《喜剧之王》

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

旧社区图书馆内的女主半身画面。她站在画面中右侧，身体自然朝前，脸微转向画面左侧的画外男主，眼神亲近而轻松，嘴角有很淡的熟悉笑意。双手自然放在身前，未手持书，也未递书。米白开衫、浅色内搭和细项链清晰可见。左侧大窗透入柔和日光，窗边木桌放着书与绿植，右后方是深色旧书架和照片墙。窗外树叶呈柔软散景，室内木材保留温暖质感，包位于画外，不提前表现哽咽或落泪。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

旧社区图书馆内，越过男主肩部看女主的近景。男主模糊的后脑、侧颈和灰白衬衣肩部只占画面左侧前景，女主清晰位于中右侧，微微抬眼看向他，露出温柔自然的浅笑。两人保持熟人谈话的近距离，女主长发、米白开衫与细项链延续确认造型。窗光从左侧柔和照亮她的脸与发丝，后方书架、旧照片、绿植和木门自然虚化。双手与包位于画外，书留在馆内；只捕捉听见应答的日常瞬间，不新增递书或拥抱。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

旧图书馆门外双人全身同框。男主在画面左侧木门旁自然站定，身体朝右，温和地看着她，左手略抬，左腕手表可见。女主位于画面右侧，身体与脚步仍朝右，头转向左侧男主，带熟悉而柔软的浅笑，左手轻扶左肩布包带，包垂在自身左胯。两人相隔几步，保持自然的六厘米身高差。门内有旧书架，门外树荫、石板路与少量落叶延续初秋日光。保留完整脚部与鞋子，人物之间有呼吸空间，不让她拿书或提前落泪。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

图书馆门外越过男主肩部拍女主的中近景。男主模糊的灰白衬衣肩颈占画面左侧前景，女主清晰落在中右侧，面向画面左侧男主，头轻轻偏一点，眼神在温柔玩笑中短暂认真，嘴角微扬，嘴唇自然微启。她用自身左手轻扶左肩布包带，米白针织、浅色内搭与细项链保持一致。后方木门、旧书架及街边橄榄绿树叶柔和虚化，日光描出发丝边缘。只呈现反问的一瞬，神情亲近，没有委屈哭脸或滑落的眼泪。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

沿原有关系轴拍摄门外双人全身远景。男主留在画面左侧图书馆木门旁，双手自然放下，视线追随离开的她。女主位于画面右侧，以三分之四背面继续向右行走，头与视线朝前，此刻没有回头；长裙随着一步轻轻摆动，棕色平底鞋接触石板路，布包仍在自身左肩左胯。两人之间留出明显空隙，以空荡的人行道呈现尚未说出口的回答。旧书架、树干、绿植与零散落叶保持空间连续，克制宁静，没有追赶、哭泣或拿书。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

承接上一镜的门外双人全身远景，构图保持男左女右。男主仍在左侧旧木门旁，站定朝右看她，表情带终于鼓起勇气的急切，身体没有追到她面前。女主已在右侧几步之外，身体和双脚仍朝右，步幅放慢，头转向左侧远处男主，眼神亲近，嘴角只有一点浅笑。布包延续自身左肩左胯，裙摆与鞋子保持一致。两人之间的石板路空隙明确可见，旧书架、树荫和落叶延续原场景，不缩短距离，不让两人突然贴近，也不提前落泪。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

稳稳拍摄男主肩部以上的近景，人物位于画面中左侧，右侧保留看向女主的空间。他仍站在旧图书馆门外，头微转向画面右侧，目光认真地望着几步外的画外女主。试探的笑已经收起，眉眼柔和却坚定，嘴唇自然微启，捕捉终于说出承诺的一瞬，不夸张张口喊叫。蓬松短黑发、灰白衬衣领口与左胸口袋延续确认造型。旧木门和书架在后方虚化，右侧是晴日树叶散景；柔和光线保留皮肤细节，女主不进入近距离前景。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

街道上的女主肩部以上近景，人物位于中右侧，朝画面左侧远处的画外男主看去，左侧留出干净的视线空间。她被承诺打动，眼神变软、稍有湿润，却没有泪水滑落；嘴角努力恢复小小的温柔笑意，笑里藏着心疼与不舍，嘴唇自然微启。保持半扎长发、脸侧碎发、米白开衫、浅色内搭与细项链。后方为沿街树干、橄榄绿叶与旧建筑柔和散景，表现她已经走远。画面只出现女主，不增加男主贴近的肩头前景，不返回馆内。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

晴日旧图书馆外的路边全身画面，灰蓝四门出租车已经停稳，车头向画面左，车尾向右，浅米色车顶灯没有文字。女主在靠人行道的右后门旁，双手自然放下，视线落向门的位置，捕捉走近准备开门的一瞬，右后门仍关闭，不再抬手招车。布包仍在自身左肩左胯，可被身体局部遮住。隔玻璃可见同一位约四十五岁中国男司机，短黑发、灰蓝衬衣，坐前排远侧左驾位朝车头看。右后座空着，车漆映出树叶与日光。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

