, soft wispy feathered ends - Hair tips are piecey and irregular (not blunt straight cut) - Slight volume at the crown, soft movement on sides - Ends slightly textured and choppy, gently flicking outward in random directions (controlled messiness) - Blunt full bangs to eyebrows, but with slight thinning and a few wispy strands breaking the line - Style: French crop / Japanese textured short cut / refined wolf-cut hybrid, salon-fresh editorial cut - Reference: Japanese magazine hair editorial 2024, layered short hair with airy texture Three-view character sheet: front view / three-quarter view / profile view. Plain ivory cotton tee + soft beige cardigan. Soft studio daylight, neutral light grey backdrop. Calm gentle gaze, lips softly closed, cool composed expression with quiet intensity. Photorealistic, 85mm portrait, f/1.4 shallow depth of field, sharp facial detail, cinematic soft color grading, luminous skin glow, high-end Japanese editorial quality. Negative: blunt bob, smooth one-length cut, helmet hair, inward curl, outward flip, wavy hair, curly hair, long hair, medium hair, colored hair, pink hair, different face, different person, desert background, outdoor background, heavy makeup change, double eyelid.

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 35mm or 50mm lens, shallow depth of field, natural skin texture, real fabric weave. NO illustration, anime, cartoon, 3D render, CGI, painted, stylized. Cinematic still from a quiet contemporary Japanese drama series, in the visual mood of Kore-eda Hirokazu, Hamaguchi Ryusuke, or NHK late-night dramas. Naturalistic, observational, slightly melancholic warmth. A young Japanese woman sits on a worn plastic chair against the wall of a late-night Japanese coin laundromat, holding a small triangular convenience-store salmon rice ball (wrapped in nori and plastic film) in both hands, eating slowly and absent-mindedly. She is the only customer. Slight slouch, knees together, feet relaxed on the linoleum floor, looking softly forward — not at the camera. A faint, unguarded expression, lost in her own thoughts. Setting: real Japanese coin laundromat at night. A row of front-loading washing machines and dryers fills the background, one drum slowly turning with clothes inside. Worn pale linoleum floor, faded yellow plastic chairs, vending-style detergent dispenser, small handwritten Japanese signs on the wall, fluorescent ceiling tubes mixed with warm tungsten accents. Faint reflections on the glass front. Empty street visible faintly through the window. Identity lock: face must match the reference face image exactly — same face shape, same eyes, same nose, same lips, same skin tone. Do not generalize or beautify. Asian Japanese features. Hair: Japanese-style straight short bob, blunt fringe, ear-covering, layered thinned ends (Aoi Yu / Mitsushima Hikari reference). OUTFIT (lock exactly, do not invent or change): sheer pale pink long-sleeve shirt worn unbuttoned and loose, white cropped ribbed tank top underneath, light-wash baggy wide-leg jeans sitting on the hip, white low-top canvas sneakers. Slim build, low body fat, visible collarbone. Composition: medium-wide shot, eye-level or very slight low angle, subject seated to the left third of the frame with washing machines as horizontal layered background to the right, generous breathing room above her head, 16:9 cinematic crop. Lighting: warm tungsten key from inside the laundromat softly illuminating her face and hands, subtle cool fluorescent fill from the ceiling, gentle rim light from the dryer drums behind, no harsh shadows, soft contrast, slightly faded film tone (mild green-and-amber color grading, low saturation, fine 35mm film grain). Style: photorealistic, real photography, cinematic editorial, Japanese late-night TV drama aesthetic, natural skin tone, real fabric, real lighting, quiet observational framing.

Horizontal 16:9 cinematic composition. EXTREME WIDE LONG SHOT — camera positioned far back, the street stretches deep into the distance. CRITICAL — BODY PROPORTIONS LOCK: The character's body proportions MUST EXACTLY match the FIRST reference image (character body board). Replicate her exact silhouette: tall slim model build, narrow shoulders, long slender legs, elongated torso, 170cm height proportions, fashion-model elongation. Do not redraw her as shorter, do not make her stockier, do not widen her shoulders, do not shorten her legs. Even at small scale in the frame, preserve the SAME tall narrow elongated silhouette as the reference body board. SCENE LOCK: Strictly follow the SECOND reference image (storyboard) for: street scenery, lighting tone, color palette, streetlamp arrangement, building style, camera angle, depth of field, atmosphere. Replicate the storyboard scene as closely as possible. Do not invent a different scene. OUTFIT LOCK: The outfit MUST exactly match the THIRD reference image (costume): oversized PINK button-up shirt draped loosely on slim frame, BLUE DENIM JEANS (full-length, straight or slightly loose, covering full legs to the ankles), white canvas low-top sneakers, black straight hair with blunt bangs (shoulder-length, viewed from behind). LIGHTING: Naturalistic photographic light, real-world ambient streetlamp glow, soft falloff, documentary-style realism. NOT stylized, NOT plastic, NOT 3D-rendered, NOT over-glowing. POSE: Place her in the CENTER of the frame, seen FROM BEHIND ONLY (full back view, NO face, NO side profile, NO turning around), walking AWAY from camera into the deep distance. She holds a bundle of folded laundry against her chest with both arms. Her figure is SMALL in the frame — occupies only about 1/12 to 1/15 of frame height — a tiny TALL NARROW silhouette in the middle distance, dwarfed by the long street perspective. Even at this small size, she clearly reads as TALL and SLIM (not as a small chubby blob), her elongated proportions visible from her silhouette alone. The street and streetlamps dominate the composition. Long deep one-point perspective showing many streetlamps receding into the distance. 35mm cinematic, film grain, soft contrast, deep DOF, naturalistic photographic look. Aspect ratio 16:9. Negative： short figure, stocky build, plump body, chubby, fat, overweight, wide shoulders, broad frame, heavy build, thick legs, short legs, wide hips, round body, square silhouette, child-like proportions, body lost in clothing, shrunken proportions when small, blob-like silhouette, pixelated body losing proportions, ignoring body reference, redrawing body shape, changing body proportions from reference, different body type from reference, bare legs, bare thighs, exposed legs, no pants, leggings, tights, stockings, shorts, mini skirt, skirt, dress, hem reaching mid-thigh, plastic lighting, fake lighting, over-rendered, CG light, 3D render, cartoonish glow, over-saturated lights, neon glow, anime lighting, video game lighting, stylized lighting, fake lens flare, halo around lamps, exaggerated bloom, medium shot, close-up, mid-distance shot, figure too large, figure dominating the frame, figure occupying more than 1/8 of frame height, vertical composition, 9:16, portrait orientation, front view, side profile, three-quarter view, face visible, turning around, looking back, different street, different buildings, modern city, neon signs, traffic, motorcycles, bicycles, cars, pedestrians, animals, black shirt, white shirt, blue shirt, jacket, coat, school uniform, high heels, boots, dark shoes, sandals, branded shoes with logos, short hair, long hair past waist, ponytail, bun, dyed hair, brown hair, blonde hair, no laundry bundle, holding bag, holding phone, empty hands, multiple people, second person, group walking, rain, snow, fog effects

She slowly takes one small, deliberate bite of the onigiri. Subtle jaw movement from chewing only, then pauses. Lips remain closed throughout. Mouth stays shut. Eyes drift slightly downward during the pause. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： talking, speaking, lip sync, mouth movement for speech, mouthing words, dialogue, parted lips, opening mouth wide, sighing, exhaling, breathing heavily, camera movement, zoom in, zoom out, pan, dolly, exaggerated chewing, multiple bites, head turning, dramatic expression, eye contact with camera, dramatic music sting

She slowly takes one small, deliberate bite of the onigiri. Subtle jaw movement as she chews once, then pauses. Eyes drift slightly downward during the pause. Static locked camera, no zoom, no pan, no camera shake. Silent video. Negative： camera movement, zoom in, zoom out, pan, dolly, exaggerated chewing, mouth wide open, sighing, head turning, multiple bites, dramatic expression, eye contact with camera

正向： The orange tabby cat walks calmly across the visible street area outside the glass door, moving steadily from right to left at a relaxed pace, side profile maintained throughout. The cat does not look toward the camera, does not stop, does not change pace. The girl in the blurred foreground stays completely still, only her shoulders rise and fall almost imperceptibly from breathing. Subtle warm reflections shift slightly on the glass door surface. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： cat looking at camera, cat stopping, cat sitting down, cat turning head, cat running, multiple cats, cat changing direction, girl turning around, girl moving, girl getting up, girl reaching for cat, face visible, sharp foreground, people walking past, cars passing, motorcycles, glass door opening, hand on glass, talking, speaking, lip sync, mouthing words, dialogue, parted lips, camera movement, zoom in, zoom out, pan, dolly, camera shake,

正向： The washer drum rotates steadily. Colorful clothes tumble and softly slap against the drum wall as they fall through the rotation. Water droplets on the foggy glass door slowly trickle downward. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： human reflection, face reflection, person reflected on glass, silhouette on glass, figure visible through glass, ghostly figure, face emerging from fog, camera angle changing, view shifting to frontal, person walking past, hand reaching in, clothes flying out, drum stopping, drum opening, camera movement, zoom in, zoom out, pan, dolly, camera shake, talking, speaking, dialogue, dramatic music sting

Horizontal 16:9 cinematic composition. TASK: Add a single human figure into the FIRST reference image (empty street scene). Do NOT redraw, recolor, or modify the scene in any way — preserve the first image EXACTLY as it is. SCENE PRESERVATION (CRITICAL): The FIRST reference image is the empty street background. Keep its lighting, streetlamps, buildings, asphalt, color palette, perspective, atmosphere, depth of field, and grain 100% identical to the original. Do not invent new buildings, do not change lamp colors, do not shift the camera angle, do not alter the existing light sources. Only ADD the figure described below; everything else must remain pixel-faithful to the first image. FIGURE TO ADD: A SLIM TALL young Japanese woman positioned in the CENTER of the frame, seen FROM BEHIND ONLY (full back view, NO face, NO side profile, NO turning around), walking AWAY from camera into the deep distance of the street. BODY PROPORTIONS LOCK (CRITICAL — MAKE HER EXTRA TALL AND THIN): The figure must be HIGH-FASHION RUNWAY MODEL build — even taller and slimmer than the body board reference: 175cm+ height proportions, EXTREMELY narrow shoulders, very long slender legs (legs accounting for ~62% of total height), elongated willowy torso, lean wrists and ankles, supermodel elongation. Picture a Tokyo runway model walking down the street. Her silhouette must read as TALL + WILLOWY + SLENDER even at small scale — the dominant visual impression is "long vertical line", never "compact shape". OUTFIT LOCK: Oversized PINK button-up shirt draped loosely on slim frame, BLUE DENIM JEANS (full-length, straight or slightly loose, covering full legs to the ankles), white canvas low-top sneakers, black straight hair with blunt bangs (shoulder-length, viewed from behind). She holds a bundle of folded laundry against her chest with both arms. SIZE & PLACEMENT: She occupies only about 1/12 to 1/15 of frame height — a tiny TALL NARROW silhouette in the middle distance. Place her standing on the road surface in the center of the depth, naturally integrated into the existing perspective. LIGHTING ON FIGURE (CRITICAL — KEEP HER DIM): She is in the SHADOW area of the street, BETWEEN the pools of streetlamp light, not directly under any lamp. Her body is dimly lit by ambient bounce light only — UNDEREXPOSED relative to the streetlamp pools, blending into the dim mid-tones of the street. Her shirt's pink reads as a desaturated muted dusty pink, not bright. Her jeans read as deep dark navy-blue, not vivid. Her white sneakers read as dim off-white grey, not glowing. She is roughly 1.5 to 2 stops darker than the brightest streetlamp pools. The dominant light from the lamps stays on the road and walls — she sits in the cooler dimmer shadow zone in between. Her overall silhouette tonality: dark mid-grey, almost reading as a near-silhouette against the warmer lamp-lit road behind her. INTEGRATION: Match her color temperature and shadow direction to the existing scene. Add a soft natural shadow under her feet matching the streetlamp light direction. 35mm cinematic, film grain, naturalistic photographic look matching the first reference image's grain and contrast. Aspect ratio 16:9. Negative： brightly lit figure, glowing figure, overlit body, spotlight on character, hero lighting, key light on character, sunlit figure, figure brighter than surroundings, vivid clothing colors, saturated pink, neon pink, bright denim, glowing white sneakers, character popping out of scene, character standing out, isolated stage lighting, studio lighting on figure, beauty lighting, rim light, halo around figure, glow effect, bloom on character, short figure, stocky build, plump body, chubby, fat, wide shoulders, broad frame, thick legs, short legs, wide hips, round body, square silhouette, child-like proportions, blob-like silhouette, body lost in clothing, shrunken proportions when small, pixelated body losing proportions, average proportions, normal build, athletic build, muscular figure, ignoring body reference, average height, medium height, short legs to torso ratio, normal leg-to-body ratio, redrawing the scene, modifying the background, changing streetlamps, changing buildings, changing lamp colors, shifting camera angle, altering perspective, adding new buildings, removing buildings, changing road texture, recoloring the scene, ignoring the first reference image, plastic lighting on figure, CG-rendered figure, cartoonish figure, anime figure, 3D-rendered figure, mismatched lighting, floating figure, no shadow, hovering, pasted-on look, bare legs, bare thighs, exposed legs, no pants, leggings, tights, stockings, shorts, mini skirt, skirt, dress, hem reaching mid-thigh, medium shot of figure, close-up of figure, figure too large, figure dominating the frame, figure occupying more than 1/8 of frame height, vertical composition, 9:16, portrait orientation, front view, side profile, three-quarter view, face visible, turning around, looking back, black shirt, white shirt, blue shirt, jacket, coat, school uniform, high heels, boots, dark shoes, sandals, branded shoes with logos, short hair, long hair past waist, ponytail, bun, dyed hair, brown hair, blonde hair, no laundry bundle, holding bag, holding phone, empty hands, multiple people, second person, group walking, rain, snow, fog effects