延续同一灰蓝出租车、路边位置与车头左车尾右的方向，拍女主在右后门旁准备入座的全身画面。靠人行道的右后门已打开，灰黑布座椅和空着的右后座清楚可见。女主轻扶门框，身体微微朝车内倾，视线落向座位，表情安静克制；布包仍在自身左肩左胯，尚未放到座椅上。前排远侧左驾位局部可见同一短黑发、灰蓝衬衣男司机，他始终朝车头看。双脚、手指和车门结构自然准确，身体不穿过车门，没有泪水。

每次上传对应镜号的 `02_分镜图` 同名图片，锁定景别、构图、朝向和空间关系；上传 `03_人物参考` 中的确认照，锁定人物身份。保持同一张脸、发型、服装和真实皮肤质感。女主172厘米，半扎深色长发、脸侧碎发、米白针织开衫、浅色内搭、细项链、深灰及小腿的长裙、棕色平底皮鞋；出门后燕麦色布包位于人物自身的左肩与左胯，上车后放到膝上。男主178厘米，蓬松短黑发、灰白卷袖衬衣、深炭灰长裤、暖灰平底鞋、左腕棕皮带手表。同框保持自然的6厘米身高差。书留在馆内。

每张为一幅独立横屏画面，比例接近16:9。真实青春电影摄影，柔和初秋自然光，暖肤色、橄榄绿环境和冷青阴影，材质细节自然，高光柔和；不加粗颗粒、字幕、图内台词、标签、箭头、水印、边框、拼图或黑边。下列提示词中的“左、右”指画面方向，包、手表、安全带的左右则明确指人物自身。

同一出租车内右后座的女主中近景，侧窗位于画面左侧，相机从后排中央方向看她的三分之四侧脸。车门已经关好，灰黑布座椅、浅灰顶棚保持一致；安全带从人物自身右肩斜至左胯，燕麦色布包放在膝上，双手自然压着包。她望向窗外，眼睛湿润、唇角收住，第一滴泪正沿脸颊滑落，捕捉段末终于卸下笑容的关键帧，神情克制而非嚎哭。窗外晴日树影略有行驶模糊，司机在相机后方画外，不增添雨夜或其他乘客。

subject_definitions:
<Subject 1> is the approved 24-year-old, 172 cm East Asian woman: face and loose half-tied dark brown-black hair from <Picture 3>, full wardrobe from <Picture 5>, and indoor presence from <Picture 1> and <Picture 2>. Preserve her cream knitted cardigan, pale inner top, fine necklace, charcoal calf-length skirt and brown flat shoes. Her oatmeal tote is initially off-screen and later settles on her anatomical left shoulder and hip. Her voice is a warm, clear young-adult Mandarin voice in a natural medium-high register, soft resonance and relaxed articulation; gentle rather than childish or whispered. She is (S1) in this clip.
<Subject 2> is the approved 27-year-old, 178 cm East Asian man: face and natural short black hair from <Picture 4>, full wardrobe from <Picture 6>, and foreground position from <Picture 2>. Preserve his gray-white cotton shirt with sleeves rolled to the forearms, charcoal trousers, gray flat canvas shoes and watch on his anatomical left wrist. His voice is a warm young-adult Mandarin voice in a natural medium-low register, light chest resonance and clear unforced articulation; conversational rather than theatrical. He is (S2) in this clip.
<Picture 1> is the indoor half-length composition reference for [Shot 1], with the standing woman looking left through soft library daylight.
<Picture 2> is the over-shoulder composition reference for [Shot 2], with the man's blurred shoulder at the left foreground and the standing woman in focus.

summary:
[reference generation] Generate one 10.0-second, two-shot library conversation. <Subject 1> and <Subject 2> retain their approved identities and familiar, gentle manner. <Picture 1> and <Picture 2> guide their respective compositions; the other pictures supply identity and wardrobe only. These photographs are references, not multiple compulsory first frames. Cut once at 00:04.000, then finish with her collecting her bag and leaving through the doorway.

retention_analysis:
<Subject 1> (appears in [Shot 1], [Shot 2]): fully_preserved - approved face, adult proportions, hair, clothing and jewelry; the tote is collected only after the goodbye.
<Subject 2> (appears as foreground and doorway presence in [Shot 2]): fully_preserved - approved identity, clothing, left-wrist watch and small height difference.
<Picture 1> ([Shot 1] composition): fully_preserved - standing half-length framing, leftward eyeline and library light; animate a natural conversation within this arrangement.
<Picture 2> ([Shot 2] composition): fully_preserved - foreground shoulder and woman's close framing; allow a small camera adjustment as she leaves.
<Picture 3> and <Picture 5>: fully_preserved - female identity and wardrobe references, not additional shots.
<Picture 4> and <Picture 6>: fully_preserved - male identity and wardrobe references, not additional shots.

detailed_description:
Create a continuous 10.0-second live-action scene in soft early-autumn daylight, with warm natural skin, cream knitwear, olive foliage beyond the windows and subtly cool shadows. Use one full landscape frame, gentle highlights and readable expressions; preserve both faces and bodies without morphing, subtitles, captions, collage panels or added grain.

[Shot 1] Use <Picture 1> for an eye-level half-length view inside the old library. <Subject 1> is already standing, long half-tied hair resting over her cardigan, hands relaxed near her waist. She looks left toward <Subject 2>, outside this framing. Warm window light brushes her cheek; wooden shelving recedes quietly behind her. The camera stays still. Her hands are empty, and the tote remains on an off-screen surface beside her. At approximately 00:00.500, <Subject 1> (S1) speaks in her established warm, clear Mandarin voice, with ordinary familiarity and a light smile: <d>[Chinese] 书先帮我留着吧。</d> Let the complete final syllable finish naturally by about 00:02.800. Her lips then settle into a small smile as she listens. Keep her eyes clear and dry, with no trembling breath or premature sadness. Books remain in the library; there is no handover, pointing demonstration or extra spoken explanation.

[Shot 2] At 00:04.000, the camera cuts to the over-shoulder arrangement guided by <Picture 2>. <Subject 2>'s gray-white shoulder is soft at the extreme left; <Subject 1> remains standing and sharply readable. At approximately 00:04.150, <Subject 2> (S2) answers from that nearby foreground position in his established warm medium-low voice: <d>[Chinese] 好，给你留着。</d> This is live conversation within the room. The woman listens with her mouth closed. After his complete reply, <Subject 1> (S1) gives a tiny familiar smile and says: <d>[Chinese] 那我先走啦。</d> Aim to complete her line by about 00:06.800, without stretching the pauses.

Only after speaking, she lowers her gaze and smoothly lifts the off-screen tote onto her anatomical left shoulder. Her body stays upright throughout; this is not a stand-up action. As she collects it, the man steps aside and just outside the doorway to clear her path. A small, slow camera ease-back accommodates her turn and first departing steps without adding a third cut. End at 00:10.000 as she passes through the doorway toward daylight, the tote settled at her left hip and both hands free of books. Keep the movement continuous and modest, with a soft cloth rustle and unhurried footfalls. Each line occurs once; preserve its wording and punctuation exactly, with no additional speech.

overall_soundscape:
Soft indoor room tone, distant subdued street noise through the entrance, light footsteps and a gentle canvas-and-knitwear rustle. Keep these sounds low and continuous beneath the conversation, with no background chatter, singing or audible radio.

non_diegetic_music:
N/A

subject_definitions:
<Subject 1> is the approved 24-year-old, 172 cm East Asian woman: face and half-tied dark brown-black hair from <Picture 4>, full wardrobe from <Picture 6>, and street placement from <Picture 1>, <Picture 2> and <Picture 3>. Preserve her cream knitted cardigan, pale inner top, fine necklace, charcoal calf-length skirt, brown flat shoes and oatmeal tote on her anatomical left shoulder and hip. Her voice is a warm, clear young-adult Mandarin voice in a natural medium-high register, soft resonance and relaxed articulation; gentle rather than childish or whispered. She is (S2) in this clip.
<Subject 2> is the approved 27-year-old, 178 cm East Asian man: face and short natural black hair from <Picture 5>, full wardrobe from <Picture 7>, and street placement from <Picture 1> and <Picture 3>. Preserve his gray-white rolled-sleeve cotton shirt, charcoal trousers, gray flat canvas shoes and watch on his anatomical left wrist. His voice is a warm young-adult Mandarin voice in a natural medium-low register, light chest resonance and clear unforced articulation; conversational rather than theatrical. He is (S1) in this clip.
<Picture 1> is the two-person street composition reference for [Shot 1]: man left by the library, woman right along the pavement.
<Picture 2> is the woman's medium-close composition reference for [Shot 2], with a blurred male shoulder at the left edge.
<Picture 3> is the wider departure composition reference for [Shot 3], with increasing distance between the stationary man and the rightward-walking woman.

summary:
[reference generation] Generate one 14.0-second street conversation with cuts at 00:05.000 and 00:11.000. <Picture 1>, <Picture 2> and <Picture 3> guide the corresponding shot compositions and spatial continuity; the remaining pictures retain <Subject 1> and <Subject 2>'s identities and clothing. No photograph is an automatic compulsory first frame, and identity portraits do not become extra shots. Begin with a gentle call, continue through her teasing response, then let her walk away.