正向： The orange tabby cat walks calmly across the visible street area outside the glass door, moving steadily from LEFT to RIGHT at a relaxed pace, side profile maintained throughout, body facing right. The cat does not look toward the camera, does not stop, does not change pace. The girl in the blurred foreground stays completely still, only her shoulders rise and fall almost imperceptibly from breathing. The lighting outside the glass door stays completely constant throughout the entire shot — no light changes, no new light sources appear, no glow, no flare, no illumination shifts. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： cat moving right to left, cat walking backwards, cat changing direction mid-walk, new light appearing, light beam, light streak, glowing trail behind cat, light following cat, headlight, car light, street lamp turning on, sudden illumination, lens flare, light burst, glow effect, bloom, light leak, sudden brightness change, exposure shift, cat looking at camera, cat stopping, cat sitting down, cat turning head, cat running, multiple cats, girl turning around, girl moving, girl getting up, girl reaching for cat, face visible, sharp foreground, people walking past, cars passing, motorcycles, glass door opening, hand on glass, talking, speaking, lip sync, mouthing words, dialogue, parted lips, camera movement, zoom in, zoom out, pan, dolly, camera shake,

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 35mm lens, natural film grain, real fabric, real light. NO illustration, anime, cartoon, 3D render, CGI, painted, stylized. Cinematic establishing shot from a quiet Japanese late-night drama (Kore-eda / Hamaguchi mood). Wide shot of an empty residential street at night, taken from across the road. On the opposite sidewalk sits a small Japanese coin laundromat — its glowing white-warm interior visible through a large glass storefront. The signage reads "コインランドリー" in plain modern Japanese type. Inside, very small in frame, a single tiny figure sits on a chair, barely identifiable — just a silhouette of a young woman with a short bob. Setting: real Japanese suburban street at night. Asphalt road with faint puddle reflections, a few parked bicycles, vending machine glowing softly further down, a power pole with tangled wires, low residential houses in the distance. No other people. No cars. Quiet. The figure inside the laundromat (matching the reference face and identity for continuity, even though small in frame): same young Japanese woman, Japanese-style straight short bob with blunt fringe. Wearing the same outfit as the reference identity board: sheer pale pink shirt unbuttoned, white cropped tank, light-wash baggy wide-leg jeans, white sneakers. She is seated, holding a small triangular convenience-store rice ball. Interior of the laundromat must match the reference scene image: row of front-loading washing machines and dryers along the back wall, faded yellow plastic chairs, fluorescent ceiling tubes mixed with warm tungsten, worn pale linoleum floor. Composition: wide establishing shot, eye-level, laundromat occupying the right two-thirds of the frame, empty street and night sky filling the left third for breathing room. 16:9. Lighting: deep night blue tone outside, warm amber laundromat interior glowing as the only bright pocket of light in the frame. Strong contrast between the cold dark street and the warm safe interior. Faint reflections of the laundromat light on the wet asphalt. Mood: lonely but not sad. Protective. Like a small lit lantern in a dark neighborhood. Style: photorealistic, real photography, cinematic editorial, Japanese late-night TV drama aesthetic, 35mm film grain, slight green-amber color grade, low saturation, soft contrast.

LOCKED-OFF STATIC CAMERA, tripod, absolutely no camera movement. NO pan, NO tilt, NO zoom, NO dolly, NO handheld breathing. The frame stays pixel-identical to the starting image throughout. Camera at strict EYE-LEVEL horizontal angle to the hands. Within the static frame, the motion choreography in two beats: BEAT 1 (seconds 0–2): Her right thumb and forefinger continue pulling the red plastic pull-tab slowly downward along the center seam — the clear film parts further along the perforated line, the tear extends from one-third down to about two-thirds down the seam. The wrapper splits open visibly, the dark green nori-wrapped rice triangle becoming more exposed through the parted film. Subtle micro-crinkles on the plastic edges as it tears. BEAT 2 (seconds 2–6, HOLD): Her hands STOP completely mid-motion. The pull-tab is frozen near the bottom of the seam, the wrapper widely parted but not yet removed. Fingers do not move further. She holds this exact stillness for the rest of the shot — a long suspended pause. Action interrupted, suspended, unfinished. CONTINUOUS BACKGROUND MOTION: in the soft warm bokeh behind the hands, one washing machine drum tumbles steadily and continuously throughout — clothes rotating as round glowing shapes in the blur, never stopping. Lighting and color: warm tungsten glow steady, green-amber 35mm film grade, subtle film grain, no flicker. End frame: hands frozen exactly as in beat 2 — pull-tab paused near the bottom of the seam, plastic widely parted exposing the nori rice triangle, washer still tumbling in soft bokeh. Action unresolved. Duration: 6 seconds, contemplative pacing, NOT slow-motion. OUTPUT: silent video, no audio track needed. NEGATIVE: face entering frame, head appearing, chin appearing, camera tilting up, camera tilting down, camera moving, pan, tilt, zoom, dolly, handheld, shake, slow motion, fully unwrapping, removing wrapper completely, peeling plastic off, hands continuing motion after pause, hands leaving frame, washer stopped, washer empty, static drum, anime, cartoon, 3D, CGI, dramatic music sting.

【生成内容】 - 光：户外巷子傍晚蓝白冷光 + 夕阳从右侧斜打，红铁楼梯阴影斜落 / 室内冰箱冷白光柱 + 纱窗夕阳暖橙打地板成格，写实纪录片质感（必写 naturalistic photographic light, NOT plastic/CG） - 关键节拍：镜 3「正好十二步」脚步定 / 镜 6 BOSS 牛奶咖啡罐特写（→ S2 闭环，与 S2 镜 12 同一瓶）/ 镜 9 枯花收尾 - 比例 16:9，silent video，日式老旧居家家居造型 - 通用 negative：sighing / parted lips / exhaling / plastic lighting / eye movement / blinking-heavy / smile / head turn / look at camera / streetlight halo / over-ear headphones --- 镜 1 · 巷子远景建立（2s） - 景别：极远景，固定机位，巷子纵深 - 内容：日式老住宅区窄巷，左侧水泥围墙+灌木+一棵大树压顶，右侧两层木造公寓挂红色户外铁楼梯，远端尽头停一辆白色面包车和自行车 - 起始帧：纯空巷，无人 - 节奏：0-1s 静止，1-2s 她从巷子尽头走出，剪影小 - 光：傍晚天色蓝白偏冷，夕阳从巷子右侧斜进来，铁楼梯阴影斜落地面 - 喂图：场景图第一位（锁构图），脸图第二位，故事板第三位 - 注意：远景小人，必走 ADD 模式（先擦背景车/人，再合成她） 镜 2 · 红铁楼梯爬升·中近景（2.5s） - 景别：腰上中景，侧后跟拍 - 内容：她踩着红色铁楼梯往上走，铁栏杆条纹从画面右侧贯穿，巷子虚化在身后 - 起始帧：脚刚踩上第三阶，手轻搭红色栏杆 - 动作：自然爬两阶，肩线松 - 光：夕阳从巷子那侧侧打过来，半边脸亮半边脸暗（光过脸） - Negative：no head turn, no eye to camera 镜 3 · 楼梯坏灯段+数步（1.5s） - 景别：低角度脚部特写，俯视斜 30° - 内容：她的鞋踩在铁楼梯踏板上，前方一段顶灯不亮，踏板由亮到暗 - 起始帧：右脚刚踩上铁阶，前方阶面发暗 - 动作：两步落下，第二步停住，对位"正好十二步" - 注意：节拍点，定格用作"门口"的隐含转场 镜 4 · 走向冰箱·过肩背拍（2s） - 景别：过肩背拍中景，室内 - 内容：日式狭长厨房，木地板，台面有滤水杯/电水壶，屋内幽暗，她背影走向深处的单门白色冰箱，冰箱白光是屋里唯一光源 - 起始帧：背影距冰箱 3 步 - 动作：自然走两步，肩线松 - Kling 提示：背后极慢推 - Negative：no plastic light, no over-bright kitchen lighting 镜 5 · 开冰箱·正面近景（2.5s） - 景别：胸上近景，平视，冰箱内部反打 - 内容：冰箱冷白光打满脸，半边亮半边暗，眼神往下沉，没表情，手伸进去 - 起始帧：冰箱门刚开 30°，光刚扫上脸 - 动作：极小幅度——手伸入 - 光：冷白光柱 - Negative 关键：no facial expression change, no smile, no eye movement 镜 6 · BOSS 牛奶咖啡罐特写 · 冰箱内部视角（3s · 必给） - 景别：冰箱内部视角，平视微仰 - 内容：BOSS 牛奶咖啡罐独自立在冰箱最里层，罐身水汽在冷光下凝出 - 起始帧：罐子静止，她的手指刚进画框 - 动作：手指碰上罐身 - 喂图：S2 镜 12 取物口那张第一位（锁颜色/品牌/标签），BOSS 罐特写第二位，故事板第三位 - 埋点：S2 闭环点，给足时长 镜 7 · 纱窗夕阳全景（2.5s） - 景别：固定全景，客厅 - 内容：榻榻米或老木地板客厅，整面纱窗发着暖橙，地板被切成一格一格光，房间空 - 起始帧：光的格子在地板上已成形 - 动作：纯静止，云移过让光的格子缓慢挪 - 光：纱窗夕阳暖橙，与镜 5-6 冰箱冷白形成对比 - Negative：no zoom, no pan 镜 8 · 地板侧背 · 横移到侧脸（3s） - 景别：从背到侧脸的腰上中景，缓慢横移 - 内容：她坐在木地板上，光的格子打在背和侧脸，手抱膝，BOSS 罐放在地板上没开 - 起始帧：背对镜头，光的格子已在背上 - 动作：镜头缓慢横移，她不动 - 注意：对位"有只猫的话该有多好" - Negative 关键：no eye movement, no blink-heavy 镜 9 · 空镜 · 窗台枯花（2.5s） - 景别：木窗台特写，逆光 - 内容：一盆早已枯掉的花在棕色陶盆里，叶子是焦的，木窗框边缘清晰，房间没人 - 起始帧：花静止 - 动作：纯静止，结束 - 埋点：对位"我连花都养不活"

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY. Shot on full-frame camera, 35mm lens, natural skin texture, real fabric weave. NO illustration, anime, cartoon, 3D render, CGI, painted, stylized. Cinematic still from a quiet Japanese short film. A young Japanese woman crouches on the floor of a late-night Japanese coin laundromat, hugging a small triangular convenience-store salmon rice ball wrapped in nori and plastic, eating slowly. She is the only customer. Setting: Japanese coin laundromat at night, row of front-loading washing machines and dryers as background, drum doors visible, warm tungsten light mixed with pale fluorescent, worn linoleum floor, empty plastic chairs along the wall, soft reflections on glass. Identity lock: face must match the reference face image exactly — same face shape, same eyes, same nose, same lips, same skin tone. Do not generalize or beautify. Hair: Japanese-style straight short bob, blunt fringe, ear-covering, layered thinned ends (Aoi Yu / Mitsushima Hikari reference). OUTFIT (lock exactly, do not invent or change): sheer pale pink long-sleeve shirt worn unbuttoned and loose, white cropped ribbed tank top underneath, light-wash baggy wide-leg jeans sitting on the hip, white low-top canvas sneakers. Slim build, low body fat, visible collarbone. Composition: medium shot, slight low angle, subject roughly centered with breathing space to her left, washing machines as horizontal layered background, 16:9. Lighting: warm tungsten key from inside, subtle cool fluorescent fill, soft glow on her face and the rice ball, gentle shadow under her crouch. Style: photorealistic, real photography, cinematic editorial, Japanese indie film aesthetic (Kore-eda / Hamaguchi mood), natural skin tone, real fabric, real lighting.

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY, single-frame cinematic still, 16:9 aspect ratio. 35mm film grain, green-amber color grade, Japanese late-night TV drama look (NHK deep-night, Kore-eda / Hamaguchi reference). NO anime, NO illustration, NO 3D render, NO grid, NO multi-panel, NO collage. Medium shot from INSIDE the laundromat, slight low angle, eye-level toward the entrance. A young Japanese woman pushes the glass door open and steps in — caught mid-motion, one hand on the brushed aluminum door handle, the other holding a small beige canvas tote with folded laundry inside. Soft pink button-up shirt over white inner top, simple light-wash wide-leg jeans. Straight short black hair, blunt fringe, white wired in-ear earphones with visible cable trailing to her pocket. Her face is partially visible — three-quarter angle, eyes softly lowered, calm neutral expression, no smile. Same character as the reference, do not change facial features. Cold blue night air spills in from the open door behind her, contrasting the warm tungsten interior glow. Edge of a washing machine and the bench visible in foreground out of focus. A small "コインランドリー" sign reflection on the glass. CORE MOOD: quiet entry, private routine, slightly melancholic — like coming home to a familiar small ritual. Composition: subject centered slightly off-axis, shallow-medium depth, warm interior bokeh, cold exterior gradient through the door. 35mm equiv, f/2.0, gentle bloom on tungsten bulbs, soft halation. OUTPUT: ONE single photographic frame only. NEGATIVE: storyboard, panels, grid, collage, multi-shot layout, anime, illustration, art book, animation studio, cartoon, 3D render, harsh light, smile, dramatic pose, action movie staging.

正向： She continues walking steadily AWAY from the camera, deeper into the street, at a slow calm natural pace. Her figure gradually becomes smaller in the frame as she moves into the distance. Both arms remain folded around the bundle of laundry against her chest, her posture relaxed. The streetlamps remain steady, their warm amber glow constant. A subtle gentle wind moves a few strands of her hair and the loose hem of her oversized shirt. After approximately 4 seconds she is noticeably smaller in frame, then for the final 2 seconds the camera holds on the empty deepening street with her tiny figure still visible far away. Static locked camera, no zoom, no pan, no camera shake, no dolly. OUTPUT: silent video, no audio track needed. Negative： camera dollying in, camera following her, camera tracking, zoom in, zoom out, pan, tilt, camera shake, drone movement, turning around, looking back, stopping, hesitating, sitting down, dropping the laundry, running, jogging, jumping, dramatic gesture, laundry falling, clothes flying, items dropped on ground, new people appearing, cars driving by, motorcycles, bicycles, cat appearing, dog appearing, streetlamps flickering, lights turning off, sudden darkness, lightning, sudden brightness change, exposure shift, lens flare, light burst, rain starting, snow falling, fog rolling in, talking, speaking, lip sync, dialogue, dramatic music sting

LOCKED-OFF STATIC CAMERA, tripod-mounted, absolutely no camera movement throughout the entire shot. NO pan, NO tilt, NO zoom in, NO zoom out, NO dolly, NO truck, NO crane, NO handheld breathing, NO parallax, NO push-in, NO pull-out. The frame stays pixel-identical to the starting image from beginning to end. Within the static frame, the only motion is the woman: - She walks slowly from frame-left along the far sidewalk toward the laundromat entrance at a calm natural pace, tote bag swaying gently with her steps. - She reaches the glass door, pauses for a brief beat (about half a second), then pushes the door open with her free hand. - The door swings inward, warm tungsten light briefly widens onto the wet asphalt outside as the door opens. - She steps inside, the glass door slowly closes behind her. - Inside the laundromat, she is now seen as a small distant figure through the glass, walking a few steps toward the wooden bench. Ambience: faint distant city hum, soft footsteps on damp asphalt, gentle door hinge sound when she opens it, no music, no dialogue, no traffic, no other pedestrians anywhere in frame. Lighting and color: tungsten storefront glow remains steady and warm, navy night exterior remains cool and quiet, green-amber 35mm film grade holds throughout, subtle film grain, no flicker, no color shift. End frame: glass door fully closed, the small figure of the woman is visible inside the laundromat through the glass walking toward the bench, street outside empty again, identical framing as the opening shot. Duration: 5 seconds, smooth real-time pacing, no slow-motion, no time-lapse. NEGATIVE: camera pan, camera tilt, camera zoom, dolly, crane, handheld, shake, push-in, pull-out, parallax shift, rack focus, whip pan, fast cuts, multiple shots, scene change, slow motion, time lapse, anime, cartoon, 3D, CGI, dramatic music sting, lens flare streak.

LOCKED-OFF STATIC CAMERA, tripod-mounted, absolutely no camera movement. NO pan, NO tilt, NO zoom, NO dolly, NO handheld breathing. The frame stays pixel-identical to the starting image throughout. Within the static frame, the only motion is the hands and the onigiri wrapper: - Her right thumb and forefinger gently grip the small red plastic pull-tab at the top of the triangular wrapper. - She slowly pulls the tab downward along the center seam of the plastic, the clear film parting cleanly along the perforated line, revealing the dark green nori sheet and white rice underneath. - The plastic film crinkles softly as it separates, tiny natural micro-movements of the wrapper edges. - Her left hand steadies the bottom of the triangle, fingertips lightly pressing the side of the rice ball. - The motion is slow, deliberate, unhurried — a private careful little ritual, not rushed. KEY SOUND: crisp plastic wrapper rustling and crinkling — the unmistakable konbini onigiri unwrap sound. The plastic film is the dominant audio. Faint background hum of washing machine drum tumbling far behind. Ambience: only the soft plastic crinkle and the distant washer hum, no music, no dialogue, no other sound. Lighting and color: warm tungsten interior glow remains steady, green-amber 35mm film grade holds throughout, subtle film grain, no flicker. End frame: the plastic wrapper pull-tab strip has come fully down the seam, the two halves of the clear film are slightly parted exposing the nori-wrapped rice triangle, her hands holding it ready to peel the plastic open in the next shot. Duration: 5 seconds, slow contemplative pacing, NOT slow-motion (no frame interpolation), just a naturally unhurried small action. NEGATIVE: camera pan, tilt, zoom, dolly, crane, handheld, shake, parallax, rack focus, whip pan, fast cuts, multiple shots, slow motion effect, time lapse, anime, cartoon, 3D, CGI, hands trembling, dropping onigiri, biting, eating, talking, music, hand leaving frame.

LOCKED-OFF STATIC CAMERA, tripod, absolutely no camera movement. NO pan, NO tilt, NO zoom, NO dolly, NO handheld breathing. The frame stays pixel-identical to the starting image throughout. Within the static frame, the motion choreography in two beats: BEAT 1 (seconds 0–2): Her right thumb and forefinger continue pulling the red plastic pull-tab slowly downward along the center seam of the triangular onigiri wrapper — the clear film parts further along the perforated line, the tear extends from one-third down to about two-thirds down the seam. The wrapper splits open visibly, the dark green nori-wrapped rice triangle becoming more exposed through the parted film. Subtle micro-crinkles on the plastic edges as it tears. BEAT 2 (seconds 2–6, HOLD): Her hands STOP completely mid-motion. The pull-tab is frozen near the bottom of the seam, the wrapper widely parted but not yet removed. Fingers do not move further. She holds this exact stillness for the rest of the shot — a long suspended pause. Eyes stay softly lowered on the half-opened wrapper, expression calm and slightly distant. Action interrupted, suspended, unfinished. Her mouth stays completely closed and still throughout — NO sighing, NO exhaling, NO parted lips, NO heavy breathing, NO chest heaving. Face calm and neutral, head gently tilted down toward her hands. CONTINUOUS BACKGROUND MOTION: behind her in soft warm bokeh, one washing machine drum tumbles steadily and continuously throughout — clothes rotating, gentle motion blur, never stopping. Lighting and color: warm tungsten glow steady, green-amber 35mm film grade, subtle film grain, no flicker. End frame: hands frozen exactly as in beat 2 — pull-tab paused near the bottom of the seam, plastic widely parted exposing the nori rice triangle, washer still tumbling. Action unresolved. Duration: 6 seconds, contemplative pacing, NOT slow-motion. OUTPUT: silent video, no audio track needed. NEGATIVE: sighing, exhaling, breathing heavily, mouth opening, parted lips, talking, biting, eating, fully unwrapping, removing wrapper completely, peeling plastic off, hands continuing motion after pause, hands leaving frame, washer stopped, washer empty, static drum, no background motion, camera movement, pan, tilt, zoom, dolly, handheld, shake, slow motion, anime, cartoon, 3D, CGI.

上一版的 i2v prompt 不用改，直接用就行。重新贴一遍方便复制： LOCKED-OFF STATIC CAMERA, tripod-mounted, absolutely no camera movement. NO pan, NO tilt, NO zoom, NO dolly, NO handheld breathing. The frame stays pixel-identical to the starting image throughout. Within the static frame, the motion choreography in two beats: BEAT 1 (seconds 0–1.5): Her right thumb and forefinger gently grip the small red plastic pull-tab at the top of the triangular wrapper resting in her lap. She slowly pulls the tab partway down the center seam, the clear film starting to part along the perforated line. Crisp plastic crinkle and rustle. BEAT 2 (seconds 1.5–6, HOLD): Her hands STOP completely mid-action. The pull-tab is frozen halfway down the seam, the wrapper half-opened. Her fingers do not move further. Her face remains softly lowered toward the onigiri in her lap, gaze fixed on the half-opened wrapper, expression calm and slightly distant. The plastic crinkle sound CUTS OFF abruptly. She holds this stillness for the rest of the shot. CONTINUOUS BACKGROUND MOTION (entire shot): One washing machine drum tumbles steadily and continuously in the background bokeh — clothes rotating, gentle motion blur, never stopping. The drum's low rhythmic hum carries through. KEY SOUND DESIGN: - Seconds 0–1.5: crisp plastic wrapper rustling, dominant. - Second 1.5: plastic sound CUTS OFF cleanly, abrupt silence on the wrapper. - Seconds 1.5–6: only steady low hum and rhythmic tumble of the washing machine, slightly amplified. - No music, no dialogue. Lighting and color: warm tungsten glow steady, green-amber 35mm film grade, subtle grain. End frame: hands frozen as in beat 2, pull-tab paused halfway, plastic half-parted, washer still tumbling, her face still softly looking down at the unfinished action. Duration: 6 seconds, contemplative pacing, NOT slow-motion. NEGATIVE: camera movement, pan, tilt, zoom, dolly, handheld, shake, slow motion effect, fast unwrap, fully unwrapping, completing the unwrap, hands continuing motion after pause, washer stopped, washer empty, static drum, no background motion, biting, eating, talking, music, hand leaving frame, plastic sound continuing through pause, face cropped out, looking at camera, anime, cartoon, 3D, CGI.

LOCKED-OFF STATIC CAMERA, tripod-mounted, absolutely no camera movement. NO pan, NO tilt, NO zoom, NO dolly, NO handheld breathing. The frame stays pixel-identical to the starting image throughout. Within the static frame, the motion choreography: BEAT 1 (seconds 0–2): She sits still, gaze softly lowered on the wrapped onigiri cupped in her lap. Her right thumb very gently strokes the top edge of the triangular plastic wrapper once, a small idle touch — feeling the texture but not yet pulling the tab. The wrapper stays fully closed. BEAT 2 (seconds 2–4): She blinks once, slowly. Her left hand subtly adjusts the onigiri in her grip, settling it more comfortably in her lap, the wrapper remains intact. Eyes stay lowered on the rice ball. BEAT 3 (seconds 4–5): Her right thumb and forefinger move up to the small red plastic pull-tab at the top point of the wrapper, just barely beginning to pinch it — the action of pulling has NOT started yet, she is just about to. CONTINUOUS BACKGROUND MOTION: behind her, in the soft warm bokeh, one washing machine drum tumbles steadily and continuously throughout the entire shot — clothes rotating, gentle motion blur, never stopping. Her mouth stays completely closed and still throughout — NO sighing, NO exhaling, NO parted lips, NO heavy breathing, NO chest heaving. Face is calm and neutral, head stays gently tilted down toward her hands. Ambience: low rhythmic hum of the washing machine drum, faint plastic wrapper rustle on the thumb stroke, no music, no dialogue. Lighting and color: warm tungsten glow steady, green-amber 35mm film grade, subtle grain. End frame: her right thumb and forefinger pinching the red pull-tab at the top of the wrapper, ready to pull — wrapper still intact, washer still tumbling. (This end frame matches the START frame of SHOT 4.) Duration: 5 seconds, contemplative pacing, NOT slow-motion. NEGATIVE: sighing, exhaling, breathing heavily, mouth opening, parted lips, talking, biting, eating, fully unwrapping, opening wrapper, removing onigiri, hands leaving frame, washer stopped, washer empty, static drum, camera movement, pan, tilt, zoom, dolly, handheld, shake, slow motion, anime, cartoon, 3D, CGI, earphones appearing, cable appearing, touching ear.