retention_analysis:
<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - approved face, hair, clothing, adult proportions and anatomical-left tote position.
<Subject 2> (appears in [Shot 1], as shoulder in [Shot 2], and in [Shot 3]): fully_preserved - approved face, wardrobe, left-wrist watch and doorway position.
<Picture 1> ([Shot 1] composition): fully_preserved - shared frame, man left, woman right, readable space between them.
<Picture 2> ([Shot 2] composition): fully_preserved - woman's leftward eyeline and out-of-focus male foreground; his speech is an on-location off-screen line.
<Picture 3> ([Shot 3] composition): fully_preserved - widening separation and woman's rightward departure.
<Picture 4> and <Picture 6>: fully_preserved - female face and wardrobe sources, not new shot locations.
<Picture 5> and <Picture 7>: fully_preserved - male face and wardrobe sources, not new shot locations.

detailed_description:
Create a 14.0-second live-action scene on the same tree-lined pavement outside the library, maintaining soft early-autumn sun, warm skin, olive foliage and cool shaded areas. Use a single landscape image at every instant, with natural skin and fabric detail, stable faces and proportions, and no subtitles, collage, photo dissolves or added grain.

[Shot 1] Frame an eye-level two-shot guided by <Picture 1>. <Subject 2> stands at screen left beside the wooden doorway, his rolled sleeves and left-wrist watch consistent with the references. <Subject 1> has just stepped outside and occupies screen right, her body directed right along the pavement, tote already on her left shoulder. The height difference remains small. He looks right at her; she starts turning her head left. Around 00:00.400, <Subject 2> (S1) calls in his established warm medium-low Mandarin voice, casually and without shouting: <d>[Chinese] 喂，去哪里啊</d> She slows, allows a small smile and replies as <Subject 1> (S2), warm and gentle: <d>[Chinese] 上班哦</d> Complete both lines comfortably before the cut. Keep their expressions readable and attention on each other, not on the camera. Light footfalls soften as she pauses.

[Shot 2] At 00:05.000, the camera cuts to the medium-close composition of <Picture 2>. <Subject 1> stays on the right with looking-room to the left. Only a blurred sliver of <Subject 2>'s shoulder enters the extreme left foreground; his mouth is outside the frame. The camera holds steady on her face. Around 00:05.250, <Subject 2> (S1), off-screen at that nearby left position, speaks his live question: <d>[Chinese] 不上班行不行啊</d> Keep the sound spatially connected to the man beside her, never narration or voiceover. She listens with closed lips, slightly inclines her head and lets a curious smile reach her eyes. After a short natural turn-taking gap, <Subject 1> (S2) answers: <d>[Chinese] 不上班你养我啊</d> Her delivery is gently testing rather than demanding, with no crying or exaggerated flirtation. Finish the entire reply by about 00:10.200, leaving a brief searching glance before she moves.

[Shot 3] At 00:11.000, the camera cuts wider, using <Picture 3> for the departure relationship. She turns naturally and walks toward screen right while he remains near the entrance at screen left. Keep the same camera side of their axis. Her tote swings slightly at her left hip and never changes shoulders; she carries no books. The man follows her with his eyes, mouth closed, gathering resolve without another utterance. Hold the increasing empty space between them until 00:14.000. Keep all four lines exactly as written, once each, and let this silent departure continue naturally into the next scene.

overall_soundscape:
A quiet daytime street bed continues across all three shots: soft leaf movement, distant unobtrusive traffic, footsteps and slight fabric movement. The foreground remains acoustically intimate, with no intelligible background speech, horns interrupting dialogue or source music.

non_diegetic_music:
N/A

subject_definitions:
<Subject 1> is the approved 24-year-old, 172 cm East Asian woman: face and half-tied dark brown-black hair from <Picture 3>, full wardrobe from <Picture 5>, and distant street position from <Picture 1>. Preserve her cream knitted cardigan, pale inner top, fine necklace, charcoal calf-length skirt, brown flat shoes and oatmeal tote on her anatomical left shoulder and hip. Her voice is a warm, clear young-adult Mandarin voice in a natural medium-high register, soft resonance and relaxed articulation; gentle rather than childish or whispered. She is (S2) in this clip.
<Subject 2> is the approved 27-year-old, 178 cm East Asian man: face and short natural black hair from <Picture 4>, full wardrobe from <Picture 6>, and serious close-up appearance from <Picture 2>. Preserve his gray-white rolled-sleeve cotton shirt, charcoal trousers, gray flat canvas shoes and watch on his anatomical left wrist. His voice is a warm young-adult Mandarin voice in a natural medium-low register, light chest resonance and clear unforced articulation; conversational rather than theatrical. He is (S1) in this clip.
<Picture 1> is the separated two-person composition reference for [Shot 1], retaining the distance established by the woman's previous rightward departure.
<Picture 2> is the male close-up composition reference for [Shot 2], with his gaze toward the woman off-screen right.