LOCKED-OFF STATIC CAMERA, tripod-mounted, absolutely no camera movement. NO pan, NO tilt, NO zoom, NO dolly, NO handheld breathing. The frame stays pixel-identical to the starting image throughout. Within the static frame, the motion choreography in two beats: BEAT 1 (seconds 0–1.5): Her right thumb and forefinger gently grip the small red plastic pull-tab at the top of the triangular wrapper. She slowly pulls the tab partway down the center seam, the clear film starting to part along the perforated line. Crisp plastic crinkle and rustle. BEAT 2 (seconds 1.5–6, HOLD): Her hands STOP completely mid-action. The pull-tab is frozen halfway down the seam, the wrapper half-opened. Her fingers do not move further. The plastic crinkle sound CUTS OFF abruptly. She holds this exact stillness for the rest of the shot — a long suspended pause. Only the faintest breath, no hand motion at all. The unwrap action is interrupted, suspended, unfinished. CONTINUOUS BACKGROUND MOTION (entire shot): In the soft warm bokeh behind her hands, the washing machine drum tumbles steadily and continuously throughout — clothes rotating inside the glass door, gentle motion blur, never stopping. The drum's low rhythmic hum carries through the entire shot, becoming the only sound after the plastic crinkle stops. KEY SOUND DESIGN: - Seconds 0–1.5: crisp plastic wrapper rustling, dominant in mix. - Second 1.5: plastic sound CUTS OFF cleanly, no fade — abrupt silence on the wrapper. - Seconds 1.5–6: only the steady low hum and rhythmic tumble of the washing machine drum, slightly amplified now that the plastic stops. Subtle, almost meditative. - No music, no dialogue, no other sound. Lighting and color: warm tungsten glow steady, green-amber 35mm film grade, subtle grain, no flicker. End frame: hands frozen exactly as in beat 2 — pull-tab paused halfway, plastic half-parted, washer still tumbling behind. The action is unresolved. Duration: 6 seconds, contemplative pacing, NOT slow-motion. NEGATIVE: camera movement, pan, tilt, zoom, dolly, handheld, shake, slow motion effect, fast unwrap, fully unwrapping, completing the unwrap, hands continuing motion after pause, washer stopped, washer empty, static drum, no background motion, biting, eating, talking, music, hand leaving frame, plastic sound continuing through pause, anime, cartoon, 3D, CGI.

Wide establishing shot of a Japanese summer festival night street, slow dolly-in (camera moves forward gently along the street center, no rotation, no tilt). Red paper lanterns overhead sway slightly in the night breeze. Foreground girl with short black hair, dark casual outfit, standing back to camera at street center, remains still — only her hair and clothes hem catch a faint breeze. Background crowd in yukata walking naturally in both directions, stall lights flickering warm yellow and red, distant lanterns glowing. Warm red and amber color palette throughout, shallow depth of field stays consistent, no zoom, no focus rack. Cinematic 16:9, silent video.

Medium shot at a Japanese summer festival, slow lateral truck (camera glides slowly to the right past the crowd, no rotation). Foreground girl with short black hair stands among yukata-wearing crowd, only slight head turn — no eye movement. Crowd around her walks naturally, paper lanterns glow softly overhead. Warm red and amber tones, shallow depth of field consistent throughout, no zoom. Cinematic 16:9, silent video.

Medium close-up of a girl with short black hair against a backdrop of glowing red paper lanterns. Camera holds steady, very subtle push-in (no rotation, no tilt). Her hair drifts faintly in the night breeze, lips parted slightly, no eye movement. Lanterns sway gently behind her, bokeh stays consistent. Warm red and amber palette, soft night ambient light. Cinematic 16:9, silent video.

正向： She remains completely still, gaze fixed off-screen toward the right edge of the frame, lips closed throughout. Only the most subtle blink occurs once during the shot, and her shoulders rise and fall almost imperceptibly from quiet breathing. A single strand of her hair shifts very slightly from the laundromat's air movement. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： head turning, head movement, looking around, gaze shifting, eyes moving, looking at camera, talking, speaking, lip sync, mouth movement for speech, mouthing words, parted lips, opening mouth, sighing, exhaling, breathing heavily, smiling, frowning, dramatic expression, emotional shift, hand entering frame, body leaning, posture change, camera movement, zoom in, zoom out, pan, dolly, camera shake, exposure shift, light change, lens flare, glow effect, dramatic music sting

正向： Her head stays completely still in pure side profile, eyes fixed steadily off-screen toward the right edge of the frame. She does not move at all — no head turn, no eye movement. Lips remain closed throughout. Mouth stays shut. Her shoulders rise and fall almost imperceptibly from quiet breathing. A single strand of her hair shifts very slightly from the laundromat's air movement. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： eyes moving, eyeballs rolling, gaze shifting, eye movement, looking around, darting eyes, blinking heavily, eyes closing, head turning, head movement, face rotating, profile changing, looking at camera, talking, speaking, lip sync, mouth movement for speech, mouthing words, parted lips, opening mouth, sighing, exhaling, smiling, frowning, dramatic expression, emotional shift, hand entering frame, body leaning, posture change, camera movement, zoom in, zoom out, pan, dolly, camera shake, exposure shift, light change, lens flare, glow effect, dramatic music sting

Wide shot of a girl with short black hair standing back-to-camera at the edge of a festival crowd, fireworks blooming in the distant night sky above. Slow dolly-in along the street center (no rotation, no tilt). Crowd in yukata moves naturally around her, paper lanterns sway overhead, distant fireworks bloom and fade in the deep background. She stays still, only hair and clothes hem catch breeze. Warm red and amber foreground, deep navy sky with gold sparks. Cinematic 16:9, silent video.

Sky-facing shot of fireworks bursting in the night sky, single firework blooms outward from center frame, sparks trail downward and fade. Camera locked off, no movement, no zoom. Deep navy night sky backdrop, warm red and gold spark tones. Cinematic 16:9, silent video.

Tight close-up of a girl with short black hair looking upward, faint reflections of fireworks flicker across her face and eyes. Camera holds steady with very slow push-in (no rotation, no tilt). Her expression shifts subtly from neutral to a faint quiet smile, lips part slightly — no eye movement, gaze stays fixed upward. Cool blue night ambient blends with warm amber rim light. Cinematic 16:9, silent video.

Extreme close-up of a girl's lower face and hand, hand resting near her chin, fingers still. Camera locked off, no movement. Her hair drifts faintly, lips remain neutral, no head turn. Soft warm amber light from the side, deep shadow on the opposite side, shallow depth of field. Cinematic 16:9, silent video.

Silhouette of a girl with short black hair on the right side of frame, facing left toward a large firework bursting in the background. Camera locked off, no movement, no zoom. Firework blooms once and fades, sparks drift downward slowly. Her silhouette stays still, only hair edge catches faint breeze. Deep navy sky, warm gold and red spark tones rim-lighting her edge. Cinematic 16:9, silent video.

She STROLLS slowly and leisurely AWAY from the camera, deeper into the quiet residential street, taking unhurried casual steps as if she has nowhere urgent to go — the relaxed evening pace of someone enjoying a quiet walk home. Each step is gentle and even, her arms lightly cradling the bundle of laundry against her chest, posture relaxed and slightly soft. Her figure becomes smaller in the frame VERY GRADUALLY over the full duration of the shot. At 5 seconds she should still be only moderately further than her starting position — she is not in a hurry. The streetlamps remain steady, their warm amber glow constant throughout, no flickering. A subtle gentle wind moves a few strands of her hair and the loose hem of her oversized shirt. Static locked camera, no zoom, no pan, no camera shake, no dolly. OUTPUT: silent video, no audio track needed. Negative： brisk walking, fast pace, hurried steps, rushing, jogging, power walking, quick stride, marching, athletic gait, military walk, determined stride, figure shrinking too fast, vanishing into distance, reaching the end of the street, disappearing, arms swinging energetically, purposeful posture, turning around, looking back, stopping, hesitating, sitting down, dropping the laundry, running, jogging, jumping, dramatic gesture, laundry falling, clothes flying, items dropped on ground, new people appearing, cars driving by, motorcycles, bicycles, cat appearing, dog appearing, camera dollying in, camera following her, camera tracking, zoom in, zoom out, pan, tilt, camera shake, drone movement, streetlamps flickering, lights turning off, sudden darkness, lightning, sudden brightness change, exposure shift, lens flare, light burst, rain starting, snow falling, fog rolling in, talking, speaking, lip sync, dialogue, dramatic music sting

Horizontal 16:9 cinematic composition. TASK: Add a single human figure into the FIRST reference image (empty street scene). Do NOT redraw, recolor, or modify the scene in any way — preserve the first image EXACTLY as it is. SCENE PRESERVATION (CRITICAL): The FIRST reference image is the empty street background. Keep its lighting, streetlamps, buildings, asphalt, color palette, perspective, atmosphere, depth of field, and grain 100% identical to the original. Do not invent new buildings, do not change lamp colors, do not shift the camera angle, do not alter the existing light sources. Only ADD the figure described below; everything else must remain pixel-faithful to the first image. FIGURE TO ADD: A SLIM TALL young Japanese woman positioned in the CENTER of the frame, seen FROM BEHIND ONLY (full back view, NO face, NO side profile, NO turning around), walking AWAY from camera into the deep distance of the street. BODY PROPORTIONS LOCK: The figure's body proportions MUST EXACTLY match the SECOND reference image (body board). Replicate her exact silhouette: tall slim model build, narrow shoulders, long slender legs, elongated torso, 170cm height proportions, fashion-model elongation. Even rendered small in the frame, preserve the SAME tall narrow elongated silhouette as the body reference — never shrink her into a stocky or chubby blob. OUTFIT LOCK: The outfit MUST exactly match the THIRD reference image (costume): oversized PINK button-up shirt draped loosely on slim frame, BLUE DENIM JEANS (full-length, straight or slightly loose, covering full legs to the ankles), white canvas low-top sneakers, black straight hair with blunt bangs (shoulder-length, viewed from behind). She holds a bundle of folded laundry against her chest with both arms. SIZE & PLACEMENT: She occupies only about 1/12 to 1/15 of frame height — a tiny TALL NARROW silhouette in the middle distance, dwarfed by the long street perspective. Place her standing on the road surface in the center of the depth, naturally integrated into the existing perspective. INTEGRATION (CRITICAL): Match her lighting, color temperature, shadow direction, and overall exposure to the existing scene's lighting in the first image — she must look photographed in that exact street, not pasted on. Add a soft natural shadow under her feet matching the streetlamp light direction. 35mm cinematic, film grain, naturalistic photographic look matching the first reference image's grain and contrast. Aspect ratio 16:9. Negative： redrawing the scene, modifying the background, changing streetlamps, changing buildings, changing lamp colors, shifting camera angle, altering perspective, adding new buildings, removing buildings, changing road texture, recoloring the scene, restyling the scene, ignoring the first reference image, short figure, stocky build, plump body, chubby, fat, wide shoulders, broad frame, thick legs, short legs, wide hips, round body, square silhouette, child-like proportions, blob-like silhouette, body lost in clothing, shrunken proportions when small, pixelated body losing proportions, ignoring body reference, redrawing body shape, changing body proportions from reference, bare legs, bare thighs, exposed legs, no pants, leggings, tights, stockings, shorts, mini skirt, skirt, dress, hem reaching mid-thigh, plastic lighting on figure, CG-rendered figure, cartoonish figure, anime figure, 3D-rendered figure, figure not matching scene lighting, mismatched lighting, floating figure, no shadow, hovering, pasted-on look, medium shot of figure, close-up of figure, figure too large, figure dominating the frame, figure occupying more than 1/8 of frame height, vertical composition, 9:16, portrait orientation, front view, side profile, three-quarter view, face visible, turning around, looking back, black shirt, white shirt, blue shirt, jacket, coat, school uniform, high heels, boots, dark shoes, sandals, branded shoes with logos, short hair, long hair past waist, ponytail, bun, dyed hair, brown hair, blonde hair, no laundry bundle, holding bag, holding phone, empty hands, multiple people, second person, group walking, rain, snow, fog effects

正向： Her LEFT foot completes its forward step, crossing over the metal threshold and landing softly on the dark asphalt road outside. The RIGHT foot then lifts from the interior tile and follows through, also stepping over the threshold onto the asphalt — a single continuous walking motion. Simultaneously, the glass automatic sliding door on the left edge of frame slowly slides closed, moving from left toward right, completing its closing motion as her second foot lands outside. The lighting transitions naturally as she steps out: warm fluorescent glow recedes, cool night street ambient light takes over on her shoes. Static locked camera at low angle, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： running, jumping, hesitating, stepping backwards, foot stomping, kicking, turning around, walking sideways, walking back inside, multiple steps, three or more steps, jogging, glass door opening, glass door reversing, glass door jolting, glass door bouncing back, hand entering frame, body bending down, person crouching, camera movement, zoom in, zoom out, pan, dolly, camera shake, camera following the feet, exposure burst, light flare, sudden brightness change, talking, speaking, dialogue, dramatic music sting

正向： She steps decisively across the threshold with a brisk, natural walking pace. Her LEFT foot quickly lifts and steps forward, landing firmly on the dark asphalt road outside within the first 1 second. Her RIGHT foot immediately follows through with a continuous brisk stride, crossing the threshold and landing on the asphalt by the second second. The motion is one fluid normal walking pace — not slow, not exaggerated — like an everyday casual step out the door. Simultaneously, the glass automatic sliding door on the left edge of frame slides closed at a normal speed, completing its closing motion as her second foot lands. Lighting transitions naturally from warm fluorescent inside to cool night street light outside. Static locked camera at low angle, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： slow motion, slow-motion effect, slow walking, hesitant step, dragging feet, stalling, freezing mid-step, time-lapse slow, running, jumping, stomping, kicking, turning around, walking sideways, walking back inside, glass door opening, glass door reversing, glass door slow motion, hand entering frame, body bending down, camera movement, zoom in, zoom out, pan, dolly, camera shake, camera following the feet, exposure burst, light flare, talking, speaking, dialogue, dramatic music sting

Horizontal 16:9 cinematic composition. EXTREME WIDE LONG SHOT — camera positioned far back, the street stretches deep into the distance. STRICTLY FOLLOW THE PROVIDED STORYBOARD REFERENCE IMAGE as the primary visual source for: street scenery, lighting tone, color palette, streetlamp arrangement, building style, camera angle, depth of field, atmosphere, overall composition. Replicate the storyboard reference as closely as possible. Do not invent a different scene. Do not redesign the street. LIGHTING MUST BE NATURALISTIC AND REALISTIC: real-world ambient streetlamp glow, soft falloff, photographic light behavior, documentary-style realism — NOT stylized, NOT plastic, NOT 3D-rendered, NOT over-glowing, NOT artificial. Match the photographic feel of the reference image. Place a SLIM TALL young Japanese woman in the CENTER of the frame, seen FROM BEHIND ONLY (full back view, NO face, NO side profile, NO turning around), walking AWAY from camera into the deep distance of the street. Body proportions: 170cm height build, narrow shoulders, slim waist, slender arms, long legs accounting for ~60% of her height, model-like elongated silhouette. Her figure is SMALL in the frame — occupies only about 1/12 to 1/15 of frame height, a tiny tall narrow silhouette in the middle distance, dwarfed by the long street perspective. The street and streetlamps dominate the composition. Outfit (MUST MATCH costume reference exactly): - Oversized PINK button-up shirt, draped loosely on slim frame - BLUE DENIM JEANS (full-length, straight or slightly loose fit, covering full legs down to the ankles) - White canvas low-top sneakers - Black straight hair with blunt bangs, shoulder-length, viewed from behind She holds a bundle of folded laundry against her chest with both arms. 35mm cinematic, film grain, soft contrast, deep DOF, naturalistic photographic look. Aspect ratio 16:9. Negative： bare legs, bare thighs, exposed legs, no pants, naked legs, leggings, tights, stockings, shorts, mini skirt, skirt, dress, hem reaching mid-thigh, legs showing below shirt, plastic lighting, fake lighting, over-rendered, CG light, 3D render, cartoonish glow, over-saturated lights, neon glow, anime lighting, video game lighting, stylized lighting, fake lens flare, halo around lamps, exaggerated bloom, perfect symmetry of light, artificial atmosphere, medium shot, close-up, mid-distance shot, figure too large, figure dominating the frame, figure occupying more than 1/8 of frame height, character close to camera, shallow depth, short street perspective, cropped street, street ending too soon, vertical composition, 9:16, portrait orientation, front view, side profile, three-quarter view, face visible, turning around, looking back, looking sideways, different street, different buildings, modern city, neon signs, busy commercial street, traffic, motorcycles, bicycles, cars, pedestrians, animals, inventing new scenery, ignoring reference scene, changing lighting tone, changing lamp colors, redesigning the street, short figure, stocky build, plump body, chubby, fat, wide shoulders, broad frame, thick legs, short legs, wide hips, round body, square silhouette, child-like proportions, black shirt, white shirt, blue shirt, jacket, coat, school uniform, high heels, boots, dark shoes, sandals, branded shoes with logos, short hair, long hair past waist, ponytail, bun, dyed hair, brown hair, blonde hair, no laundry bundle, holding bag, holding phone, empty hands, multiple people, second person, group walking, rain, snow, fog effects

正向： The orange tabby cat walks calmly across the visible street area outside the glass door, moving steadily from right to left at a relaxed pace, side profile maintained throughout. The cat does not look toward the camera, does not stop, does not change pace. The girl in the blurred foreground stays completely still, only her shoulders rise and fall almost imperceptibly from breathing. The lighting outside the glass door stays completely constant throughout the entire shot — no light changes, no new light sources appear, no glow, no flare, no illumination shifts. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： new light appearing, light beam, light streak, glowing trail behind cat, light following cat, headlight, car light, street lamp turning on, sudden illumination, lens flare, light burst, glow effect, bloom, light leak, sudden brightness change, exposure shift, cat looking at camera, cat stopping, cat sitting down, cat turning head, cat running, multiple cats, cat changing direction, girl turning around, girl moving, girl getting up, girl reaching for cat, face visible, sharp foreground, people walking past, cars passing, motorcycles, glass door opening, hand on glass, talking, speaking, lip sync, mouthing words, dialogue, parted lips, camera movement, zoom in, zoom out, pan, dolly, camera shake, dramatic music sting

PHOTOREALISTIC REAL PHOTOGRAPHY ONLY, single-frame cinematic still, 16:9 aspect ratio. 35mm film grain, green-amber color grade, Japanese late-night TV drama look (NHK deep-night, Kore-eda / Hamaguchi reference). NO anime, NO illustration, NO 3D render, NO grid, NO multi-panel, NO collage. Wide establishing shot, across-the-street POV, eye-level — a small neighborhood coin laundromat at night, framed from the opposite sidewalk. The storefront: full-height glass window, warm tungsten interior glow spilling onto damp asphalt, soft "コインランドリー" signage above, row of front-loading washers inside, one tumbling. Inside the laundromat, a young Japanese woman sits alone on the bench facing the machines — at this distance she is a small distant silhouette, NOT a portrait, no facial detail. Soft pink button-up shirt over white inner top. Straight short black hair, blunt fringe, white wired in-ear earphones with visible cable. Outside: empty quiet street, deep navy upper sky, single overhead streetlamp, faint reflections on wet asphalt, no other people, no cars. CORE MOOD: solitary, calm, slightly melancholic — observing a stranger's small private moment from across the street. Composition: storefront centered like a lit diorama in surrounding darkness, deep depth, cold-blue exterior vs warm-amber interior contrast. 35mm equiv, f/2.8, gentle bloom and halation on tungsten bulbs. OUTPUT: ONE single photographic frame only. NEGATIVE: storyboard, panels, grid, collage, multi-shot layout, anime, illustration, art book, animation studio, cartoon, 3D render, close-up face, sharp facial features.

Prompt： ▎ Create a single starting frame: a young Japanese woman sits on an old wooden floor of a quiet Japanese apartment living room at dusk. She is seen at a three-quarter back angle, her back mostly toward the camera but turned just enough that her left side jawline, cheek, ear, and a sliver of profile are visible. Her facial features must exactly match reference 1 — same single-eyelid eye shape, same nose, same lip line, same youthful Japanese face. Same short black bob with blunt bangs as reference 1. ▎ ▎ She wears the exact outfit from reference 2: white loose A-line sleeveless slip dress with thin straps. Her knees are drawn up, both arms loosely wrapped around her shins. Posture soft, shoulders relaxed, head tilted very slightly down. A BOSS Café au Lait coffee can sits unopened on the wooden floor just beside her right hip. ▎ ▎ Behind her, an entire paper shoji screen (障子) on the wall glows warm amber from the setting sun outside; the slatted wooden frame casts a grid of warm orange light squares across the wooden floor, across her bare back of arms, and across the visible side of her face and dress. The light on her is gentle, soft, real. The rest of the room is dim and empty. ▎ ▎ Use the storyboard from reference 3 only as a composition reference — exact pose and framing. ▎ ▎ 16:9 cinematic aspect ratio. Naturalistic photographic light, NOT CG, NOT plastic. Soft contrast, slightly desaturated, intimate observational tone. Negative： ▎ face fully visible, full frontal face, face to camera, eye contact, smiling, opened coffee can, can being held, multiple people, person standing, kneeling pose, lying down, formal kimono, modern bright apartment, fluorescent lighting, overhead ceiling light, harsh shadow, dramatic backlight rim, anime style, illustration, painterly, CG plastic skin, oversaturated colors, large face mismatch

Positive： ▎ Wide static shot of an empty Japanese apartment living room at dusk. An entire paper shoji screen (障子) on the far wall glows warm amber from the setting sun outside, the slatted wooden frame casting a grid of warm orange light squares across an old wooden floor. The room is otherwise empty — perhaps a low wooden cabinet or tatami edge at the side, quiet and hushed. ▎ ▎ Subtle micro motion only: a thin cloud drifts unseen outside, causing the grid of orange light on the floor to creep extremely slowly across the floorboards by a tiny amount over the clip. Dust particles drift softly in the warm light beam. Nothing else moves. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Warm amber dusk light from the shoji, deep dim shadow in the rest of the room — strong contrast between glowing screen and dim interior. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, low-key with warm key. Negative： ▎ zoom in, zoom out, pan, tilt, dolly, push, pull, camera shake, parallax shift, perspective warp, shoji frame bending, floor warping, light flickering, sudden brightness change, person entering, person appearing, shadow of person, fast cloud movement, dramatic light change, fluorescent overhead light, ceiling light, bright background, plastic CG lighting, glow halo, lens flare, oversaturated colors, painterly stylization, vintage filter, anime style

Positive： ▎ Extreme close-up from inside a refrigerator looking outward at slightly upward angle. A single Japanese BOSS Café au Lait coffee can stands alone at the back shelf, light blue and silver label clearly readable, bearded mascot icon visible, cold mist condensation beading and clinging to the metallic can surface. A woman's slender fingers enter from the lower right edge of the frame and slowly, gently come to rest against the side of the can in a soft tentative touch — the fingers stop there, holding contact without grasping or lifting. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Cool clean white refrigerator interior light pours from above onto the can, the rest of the fridge interior darker around it. A faint warm dim kitchen ambient glows softly out of focus beyond the open door at the edge of frame. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, hero product framing, shallow depth of field on the can label. Negative： ▎ can falling, can wobbling, can rolling, can being lifted, can being grabbed, hand wrapping around can fully, fingers gripping, fingers squeezing, hand pulling back, hand withdrawing, fast hand motion, hand shaking, label changing, brand changing, generic coffee can, different drink, multiple cans, other items appearing in fridge, fridge interior crowded, fridge door closing, fridge door swinging, overhead ceiling light, bright kitchen lighting, over-bright background, plastic CG lighting, glow halo, lens flare, camera shake, pan, tilt, zoom in, zoom out, dolly, perspective warp, can geometry warping, label distortion, oversaturated colors, painterly stylization, vintage filter, anime style, condensation evaporating, can fully dry

Positive： ▎ Close-up still life of an old wooden window sill in a Japanese apartment at dusk. A single small terracotta brown ceramic pot sits on the sill, holding a long-dead potted plant — the stems and leaves are dry, brown, brittle, curled, completely withered, not a single living green shoot. The wood window frame is rough and worn at the edges. Behind the pot, a paper shoji screen glows warm amber with dusk backlight, the dead leaves appearing as silhouette shapes against the warm glow. ▎ ▎ Subtle micro motion only: a faint slow drift of warm dust particles in the backlit air. A single dry leaf may quiver almost imperceptibly. Nothing else moves. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Strong warm amber backlight from the shoji behind, dim cool shadow on the foreground side of the pot — silhouette-style lighting on the dead plant. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained closing tone, soft contrast, slightly desaturated, melancholic stillness. Negative： ▎ green leaves, fresh plant, healthy plant, blooming flower, new buds, growing shoots, watering, water droplets, person appearing, hand entering frame, picking up pot, pot wobbling, pot falling, leaves falling dramatically, leaves blowing, strong wind, fast dust motion, sudden brightness change, fluorescent light, overhead ceiling light, bright lighting, plastic CG lighting, glow halo, lens flare, camera shake, pan, tilt, zoom in, zoom out, dolly, perspective warp, shoji bending, oversaturated colors, painterly stylization, vintage filter, anime style, animated cartoon look

参考图人物形象，和以下参考提示词结构，不用精简，帮我重新生成一段 10宫格分镜图

Positive： ▎ A quiet late-afternoon Japanese residential alley. Concrete wall and shrubs on the left, a large overhead tree softly framing the top, a two-story wooden apartment with a red iron exterior staircase on the right casting a long diagonal shadow on the ground. At the far end of the alley deep in the background, a small distant female figure in a white loose A-line sleeveless slip dress, white ankle socks, beige sandals, short black bob with blunt bangs, stands as a tiny silhouette. She walks one to two gentle steps forward toward the camera, her scale staying small. ▎ ▎ The camera performs an extremely subtle slow dolly-in along the alley center line, advancing only a small amount across the full clip, maintaining the deep alley perspective. No zoom. No pan. No tilt. ▎ ▎ Overhead tree leaves micro-sway in a light breeze. Everything else remains still — the parked white van and bicycle do not move. ▎ ▎ Cool blue-white evening sky, warm low side sunlight from the right, red iron staircase casting diagonal shadow. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated. Negative： ▎ fast camera push, fast dolly, big push-in, zoom in, zoom out, wide zoom, lens distortion, fisheye, perspective warp, depth distortion, walls bending, alley narrowing, figure becoming large quickly, figure running, pan, tilt, parallax shift, plastic CG lighting, glow halo, lens flare, extra people appearing, parked van moving, bicycle moving, sky color shift, oversaturated colors, painterly stylization, vintage filter, head turn, look at camera, smile, parted lips, sighing