summary:
[reference generation] Generate one 10.0-second scene with a single cut at 00:04.000. <Picture 1> guides the separated street relationship and <Picture 2> guides the serious male close-up; neither is an obligatory first-frame restart. The remaining pictures preserve <Subject 1> and <Subject 2>'s faces and clothing only. The man calls her back, hears her response, then makes one sincere promise, completing it by approximately 00:06.800 and retaining a quiet reaction tail through the end.

retention_analysis:
<Subject 1> (appears in [Shot 1], remains off-screen in [Shot 2]): fully_preserved - same approved adult woman, outfit, left-side tote and already-established distant position.
<Subject 2> (appears in [Shot 1], [Shot 2]): fully_preserved - same approved adult man, identity, outfit and left-wrist watch; the serious expression develops naturally.
<Picture 1> ([Shot 1] composition): fully_preserved - man screen left, woman farther screen right, open pavement between them and matching eyelines.
<Picture 2> ([Shot 2] composition): fully_preserved - intimate male close-up, rightward gaze and coherent daylight; he does not move closer to her at the cut.
<Picture 3> and <Picture 5>: fully_preserved - female face and full-costume sources, not extra shots.
<Picture 4> and <Picture 6>: fully_preserved - male face and full-costume sources, not extra shots.

detailed_description:
Create a 10.0-second live-action continuation of the same early-autumn street encounter, retaining soft natural daylight, warm skin, olive foliage and subtly cool shadows. Keep a single landscape frame, realistic facial texture and stable identities throughout; no captions, subtitles, collage, face morphing, excessive grain or automatic cuts to each reference photograph.

[Shot 1] Use <Picture 1> to frame the established distance in an eye-level two-person view. <Subject 2> remains screen left at the library entrance, while <Subject 1> is farther screen right along the pavement, several metres away. Her feet and lower body still point along her rightward route; her long hair and anatomical-left tote settle naturally as she slows. Do not teleport either person closer or reverse their screen positions. Around 00:00.350, <Subject 2> (S1) uses slightly greater conversational projection to reach her, retaining his warm medium-low Mandarin timbre: <d>[Chinese] 喂！</d> The impulse is gentle urgency, never anger. She checks her step and turns her head and shoulders left without turning her entire body back toward the doorway. As <Subject 1> (S2), she responds in her established warm, clear voice, with familiar curiosity: <d>[Chinese] 又怎么啦</d> Let her full response finish by approximately 00:03.200. Both then listen silently. Leaves and a distant vehicle remain soft behind them; there is no extra call or spoken reaction.

[Shot 2] At 00:04.000, the camera cuts to a steady upper-chest close-up of <Subject 2>, guided by <Picture 2>. He occupies the left-middle of the frame with looking-room toward screen right. The woman's actual position remains several metres away and off-screen; the close-up changes viewing distance, not their physical separation. Soft reflected daylight makes his eyes readable, while the library's wood and olive foliage fall gently out of focus. Keep the exact approved hairline, eyes, nose, lips and gray-white shirt collar.

His tentative smile settles. He takes one short, quiet breath, steadies his gaze toward her and releases his jaw without a theatrical gesture. Around 00:05.000, <Subject 2> (S1) speaks once, clearly and sincerely: <d>[Chinese] 我养你啊。</d> Preserve his normal voice character while giving the promise firmer support and calm emphasis. Complete every syllable by approximately 00:06.800; do not stretch the pause until the sentence is forced against the ending.

After finishing, his lips close naturally. Hold his attentive, slightly exposed expression until 00:10.000, allowing a small exhale and almost imperceptible eye movement as he waits for her. Keep this reaction tail on the man; no new shot, repeated promise, female reply or additional sentence. Each of the three lines is spoken exactly once, with the supplied punctuation unchanged.

overall_soundscape:
Continuous low daytime street ambience, a light breeze in leaves and a few soft footsteps as the woman slows. Retain a quiet natural intake and release of breath around the close-up, without amplifying it into panting; no background chatter or source music.

non_diegetic_music:
N/A

subject_definitions:
<Subject 1> is the same 24-year-old East Asian woman in <Picture 1>, with her exact facial identity reinforced by <Picture 2> and her complete wardrobe by <Picture 3>. She is 172 cm tall, with long nearly black hair loosely half tied, soft strands beside her cheeks, a delicate necklace, a cream cable-knit cardigan over a pale inner top, a charcoal long A-line skirt, brown flat leather shoes, and an oatmeal canvas tote on her anatomical LEFT shoulder. Preserve one identity and one outfit throughout.
<Picture 1> is the B3 solo close-up storyboard reference for [Shot 1], defining the woman's screen-right placement, her screen-left eyeline toward the distant offscreen man, soft outdoor foliage, natural daylight, and restrained expression before tears fall.

summary:
[reference generation] Generate a 10-second continuous solo reaction shot of <Subject 1>, guided by <Picture 1>. She silently absorbs a sincere promise, gives her complete gentle reply, then breaks eye contact and begins leaving toward screen-right. Her full response and the beginning of her departure fit before approximately 6.5 seconds; the remaining time supplies natural movement for editing. She has not begun crying.

retention_analysis:
<Subject 1> (appears in [Shot 1]): fully_preserved - retain her exact referenced face, half-tied hair, necklace, outfit, age, proportions, and anatomical-left tote position while her expression and movement develop naturally.
<Picture 1> ([Shot 1] composition and emotional-state reference): fully_preserved - retain the solo outdoor close-up, screen-left look-space, soft foliage, warm skin, and gently held emotion; the man remains several metres away and completely outside the frame.

detailed_description:
Create a live-action 16:9 scene in gentle early-autumn daylight, with warm realistic skin, cream knit texture, olive foliage, soft highlights, and subtly cool shadows. Preserve natural skin detail and continuous photographic lighting without beauty-filter smoothing, added grain, or changes of time and weather.

[Shot 1] The camera holds a steady eye-level close-up following the composition of <Picture 1>. <Subject 1> occupies the right-middle of the frame, her face and cream cardigan readable against softly blurred outdoor pavement and trees. Her half-tied dark hair, loose cheek strands, and fine necklace match the references. Her tote remains on her anatomical left shoulder, even when most of it falls below the crop. Leave airy space on screen-left. She looks toward the man several metres away in that direction; no male face, shoulder, hand, reflection, or foreground silhouette enters this solo shot.

From 0.0 to 1.1 seconds, she stays silent and lets the promise register. Her eyes soften, her breath briefly catches without becoming a sob, and the corners of her mouth attempt a small familiar smile. Her chin stays almost level. The eyes may look moist, but there is no running tear or wet track on either cheek. A faint breeze moves only the loose hair strands. Quiet leaves and distant road ambience remain below her breathing.

From 1.2 to 5.6 seconds, <Subject 1> (S1) uses a warm, clear young-adult Mandarin voice in a natural medium-high register, soft resonance and relaxed articulation; gentle rather than childish or whispered. She says the complete line once: <d>[Chinese] 你先养好你自己吧，傻瓜</d> Her lips synchronize precisely with every syllable. The first clause sounds like a tender practical reminder, with light emphasis on looking after himself. Keep the comma pause brief and connected. The final affectionate address is quieter and warmer, carrying concern through softened articulation and a small release of breath; it is neither an insult nor a teasing laugh. Keep her face turned toward screen-left until the entire final word has finished. Do not rush, repeat, whisper away, or truncate the ending.

From 5.6 to 6.5 seconds, her lips close and the held smile becomes smaller. Only now does she lower her eyes slightly, release eye contact, and begin turning her head and shoulders toward screen-right. Her weight shifts into departure without an abrupt emotional change. From 6.5 to 10.0 seconds, she continues that same quiet turn and takes natural steps toward screen-right, with a slight cardigan and tote-strap movement. The camera stays steady as she moves toward the edge, leaving a clean continuation of the pavement behind. No second reply, new gesture, cut, or fresh emotional climax is introduced. Keep her cheeks free of tears throughout. Produce a single full-frame scene without subtitles, captions, arrows, panels, watermarks, or graphic overlays.

overall_soundscape:
Soft wind in street trees and distant traffic continue beneath a quiet breath and slight knit-and-canvas movement. Natural light footsteps begin only after she completes her response and starts leaving; avoid laughter, sobs, intelligible background voices, and artificial dramatic effects.

non_diegetic_music:
N/A

subject_definitions:
<Subject 1> is the same 24-year-old East Asian woman in <Picture 1> and <Picture 2>, with exact facial identity from <Picture 3> and full wardrobe from <Picture 4>. She is 172 cm tall, with long nearly black hair loosely half tied, a delicate necklace, a cream cable-knit cardigan, pale inner top, charcoal long A-line skirt, brown flat leather shoes, and one oatmeal canvas tote initially carried on her anatomical LEFT shoulder.
<Subject 3> is the same approximately 45-year-old Chinese male taxi driver visible in <Picture 1> and <Picture 2>, with short black hair and a gray-blue shirt. He occupies the far-side FRONT LEFT driving seat, faces the vehicle's front, and remains silent. Preserve the referenced seat-depth and occlusion relationships.
<Subject 4> is the same gray-blue four-door taxi in <Picture 1> and <Picture 2>, with its front at screen-left, rear at screen-right, ordinary dark fabric seats, and the curbside RIGHT REAR passenger door used by the woman. Preserve the car's body, window pillars, door hinge, handle, seat positions, and parked orientation.
<Picture 1> is the C1 opening-state anchor for [Shot 1]: the taxi has already stopped, the woman has lowered both hands and is approaching the closed right rear door, and the driver is visible through reflective glass.
<Picture 2> is the C2 opening-state anchor for [Shot 2] at 2.5 seconds: the same right rear door is open, the woman is preparing to enter, and the same driver remains partially visible in the far-side front left seat.