Positive： ▎ A young Japanese woman climbs a red iron exterior staircase of an old two-story wooden apartment, seen from a side-rear angle, waist-up framing. She wears a white loose A-line sleeveless slip dress, white ankle socks, beige sandals, short black bob with blunt bangs. One hand rests lightly on the red iron handrail. She climbs two steps naturally and slowly, the motion fluid and unhurried, shoulders relaxed and soft. ▎ ▎ The camera tracks her from behind-right with an extremely subtle slow rise, following her ascent only slightly. No zoom. No pan sideways. No tilt down. Camera stays close and intimate. ▎ ▎ Red iron handrail bars vertically streak the right side of the frame. The alley background remains softly out of focus behind her. ▎ ▎ Warm low side sunlight from camera right rim-lights her shoulder and dress edge in warm orange; the rest of her body falls into cool soft shadow. Cool blue-white sky overhead. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, shallow depth of field. Negative： ▎ head turn, look at camera, face to camera, smile, parted lips, sighing, exhaling, blinking heavy, eye movement, hair flip, dress flying dramatic, fast climb, running, skipping steps, hand leaving handrail, camera shake, fast pan, fast tilt, zoom in, zoom out, lens distortion, fisheye, perspective warp, handrail bending, staircase warping, plastic CG lighting, glow halo, lens flare, oversaturated colors, painterly stylization, vintage filter, extra people appearing, anime style

Positive： ▎ Close-up of feet on a red iron exterior staircase of an old Japanese apartment, viewed from above and slightly behind at about 30 degrees downward angle. White ankle socks and beige cream cream sandals step onto iron mesh treads. Two steps land in succession naturally, the second step coming to a complete halt. The step currently under her foot is softly lit by warm low side sunlight; the step beyond is in cool dim shadow because the overhead corridor light is broken, creating a clear bright-to-dark transition across the treads. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained tone, soft contrast, slightly desaturated, shallow depth of field on the foreground tread. Negative： ▎ camera shake, pan, tilt, zoom in, zoom out, dolly, perspective warp, staircase warping, tread bending, blurred motion, fast steps, running, skipping a step, hesitating mid-air, additional feet appearing, socks falling off, sandal strap loose, foot leaving frame, other people walking by, plastic CG lighting, glow halo, lens flare, oversaturated colors, vintage filter, anime style, painterly, overhead light flickering

Positive： ▎ Interior of a narrow Japanese apartment kitchen at dusk, dimly lit. A young woman seen from behind walks slowly toward a single-door white refrigerator standing at the depth of the room. She wears a white loose A-line sleeveless slip dress, white ankle socks, beige sandals, short black bob with blunt bangs. Her shoulders are relaxed, posture soft and unhurried. She takes two natural steps forward, keeping a steady pace. ▎ ▎ Wooden floor, a small countertop on the side with a filter water jug and an electric kettle. The room is hushed and quiet. ▎ ▎ The camera performs an extremely subtle slow dolly-in from directly behind her, advancing only a small amount across the full clip, maintaining the deep kitchen perspective. No zoom. No pan. No tilt. ▎ ▎ Warm faint dusk twilight bleeds in softly from the side window behind camera, casting low amber ambient onto the wooden floor. The white refrigerator body in the depth catches a cool reflection, standing out as the brightest surface in the dim room. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, low-key dim interior. Negative： ▎ head turn, look at camera, face to camera, smile, parted lips, sighing, exhaling, hair flip, dress flying dramatic, fast walking, running, hesitating mid-step, fridge door opening, fridge interior light spilling, overhead ceiling light, fluorescent light on, bright kitchen lighting, over-bright interior, plastic CG lighting, glow halo, lens flare, fast camera push, zoom in, zoom out, pan, tilt, parallax shift, perspective warp, kitchen walls bending, floor warping, oversaturated colors, painterly stylization, vintage filter, anime style, extra people appearing

Positive： ▎ Chest-up close-up of a young Japanese woman seen from the refrigerator's interior reverse angle, eye-level framing. She wears a white loose A-line sleeveless slip dress with thin straps, short black bob with blunt bangs, single-eyelid features. Cool clean white refrigerator light from camera direction washes the left half of her face brightly; the right half falls into dim warm shadow from the kitchen behind her. Her gaze stays softly lowered, looking down into the refrigerator, expression completely neutral and still — not sad, not curious, just quiet. ▎ ▎ Her hand slowly and gently extends forward into the lower frame, reaching into the refrigerator interior with a minimal natural motion. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Behind her, the dim narrow kitchen recedes into low warm dusk shadow. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, low-key with cool key light contrast. Negative： ▎ head turn, look at camera, eye to camera, eye movement, eyes shifting, glancing, blinking heavy, smile, parted lips, sighing, exhaling, frowning, raised eyebrow, lip bite, surprised look, mouth opening, hair flip, dress strap slipping, hand pulling back, fast hand movement, hand waving, fridge door swinging wider, fridge door closing, overhead ceiling light, fluorescent kitchen light, bright background, over-bright fridge interior, plastic CG lighting, glow halo, lens flare, camera shake, pan, tilt, zoom in, zoom out, dolly, perspective warp, face geometry warping, oversaturated colors, painterly stylization, vintage filter, anime style, extra people appearing, reflection of camera in fridge

Positive： ▎ A quiet late-afternoon Japanese residential alley. On the left, a low concrete wall with shrubs and a large overhead tree casting soft leaves into frame. On the right, a two-story wooden apartment building with a red iron exterior staircase, casting a long diagonal shadow on the ground. At the far end of the alley, a small distant female figure stands as a tiny silhouette, slowly walking two gentle steps forward toward the camera, her scale staying small. Overhead tree leaves micro-sway in a light breeze; everything else remains perfectly still. Cool blue-white evening sky with warm low side sunlight from the right. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated. Locked tripod camera, fixed framing, no movement. Negative： ▎ camera push, camera pull, zoom in, zoom out, pan, tilt, dolly, parallax shift, plastic lighting, CG lighting, glow halo around streetlight, extra people appearing, parked car driving away, bicycle moving, sky color shift, sun flare, lens flare, heavy motion blur, figure becoming larger, figure facing camera, head turn, look at camera, smile, parted lips, exhaling, sighing, oversaturated colors, painterly stylization, vintage filter

Prompt： ▎ Add the woman from reference 2 into the alley scene from reference 1. She must stand at the far end of the alley deep in the background, appearing very small as a distant silhouette, only roughly 1/8 the frame height. She wears the exact same outfit as reference 2: white loose A-line cotton-linen sleeveless slip dress, white ankle socks, beige cream sandals, short black bob with straight blunt bangs, Japanese single-eyelid features. She faces toward the camera but is too far to see facial detail. ▎ ▎ The alley composition from reference 1 must remain completely unchanged — same concrete wall on the left, same overhead tree, same red iron exterior staircase on the right, same parked white van and bicycle at the alley's end, same ground perspective. Do not move, remove, or add anything else. Do not replace any existing element. ▎ ▎ Late afternoon Japanese residential alley. Cool blue-white sky overhead, warm low side sunlight from the right, red iron staircase casting a long diagonal shadow on the ground. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained tone, soft contrast, slightly desaturated. ▎ ▎ 16:9 cinematic aspect ratio. Negative： ▎ face close-up, large figure, figure in foreground, figure mid-frame, multiple people, no people, extra bicycle, modified buildings, removed tree, changed staircase, oversaturated colors, plastic CG lighting, glow halo, lens flare, painterly stylization, vintage filter, anime style, illustration

LOCKED-OFF STATIC CAMERA, tripod, absolutely no camera movement. NO pan, NO tilt, NO zoom, NO dolly, NO handheld breathing. The frame stays pixel-identical to the starting image throughout. Within the static frame, the motion choreography in two beats: BEAT 1 (seconds 0–1.5): Her right thumb and forefinger gently grip the small red plastic pull-tab at the top point of the triangular wrapper. She slowly pulls the tab partway down the center seam, the clear film starting to part along the perforated line — about a third of the way down. Crisp plastic crinkle and rustle. BEAT 2 (seconds 1.5–5, HOLD): Her hands STOP completely mid-action. The pull-tab is frozen partway down the seam, the wrapper half-opened. Her fingers do not move further. She holds this exact stillness for the rest of the shot — a long suspended pause. Eyes stay softly lowered on the half-opened wrapper, expression calm. The plastic crinkle sound CUTS OFF abruptly at second 1.5. The unwrap action is interrupted, suspended, unfinished. Her mouth stays completely closed and still throughout — NO sighing, NO exhaling, NO parted lips, NO heavy breathing, NO chest heaving. CONTINUOUS BACKGROUND MOTION: behind her in the soft warm bokeh, one washing machine drum tumbles steadily and continuously throughout — clothes rotating, gentle motion blur, never stopping. Drum's low rhythmic hum carries through the entire shot. KEY SOUND DESIGN: - Seconds 0–1.5: crisp plastic wrapper rustling, dominant. - Second 1.5: plastic sound CUTS OFF cleanly, abrupt silence on the wrapper. - Seconds 1.5–5: only the steady low hum and rhythmic tumble of the washer drum, slightly amplified. Subtle, meditative. - No music, no dialogue. Lighting and color: warm tungsten glow steady, green-amber 35mm film grade, subtle grain. End frame: hands frozen exactly as in beat 2 — pull-tab paused partway down, plastic half-parted, washer still tumbling. Action unresolved. Duration: 5 seconds, contemplative pacing, NOT slow-motion. NEGATIVE: sighing, exhaling, breathing heavily, mouth opening, parted lips, talking, biting, eating, fully unwrapping, completing the unwrap, hands continuing motion after pause, hands leaving frame, plastic sound continuing through pause, washer stopped, washer empty, static drum, no background motion, camera movement, pan, tilt, zoom, dolly, handheld, shake, slow motion, anime, cartoon, 3D, CGI, earphones appearing, cable appearing, touching ear.

正向： Her head stays completely still in pure side profile. Only her EYES slowly move: starting at the left edge of her gaze range, her eyes drift smoothly from LEFT to RIGHT across the 5 seconds, as if her gaze is following something invisible passing by outside the frame from left to right. The eye movement is slow, continuous, and gentle — no jerky shifts, no sudden snap. Lips remain closed throughout. Mouth stays shut. Her shoulders rise and fall almost imperceptibly from quiet breathing. A single strand of her hair shifts very slightly from the laundromat's air movement. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： head turning, head movement, face rotating, profile changing, looking at camera, looking up, looking down, eyes moving right to left, eyes snapping, jerky eye movement, eyes darting, multiple eye movements, eyes closed, blinking heavily, talking, speaking, lip sync, mouth movement for speech, mouthing words, parted lips, opening mouth, sighing, exhaling, smiling, frowning, dramatic expression, emotional shift, hand entering frame, body leaning, posture change, camera movement, zoom in, zoom out, pan, dolly, camera shake, exposure shift, light change, lens flare, glow effect, dramatic music sting

正向： The single small RED indicator LED on the washing machine control panel slowly fades out and transitions into a soft GREEN glow during the middle portion of the shot. The color change is gentle and continuous — red dimming, then green emerging in the same exact spot, no flicker, no flash. After the green light settles, it stays lit and stable for the rest of the shot. The drum visible at the edge of frame slows its rotation almost imperceptibly during the color change, then comes to a complete stop. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： LED flickering, multiple flashes, strobe effect, sudden color change, abrupt cut, light burst, lens flare, glow expanding, halo effect, multiple LEDs lighting up, panel fully lighting up, other buttons activating, display changing numbers dramatically, hand entering frame, finger pressing, person appearing, drum speeding up, drum reversing, drum jolting, camera movement, zoom in, zoom out, pan, dolly, camera shake, exposure shift, brightness change, talking, speaking, dialogue, dramatic music sting