summary:
[keyframe completion + reference generation] Generate a 10-second, two-shot boarding sequence using <Picture 1> and <Picture 2>. <Subject 1> approaches the already parked <Subject 4>, opens its curbside right rear door, sits in the empty right rear seat, places her tote on her lap, and closes the door by approximately 6.7 seconds. <Subject 3> remains in the correct front left driving position. Nobody speaks, and the final seconds hold the completed closed-door state for editing.

retention_analysis:
<Subject 1> (appears in [Shot 1], [Shot 2]): fully_preserved - retain her face, proportions, hairstyle, necklace, outfit, and single tote; move the tote naturally from her anatomical left shoulder to her lap during boarding.
<Subject 3> (partly visible in [Shot 1], [Shot 2]): fully_preserved - preserve his identity, short hair, gray-blue shirt, front left driving position, forward orientation, and natural occlusion by the nearer seat and pillars.
<Subject 4> (appears in [Shot 1], [Shot 2]): fully_preserved - preserve taxi identity, front-left/rear-right screen direction, left-hand-drive layout, empty right rear boarding seat, and physically connected door geometry.
<Picture 1> ([Shot 1] opening state): fully_preserved - retain the stopped car, lowered hands, closed door, curb, street, and lighting.
<Picture 2> ([Shot 2] opening state at 2.5 seconds): fully_preserved - retain the already-open right rear door, same exterior viewpoint, entry position, and partial driver visibility.

detailed_description:
Create a live-action 16:9 sequence in continuous early-autumn daylight. Keep the warm cream clothing, gray-blue paint, olive street trees, natural glass reflections, and softly cool shadows consistent with both storyboard images; preserve realistic anatomy and the taxi's fixed physical structure.

[Shot 1] Begin from the parked state in <Picture 1>. A steady exterior medium-wide view shows <Subject 1>, with her referenced half-tied dark hair, cream cardigan, charcoal skirt, and brown flats, beside <Subject 4>. The car's front points screen-left and its rear remains screen-right. Both of her hands are already lowered; hailing has finished. Her single tote stays on her anatomical left shoulder, partly hidden on the side nearer the vehicle. Through the glass, <Subject 3> is a subdued, partially obscured figure in the far-side front left driving seat, facing forward in his gray-blue shirt.

From 0.0 to 2.5 seconds, she looks toward the right rear door and takes the small steps needed to reach it. Only after her body is within comfortable reach does one hand move to the exterior handle. Her fingers grip, the latch releases, and the door begins opening on its existing forward hinge. Keep her arms ordinary in length and her feet grounded beside the curb. The car remains stopped throughout. There is no raised-arm wave, driver greeting, horn, or crying. Hear two light footsteps, restrained fabric movement, and the small latch click.

[Shot 2] At 00:02.500, the shot cuts to the already-open door state established by <Picture 2>, continuing the same boarding action. Maintain the same street side, car direction, lighting, woman, and driver; the edit skips the middle of the door swing without resetting the action. Her posture is composed and her eyes are still free of running tears. <Subject 3> remains farther forward and across the cabin, naturally screened by the near-side seat and pillars; he does not occupy the rear doorway, right rear bench, or front right passenger seat, and does not turn to her.

From 2.5 to approximately 5.8 seconds, <Subject 1> steadies the tote, releases its strap from her left shoulder, lowers herself through the clear doorway, and brings both legs inside to sit in the empty right rear seat. She places the single tote on her lap. Keep the open door clear of her knees, skirt, hands, and bag, with continuous joints and believable seat depth. From approximately 5.8 to 6.7 seconds, after her body and belongings are fully inside, she pulls the door inward using its interior grip. It closes once with a natural hinge movement and a muted latch thud.

From 6.7 to 10.0 seconds, hold the stable closed-door exterior. The driver stays forward-facing and the woman is inside, softened by window reflections; no one changes seats or repeats the boarding. Leave seat-belt fastening for the time omitted before the next interior scene. Carry the low engine idle and street ambience continuously across the cut. No dialogue, radio speech, subtitles, arrows, split screens, watermarks, or added graphic elements appear.

overall_soundscape:
Maintain quiet roadside air, rustling trees, distant traffic, and the parked taxi's low engine idle. Synchronize short footsteps, canvas and cardigan movement, the exterior latch, subdued seat-fabric movement, and one final door-closing thud to the visible actions; all occupants remain silent.