Horizontal 16:9 cinematic composition. Wide long shot, fixed locked camera at eye level, looking straight down a quiet residential side street at night. A SLIM, TALL young Japanese woman positioned in the CENTER of the frame, seen from BEHIND, walking AWAY from camera into the depth of the street. Body proportions: 170cm height build, narrow shoulders, slim waist, slender arms, long legs accounting for ~60% of her height, model-like elongated silhouette. She occupies about 1/7 of the frame height vertically — appearing as a tall narrow figure (significantly taller than wide), not a small chubby blob. Her back fully turned to camera. She wears an oversized pink button-up shirt that hangs loosely from her slim frame, the shirt drapes softly without adding bulk, hem reaching mid-thigh, exposing slim bare legs below. White canvas low-top sneakers on slim ankles. Black hair straight with blunt bangs, hair length to shoulders. Holding a bundle of folded laundry against her chest with both arms. The street stretches into deep one-point perspective: rows of warm amber streetlamps line both sides, lights diminishing into the distance, creating a chain of glows. Asphalt road surface, low residential buildings on both sides, a few darkened shop signs, slight haze in the distance. Cool blue night ambient with warm amber pools of light from each streetlamp. Deep DOF, focus from mid-distance to far. 35mm cinematic, film grain, quiet melancholic Kore-eda aesthetic, soft contrast, slightly desaturated. Aspect ratio 16:9. Negative： vertical composition, 9:16, portrait orientation, short figure, stocky build, plump body, chubby, fat, overweight, wide shoulders, broad frame, heavy build, thick legs, short legs, wide hips, round body, square silhouette, child-like proportions, kid body, tiny figure, oversized shirt making body look bulky, shirt fully covering legs, shirt swallowing the body, body lost in clothing, face visible, turning around, looking back, side profile, three-quarter back, multiple people on street, cars passing, motorcycles, bicycles, pedestrians, bright neon signs, modern city skyline, busy commercial street, billboards, traffic lights cycling, vehicle headlights, cat, dog, animals on street, camera tilt, dutch angle, drone shot, overhead view, rain, snow, fog effects

正向： She STROLLS slowly and leisurely AWAY from the camera, deeper into the quiet residential street, taking unhurried casual steps as if she has nowhere to go — the relaxed evening pace of someone walking home after a long day. Each step is gentle and even, her arms lightly cradling the bundle of laundry against her chest, posture relaxed and slightly soft. Her figure becomes smaller in the frame VERY GRADUALLY across the full duration of the shot. At 5 seconds she should still be only moderately further than her starting position — she is in no hurry, just strolling. The streetlamps remain steady throughout, their warm amber glow constant, no flickering. A subtle gentle wind moves a few strands of her hair and the loose hem of her oversized shirt. Her dim silhouette stays integrated into the dim mid-tones of the street, never brightening. Static locked camera, no zoom, no pan, no camera shake, no dolly. OUTPUT: silent video, no audio track needed. Negative： brisk walking, fast pace, hurried steps, rushing, jogging, power walking, quick stride, marching, athletic gait, military walk, determined stride, purposeful posture, figure shrinking too fast, vanishing into distance, reaching the end of the street, disappearing, arms swinging energetically, turning around, looking back, stopping, hesitating, sitting down, dropping the laundry, posture change, running, jogging, jumping, dramatic gesture, laundry falling, clothes flying, items dropped on ground, new people appearing, cars driving by, motorcycles, bicycles, cat appearing, dog appearing, camera dollying in, camera following her, camera tracking, zoom in, zoom out, pan, tilt, camera shake, drone movement, streetlamps flickering, lights turning off, sudden darkness, lightning, sudden brightness change, exposure shift, lens flare, light burst, glow effect, bloom, character brightening, figure lighting up, rain starting, snow falling, fog rolling in, talking, speaking, lip sync, dialogue, dramatic music sting

正向： She walks AWAY from the camera at a normal natural walking pace, deeper into the quiet residential street. Her pace is the everyday casual walking speed of someone heading home — comfortable, steady, neither rushed nor dragging — just normal human walking rhythm. Her arms cradle the bundle of laundry against her chest in a relaxed natural way, shoulders soft, hips moving subtly with each step in the way real walking looks. Each step is regular and even, with normal stride length. Her figure becomes gradually smaller in the frame across the 5 seconds as she moves naturally further into the street. Her hair stays completely still — no wind, no movement in the hair. Her shirt hem stays still — no fabric movement. The streetlamps remain steady throughout, their warm amber glow constant, no flickering, no light changes. Her dim silhouette stays integrated into the dim mid-tones of the street, never brightening. Static locked camera, no zoom, no pan, no camera shake, no dolly. OUTPUT: silent video, no audio track needed. Negative： slow motion, slow-motion effect, dragging feet, sluggish walk, hesitant steps, dawdling, barely moving, stalling, frozen mid-step, time-lapse slow, exaggerated slow walk, walking too slow, fast pace, brisk walking, hurried steps, rushing, jogging, power walking, quick stride, marching, athletic gait, military walk, determined stride, runway walk, model walk, exaggerated hip sway, catwalk, theatrical walking, stiff walking, robotic walking, mannequin-like motion, gliding, floating, sliding without footsteps, hair blowing, hair moving, hair flowing, wind in hair, hair fluttering, strands of hair lifting, hair tossed by wind, shirt fluttering, fabric blowing, hem flapping, clothes moving in wind, wind effect, breeze on character, arms swinging energetically, arms swinging wide, posture change, body twisting, leaning forward, leaning back, turning around, looking back, stopping, hesitating, sitting down, dropping the laundry, running, jogging, jumping, dramatic gesture, laundry falling, clothes flying, items dropped on ground, new people appearing, cars driving by, motorcycles, bicycles, cat appearing, dog appearing, camera dollying in, camera following her, camera tracking, zoom in, zoom out, pan, tilt, camera shake, drone movement, streetlamps flickering, lights turning off, sudden darkness, lightning, sudden brightness change, exposure shift, lens flare, light burst, glow effect, bloom, character brightening, figure lighting up, rain starting, snow falling, fog rolling in, talking, speaking, lip sync, dialogue, dramatic music sting

LOCKED-OFF STATIC CAMERA, tripod-mounted, absolutely no camera movement throughout the entire shot. NO pan, NO tilt, NO zoom in, NO zoom out, NO dolly, NO truck, NO crane, NO handheld breathing, NO parallax, NO push-in, NO pull-out. The frame stays pixel-identical to the starting image from beginning to end. Within the static frame, the only motion is the woman: - She walks at a VERY SLOW, unhurried, almost contemplative pace from frame-RIGHT along the far sidewalk toward the laundromat entrance at center, moving right-to-left across the frame. Her steps are soft and deliberate, about half the speed of normal walking, tote bag swaying gently and slowly with each step. - She reaches the glass door, pauses for a longer beat (about one full second), then pushes the door open softly with her free hand. - The door swings inward slowly, warm tungsten light gradually widens onto the wet asphalt outside as the door opens. - She steps inside calmly, the glass door slowly closes behind her. - Inside the laundromat, she is now seen as a small distant figure through the glass, walking a few unhurried steps toward the wooden bench. Ambience: faint distant city hum, soft slow footsteps on damp asphalt, gentle door hinge sound when she opens it, no music, no dialogue, no traffic, no other pedestrians anywhere in frame. Lighting and color: tungsten storefront glow remains steady and warm, navy night exterior remains cool and quiet, green-amber 35mm film grade holds throughout, subtle film grain, no flicker, no color shift. End frame: glass door fully closed, the small figure of the woman is visible inside the laundromat through the glass walking toward the bench, street outside empty again, identical framing as the opening shot. Duration: 5 seconds, smooth slow real-time pacing, NOT slow-motion (no frame interpolation), just naturally slow walking speed. NEGATIVE: fast walking, hurried pace, brisk steps, normal walking speed, camera pan, camera tilt, camera zoom, dolly, crane, handheld, shake, push-in, pull-out, parallax shift, rack focus, whip pan, fast cuts, multiple shots, scene change, slow motion effect, time lapse, anime, cartoon, 3D, CGI, dramatic music sting, lens flare streak, subject entering from left.

Extreme close-up, frame entirely filled by the inside of a commercial laundromat front-loading washer drum and its glass door, viewed at 30-degree angle from lower-left. The chrome metal trim of the drum opening occupies the bottom-left corner. Foggy water condensation completely covers the entire glass surface, no clear glass area visible anywhere, the fog is dense and uniform. Behind the foggy glass, only colorful clothes (pink, mint green, pale yellow, white) tumbling inside the drum are visible — no room visible, no exterior visible, no space behind the glass other than the drum interior. Overhead fluorescent tube lighting, slight green-yellow cast. Shallow DOF f/2.0, focus on the foggy glass texture and clothes inside. 35mm cinematic, film grain, quiet melancholic Kore-eda aesthetic, soft contrast. Aspect ratio 16:9.

Positive： ▎ A young Japanese woman sits on an old wooden floor of a quiet Japanese apartment living room at dusk, three-quarter back angle, knees drawn up, both arms loosely wrapped around her shins. She wears a white loose A-line sleeveless slip dress, short black bob with blunt bangs. A BOSS Café au Lait coffee can rests unopened on the floor beside her. ▎ ▎ The camera performs an extremely slow lateral truck (horizontal slide) from her back side around toward her side profile, revealing more of her cheek, jawline and ear as the clip progresses. The arc of motion is gentle and short — about 15 degrees of revealing rotation only. No zoom. No tilt. No push. ▎ ▎ She remains completely still — no head turn, no shoulder shift, no breathing motion exaggerated. Only the warm orange grid of shoji light slowly creeps a tiny amount across the floor behind her as a thin cloud drifts outside. ▎ ▎ Warm amber shoji light from behind-right casts a soft grid of orange squares onto her back, dress, and side face. Rest of room dim. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, intimate framing. Negative： ▎ head turn, looking around, looking at camera, eye to camera, eyes opening wide, eye movement, blinking heavy, smile, parted lips, sighing, exhaling, body shifting, leg uncrossing, arm releasing knees, hand reaching for can, picking up can, opening can, drinking, fast camera move, fast truck, full 180 reveal, face fully revealed, dolly in, dolly out, zoom, pan up, pan down, tilt, perspective warp, floor warping, shoji bending, light flickering, fluorescent overhead light, plastic CG lighting, glow halo, lens flare, oversaturated colors, painterly, vintage filter, anime style, extra people

Positive： ▎ Close-up still life of an old wooden window sill in a Japanese apartment at dusk. A single small terracotta brown ceramic pot sits on the sill, holding a long-dead potted plant — the stems and leaves are dry, brown, brittle, curled, completely withered, not a single living green shoot. The wood window frame is rough and worn at the edges. Behind the pot, a paper shoji screen glows warm amber with dusk backlight, the dead leaves appearing as silhouette shapes against the warm glow. ▎ ▎ The camera holds in extremely subtle organic handheld micro-float — a gentle, barely-perceptible breathing motion, as if a person is simply standing still and observing, the slight rise and fall of breath visible only as a tiny drift in frame. No wobble, no shake, no jitter, no swaying, no walking. Framing stays effectively the same throughout the clip. ▎ ▎ Subtle micro motion: a faint slow drift of warm dust particles in the backlit air. A single dry leaf may quiver almost imperceptibly. Nothing else moves. ▎ ▎ Strong warm amber backlight from the shoji behind, dim cool shadow on the foreground side of the pot — silhouette-style lighting on the dead plant. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained closing tone, soft contrast, slightly desaturated, melancholic stillness. Negative： ▎ strong handheld, shaky cam, shaky camera, camera shake, jitter, wobble, swaying camera, walking camera, footstep camera bump, vlog camera, run-and-gun, fast handheld, dramatic camera movement, push in, push out, zoom in, zoom out, pan, tilt, dolly, drone shot, gimbal sway, gimbal float, big movement, green leaves, fresh plant, healthy plant, blooming flower, new buds, growing shoots, watering, water droplets, person appearing, hand entering frame, picking up pot, pot wobbling, pot falling, leaves falling dramatically, leaves blowing, strong wind, fast dust motion, sudden brightness change, fluorescent light, overhead ceiling light, bright lighting, plastic CG lighting, glow halo, lens flare, perspective warp, shoji bending, oversaturated colors, painterly stylization, vintage filter, anime style, animated cartoon look

Positive： ▎ Wide shot of an empty Japanese apartment living room at dusk. An entire paper shoji screen on the far wall glows warm amber from the setting sun outside, the slatted wooden frame casting a grid of warm orange light squares across an old wooden floor. The room is empty — perhaps a low wooden cabinet or tatami edge at the side, quiet and hushed. ▎ ▎ The camera performs an extremely subtle slow dolly-in toward the shoji screen, advancing only a small amount across the full clip, keeping the deep room perspective and the grid of light on the floor steady in framing. No zoom. No pan. No tilt. ▎ ▎ Subtle micro motion: a thin cloud drifts unseen outside, causing the grid of orange light on the floor to creep extremely slowly across the floorboards by a tiny amount over the clip. Dust particles drift softly in the warm light beam. Nothing else moves. ▎ ▎ Warm amber dusk light from the shoji, deep dim shadow in the rest of the room — strong contrast between glowing screen and dim interior. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, low-key with warm key. Negative： ▎ fast push, fast dolly, big push-in, large camera move, zoom in, zoom out, wide zoom, pan, tilt, parallax shift, perspective warp, shoji frame bending, floor warping, light flickering, sudden brightness change, person entering, person appearing, shadow of person, fast cloud movement, dramatic light change, fluorescent overhead light, ceiling light, bright background, plastic CG lighting, glow halo, lens flare, oversaturated colors, painterly stylization, vintage filter, anime style, camera shake, jitter, handheld wobble