non_diegetic_music:
N/A

subject_definitions:
<Subject 1> is the same 24-year-old East Asian woman shown in <Picture 1>, with her exact facial identity reinforced by <Picture 2> and her outfit and body proportions by <Picture 3>. She is 172 cm tall, with loosely half-tied long nearly black hair, cheek strands, a delicate necklace, cream cable-knit cardigan, pale inner top, charcoal long skirt, and brown flat leather shoes. Her single oatmeal canvas tote is now on her lap, with its strap off her shoulder.
<Subject 4> is the same gray-blue taxi whose right rear interior appears in <Picture 1> and whose exterior and boarding geometry are established by <Picture 4>. Retain the dark fabric rear bench, gray trim, rear windshield behind the bench, closed right-side passenger window at screen-left, and right rear seating position. The woman's belt runs from her anatomical RIGHT shoulder toward her anatomical LEFT hip.
<Picture 1> is the C3 emotional keyframe reference for [Shot 1] at approximately 6.5 seconds, defining the composition, right rear seat, screen-left window, belt direction, lap-held tote, and first fine tear track. It is a later emotional state, not the target video's already-crying first frame.

summary:
[keyframe completion + reference generation] Generate a 10-second continuous interior shot of <Subject 1> seated in <Subject 4> after boarding and fastening her belt during the omitted transition. Begin before tears have fallen, build through quiet restraint and a slight breath tremor, let the first tear move at approximately 5 seconds, and reach the emotional state of <Picture 1> near 6.5 seconds. Continue restrained silent reactions through the end. The daytime taxi is moving, and the driver remains outside the camera's backward-facing view.

retention_analysis:
<Subject 1> (appears in [Shot 1]): fully_preserved - retain the exact woman, hair, necklace, clothing, and single tote; her tears develop only during the specified timeline, while the tote stays on her lap.
<Subject 4> (interior visible in [Shot 1]): fully_preserved - retain right rear seat placement, screen-left side window, correctly oriented belt, dark rear bench, gray trim, and rear windshield, with consistent daylight and moving roadside trees.
<Picture 1> ([Shot 1] emotional keyframe at approximately 6.5 seconds): fully_preserved - match its composition, restrained face, lap-held tote, and first tear track at that later point; begin with compatible dry cheeks rather than copying the tearful state into the opening.

detailed_description:
Create a live-action 16:9 daytime interior with soft window light, warm natural skin, cream knit texture, dark woven upholstery, and passing olive-green foliage. Maintain the same gentle photographic treatment as the street sequence without a change to night, rain, artificial mood lighting, or a beauty-filter appearance.

[Shot 1] Hold a steady close-medium view from near the front-centre of <Subject 4>, looking backward toward the right rear passenger and the rear windshield. <Subject 1> sits in the right half of the frame, beside the CLOSED right-side window appearing at screen-left, following the composition of <Picture 1>. Her referenced half-tied dark hair, fine necklace, cream cardigan, and pale inner top remain unchanged. The dark rear bench extends behind her toward screen-right. Her safety belt is already fastened, descending from her anatomical RIGHT shoulder at viewer-left toward her anatomical LEFT hip at viewer-right. Her one oatmeal tote rests on her lap, its shoulder strap released, and both hands lie quietly on it. The driver remains in the front left driving seat behind the camera, entirely outside the image; no driver or additional passenger appears in the rear bench or window reflection.

From 0.0 to 2.0 seconds, begin with dry cheeks and no existing tear track, despite the later tearful state in <Picture 1>. She looks toward the passing street through the screen-left window, her lips closed and her face composed by effort. Her eyes are slightly wet, but she does not smile. Her shoulders stay still beneath the belt. The camera has only minute vibration consistent with the moving car, with no orbit, reverse angle, or abrupt push-in. Soft engine and road noise replace the more open roadside ambience.

From 2.0 to 4.8 seconds, her lower eyelids gather moisture. A small breath catches and then releases, producing a slight visible tremor at the mouth and upper chest without a voiced sob. Her fingers press very gently into the canvas bag and settle again. Her gaze remains near the window rather than seeking the camera. Keep the change gradual: no face-covering gesture, violent shaking, open-mouthed crying, or sudden shower of tears.

At approximately 5.0 seconds, the first visible tear leaves the lower eyelid and begins a narrow, natural path down the cheek facing the camera. By approximately 6.5 seconds, reach the restrained wet-eye expression and fine tear-track state of <Picture 1>, with the same window side, belt direction, posture, and lap-held bag. From 6.5 to 10.0 seconds, continue that quiet state: a small blink, shallow recovering breath, and minimal downward movement of the tear. She neither wipes it away nor adds another action.

Keep daylight and tree reflections drifting smoothly across the screen-left glass throughout, preserving exposure and the cabin geometry. No spoken line, whisper, narration, flashback, new cut, night rain, exaggerated sobbing, subtitle, arrow, collage, watermark, or graphic overlay is introduced. End with her still silently looking outward, leaving calm motion and breathing for the final edit.

overall_soundscape:
A subdued moving engine, soft tire noise, light cabin vibration, and faint air flow remain continuous. Her unvoiced breathing develops one small tremor before settling, with a barely audible canvas movement under her fingers; there is no radio, driver speech, intelligible background dialogue, or theatrical sobbing.

non_diegetic_music:
N/A