Positive： ▎ Extreme close-up from inside a refrigerator looking outward at slightly upward angle. A single Japanese BOSS Café au Lait coffee can stands alone at the back shelf, light blue and silver label clearly readable, bearded mascot icon visible, cold mist condensation beading and clinging to the metallic can surface. ▎ ▎ A woman's slender hand enters from the lower right edge of the frame at natural everyday speed — the way someone reaches into their fridge to grab a drink. Her fingers wrap firmly around the cold can in a single fluid motion, lift it cleanly upward, and carry it up out of the frame toward camera. The shelf is left empty where the can stood, only a faint wet condensation ring remaining on the metal rack. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Cool clean white refrigerator interior light pours from above, the rest of the fridge interior darker around it. A faint warm dim kitchen ambient glows softly out of focus beyond the open door at the edge of frame. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, hero product framing, shallow depth of field on the can label. Negative： ▎ slow motion, slow-mo, sluggish hand, dragging hand, hovering hand, floating hand, lethargic reach, ghost-like motion, dreamy slow motion, hesitating mid-air, hand pausing mid-grip, time-lapse, exaggerated slow, hand only touching without grabbing, fingers brushing without contact, hand pulling back empty, hand withdrawing without can, can staying on shelf, can left behind, can falling, can wobbling, can rolling, can dropping, can slipping out of grip, label changing, brand changing, generic coffee can, different drink, multiple cans, other items appearing in fridge, fridge interior crowded, fridge doing, overhead ceiling light, bright kitchen

Positive： ▎ Close-up of feet ascending upward on a red iron exterior staircase of an old Japanese apartment, viewed from above and slightly behind at about 30 degrees downward angle. White ankle socks and beige cream sandals step UP onto the next iron mesh tread, then UP onto the one above it, climbing two steps in a natural everyday pace. The second upward step comes to a calm halt on the higher tread. ▎ ▎ The step currently underfoot is softly lit by warm low side sunlight; the higher step ahead is in cool dim shadow because the corridor light is broken, creating a clear bright-to-dark transition going upward across the treads. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained tone, soft contrast, slightly desaturated, shallow depth of field on the foreground tread. Negative： ▎ descending, going down, stepping down, walking backwards, reverse direction, retreat, going back, foot moving down, walking downstairs, downward steps, reversed playback, camera shake, pan, tilt, zoom in, zoom out, dolly, perspective warp, staircase warping, tread bending, blurred motion, fast steps, running, skipping a step, hesitating mid-air, additional feet appearing, socks falling off, sandal strap loose, foot leaving frame, other people walking by, plastic CG lighting, glow halo, lens flare, oversaturated colors, vintage filter, anime style, painterly, overhead light flickering

Positive： ▎ Interior of a narrow Japanese apartment kitchen at dusk, dimly lit. A young woman seen from behind walks toward a single-door white refrigerator standing at the depth of the room. She wears a white loose A-line sleeveless slip dress, white ankle socks, beige sandals, short black bob with blunt bangs. She walks at a natural everyday pace, two purposeful steady steps forward, the way someone walks in their own home — relaxed but with normal momentum. Posture upright, arms naturally swinging slightly with her gait. ▎ ▎ Wooden floor, a small countertop on the side with a filter water jug and an electric kettle. The room is hushed and quiet. ▎ ▎ The camera performs an extremely subtle slow dolly-in from directly behind her, advancing only a small amount across the full clip — the camera is slow, but the woman walks at normal speed. No zoom. No pan. No tilt. ▎ ▎ Warm faint dusk twilight bleeds in softly from the side window behind camera, casting low amber ambient onto the wooden floor. The white refrigerator body in the depth catches a cool reflection, standing out as the brightest surface in the dim room. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, low-key dim interior. Negative： ▎ slow motion, slow-mo, slow walking, sluggish walking, dragging feet, zombie walking, floating gait, hovering, lethargic, exaggerated slow, dreamy slow motion, frozen mid-step, time-lapse, hesitation, ghost-like movement, drifting forward, gliding, sliding instead of walking, head turn, look at camera, face to camera, smile, parted lips, sighing, exhaling, hair flip, dress flying dramatic, running, fridge door opening, fridge interior light spilling, overhead ceiling light, fluorescent light on, bright kitchen lighting, over-bright interior, plastic CG lighting, glow halo, lens flare, fast camera push, zoom in, zoom out, pan, tilt, parallax shift, perspective warp, kitchen walls bending, floor warping, oversaturated colors, painterly stylization, vintage filter, anime style, extra people appearing

Positive： ▎ Chest-up close-up of a young Japanese woman seen from the refrigerator's interior reverse angle, eye-level framing. She wears a white loose A-line sleeveless slip dress with thin straps, short black bob with blunt bangs, single-eyelid features. Cool clean white refrigerator light from camera direction washes the left half of her face brightly; the right half falls into dim warm shadow from the kitchen behind her. Her gaze stays softly lowered, looking down into the refrigerator, expression completely neutral and still — not sad, not curious, just quiet. ▎ ▎ Her hand reaches naturally forward into the lower frame at normal everyday speed, the way someone reaches into their own fridge to grab a drink — calm but not slow. The hand enters fridge interior with a single fluid motion. ▎ ▎ Camera fully locked. No movement. ▎ ▎ Behind her, the dim narrow kitchen recedes into low warm dusk shadow. Naturalistic photographic light, NOT plastic, NOT CG. Cinematic restrained observational tone, soft contrast, slightly desaturated, low-key with cool key light contrast. Negative： ▎ slow motion hand, slow-mo, sluggish hand, dragging hand, hovering hand, floating hand, lethargic reach, ghost-like motion, dreamy slow motion, hesitating mid-air, hand pausing mid-reach, frozen hand, time-lapse, exaggerated slow, head turn, look at camera, eye to camera, eye movement, eyes shifting, glancing, blinking heavy, smile, parted lips, sighing, exhaling, frowning, raised eyebrow, lip bite, surprised look, mouth opening, hair flip, dress strap slipping, hand pulling back, hand waving, fridge door swinging wider, fridge door closing, overhead ceiling light, fluorescent kitchen light, bright background, over-bright fridge interior, plastic CG lighting, glow halo, lens flare, camera shake, pan, tilt, zoom in, zoom out, dolly, perspective warp, face geometry warping, oversaturated colors, painterly stylization, vintage filter, anime style, extra people appearing, reflection of camera in fridge

【生成内容】 # S6 · 夏日祭烟火 · 夜 — 分镜脚本 v1（2026-06-06） ## 场景定位 全片唯一的"破"。她本来不想来，但站到一半被烟花砸住，"哇"轻声出口，嘴角松。\ 触发回扣 S1 的咸米——不硬切，靠手指碰嘴角的微动作暗示。 ## 全场基调 - 比例 16:9 cinematic - 主光：夜空烟花（红/金/蓝/紫，按节拍变色），人群灯笼暖色补光 - 镜头语言：人群中以中景/特写为主，烟花镜走仰拍空镜 - 造型：按已定稿（不写进 prompt，交给 ref 图锁） - silent video（音轨后期） --- ## 镜 1 · 远景挤进 - **景别**：远景（祭典街道全景） - **时长**：4–5s - **画面**：祭典街道，两侧浴衣人群密集，红灯笼一串串挂头顶，远端摊位灯光暖橘色。她从画面右下进画，背影、低肩、人群里慢慢挤 - **镜头运动**：极慢推近（push in 5%），不晃 - **光线**：灯笼暖橘 + 摊位高光散点 - **备注**：她在画面里只占 1/8，气氛压过人物 ## 镜 2 · 中景稳住 - **景别**：中景（腰以上） - **时长**：3–4s - **画面**：人群从左右挤过她，她肩膀被蹭一下，脚下顿一拍，稳住，继续往前 - **镜头运动**：手持微晃（跟拍 1 步幅） - **光线**：暖橘灯笼从上方泼下，脸 70% 在阴影里 - **备注**：信息点是"她还在"，别让人物太亮 ## 镜 3 · 仰头预备 - **景别**：仰角中景（从下往上看她） - **时长**：3s - **画面**：她停步，慢慢抬下巴，视线越过灯笼往天上去。天还没炸，黑底 - **镜头运动**：固定不动 - **光线**：灯笼侧光勾下颌线，背景纯黑夜空 - **备注**：脖颈到下颌的线条干净；眼神不写动（\[\[feedback_kling_no_eye_movement\]\]） ## 镜 4 · 首炸空镜 - **景别**：纯空镜（仰拍夜空） - **时长**：3–4s - **画面**：黑色夜空中央，一发烟花从下方升起，到顶端炸开，金红色花瓣展开扩散，余烬下落 - **镜头运动**：固定（极轻微随炸开的震感推近 2%） - **光线**：纯黑底 + 烟花自发光 - **备注**：核心炸点 1。这一镜不能有任何人物/前景遮挡 ## 镜 5 · 仰头特写"哇" - **景别**：脸特写（鼻梁以上，下颌切掉） - **时长**：4s - **画面**：烟花光打满整张脸，光由暗→金红→蓝紫渐变铺过来，她仰着，嘴轻轻张开一点点 - **镜头运动**：极慢推近（push in 3%） - **光线**：烟花顶光主导，脸基底冷调灰，烟花色按节奏铺 - **备注**：核心炸点 2。颜色变化要明显（金红→蓝紫），i2v prompt 里把这个色序写死 ## 镜 6 · 嘴角松 - **景别**：脸近景（眼睛到下颌） - **时长**：3s - **画面**：烟花光仍在脸上变色，她嘴角不自觉松开一点点，不是笑，只是不绷了 - **镜头运动**：固定 - **光线**：烟花光为蓝紫尾色 → 转回暖灯笼底光 - **备注**：留白镜，演技镜，i2v 容易跑偏，可能要多跑两版选 ## 镜 7 · 手指碰嘴角 - **景别**：手部+下颌的近景（脸只入下半张） - **时长**：3s - **画面**：她的手指从画面下方升起来，指尖很轻地碰了一下嘴角，停一拍，落下 - **镜头运动**：固定 - **光线**：烟花余光，偏暖 - **备注**：这是咸米的暗示触发（\[\[feedback_kling_i2v_advanced_pitfalls\]\] 时序别分步——一气呵成"升起→碰→落下"，不要写成三段） ## 镜 8 · 失焦剪影 - **景别**：前景剪影 + 背景虚焦烟花 - **时长**：4s - **画面**：她在前景失焦成深色剪影（占画面右 1/3），背景中央第二轮烟花虚焦炸开，光斑大颗（f1.2 大光圈感） - **镜头运动**：固定 - **光线**：背景烟花为唯一光源 - **备注**：呼吸镜，让"哇"那一下有回落空间 ## 镜 9 · 余烟散尽 - **景别**：中远景（侧面 3/4 身） - **时长**：4–5s - **画面**：烟花散尽，夜空只剩一缕白烟。她仍仰着，肩膀松下来，人群嘈杂从画面外回来（视觉上：左右画面边缘有人影晃过） - **镜头运动**：极慢拉远（pull out 5%） - **光线**：烟花已无，回到灯笼暖橘 + 夜空暗底 - **备注**：收束镜。她不动。下一段（S7）的"扣手机"主动姿态从这里起势 --- ## 总时长 约 31–36 秒（按 i2v 5s/镜，正片可在剪辑里压缩） ## 起始帧出图清单（\*\*\*\*\*\*\*\*\*\* 三参考图） 按 \[\[feedback\_\*\*\*\*\*\*\*\*\*\*\_3ref_order\]\] 喂图顺序： 镜第一张（锁脸/锁构图）第二张（造型/场景辅）第三张（故事板大图）1场景图：灯笼街远景造型定稿图故事板 S6 大图2脸基准造型定稿图故事板 S6 大图3脸基准（仰角）造型定稿图故事板 S6 大图4烟花空镜参考——5脸基准（特写）造型定稿图故事板 S6 大图6脸基准（近景）—故事板 S6 大图7手部+下颌参考造型定稿图故事板 S6 大图8剪影参考造型定稿图—9脸基准（侧 3/4）造型定稿图故事板 S6 大图 镜 4 烟花纯空镜不锁脸；镜 8 是剪影也不需要脸，但造型轮廓要留。 ## 已避坑 - ❌ 不硬切 S1 饭团闪回 → ✅ 镜 7 手指碰嘴角暗示 - ❌ 不写眼神动 → 凝视靠起始帧锁 - ❌ 不写音效 → silent video，独白/烟花声/人群声后期 - ❌ 不分步写时序 → 镜 7 一气呵成

正向： The washer drum rotates steadily inside. Colorful clothes tumble and softly slap against the drum wall as they fall through the rotation, visible through the dense fog on the glass. Water droplets on the foggy glass surface slowly trickle downward. The fog stays dense and uniform throughout, no clear area opens up. Static locked camera, no zoom, no pan, no camera shake. OUTPUT: silent video, no audio track needed. Negative： fog clearing, defogging, glass becoming transparent, clear area appearing on glass, human reflection, face reflection, person reflected on glass, silhouette on glass, figure visible through glass, ghostly figure, face emerging from fog, eyes in fog, recognizable face, facial features, room visible behind glass, exterior visible, street visible, camera angle changing, view shifting to frontal, person walking past, hand reaching in, clothes flying out, drum stopping, drum opening, camera movement, zoom in, zoom out, pan, dolly, camera shake, talking, speaking, dialogue, dramatic music sting