视频 · Seedance 2.5
纽约地铁里的座位正义
制作一段 30 秒写实动作喜剧:通勤者在深夜地铁中制止霸占座位和欺凌行为,并以克制、非伤害性的特技动作收束冲突。
Prompt
超写实美国真人实拍,符合现实的特技动作编排,动态手持摄影。
场景:深夜的纽约地铁里,角色A平静地坐着,一个高大傲慢的男人用背包占住两个座位,欺负一位乘客。她站起来说:“Move your bag.” 他居高临下地对她说:“Make me.”
接下来是一段节奏快速、具有连续感的逼真近身动作。男人伸手抓她,笨拙地挥拳、冲撞,并多次试图推搡她。角色A娴熟地躲闪,绕着地铁立柱转身,借力改变他的冲势,并以克制的动作反击。他多次踉跄撞上空座位和扶手,最后安全地落在软垫长椅上,背包弹落到他膝上。使用动态跟拍、紧凑特写、清晰可辨的全身动作编排、真实的平衡感、具有实际重量感的特技动作和肢体喜剧。
在第21秒,他试图起身。她平静地举起一只摊开的手,说:“We're done.” 他被围观的乘客看得尴尬,把背包抱到膝上,挪开位置,让被挤走的乘客坐下。最后三秒重现开场远景:她平静地坐在他对面,他举止礼貌,留出空座。车门打开时,她头也不回地离开。他轻声向那位乘客道歉。
视觉与声音规则:时长恰好30秒,24帧/秒,16:9横幅,人物身份与服装保持一致,真实的纽约地铁,冷色荧光灯照明,电影感35mm胶片质感,细微胶片颗粒,自然的列车运动,真实的反应,以及贴近现实的美国动作喜剧基调。仅使用以下声音:列车隆隆声、车轮尖啸声、脚步声、衣物摩擦声、座椅吱嘎声、用力声、对白和车门提示音。不要音乐、字幕、旁白、血液、血腥伤情、武器、超人般的动作、慢动作、变速效果或夸张的撞击。确保乘客安全,地铁内部布局保持一致。Ultra-realistic American live-action, grounded stunt choreography, dynamic handheld cinematography.
Scene: Inside a late-night NYC subway, Character A sits calmly as a huge, arrogant man blocks two seats with his backpack and bullies a commuter. She stands and says, “Move your bag.” He towers over her: “Make me.”
A fast, continuous-feeling sequence of realistic close-quarters action follows. The man reaches for her, swings clumsy punches, charges, and attempts repeated shoves. Character A skillfully dodges, pivots around subway poles, redirects his momentum, and uses controlled counters. He repeatedly stumbles into empty seats and handrails, finally landing safely on a padded bench with his backpack bouncing onto his lap. Dynamic tracking shots, tight close-ups, readable full-body choreography, realistic balance, practical stunt weight, and physical comedy.
At 21 seconds, he tries to rise. She calmly raises an open hand and says, “We're done.” Embarrassed by watching commuters, he pulls his backpack onto his lap and moves over to let the displaced commuter sit. In the final three seconds, recreate the opening wide shot: she sits calmly opposite him while he behaves respectfully, leaving seats available. As the doors open, she exits without looking back. He quietly apologizes to the commuter.
Visual & Audio Rules: Exactly 30 seconds, 24 fps, 16:9 landscape, consistent character identity and wardrobe, authentic NYC subway, cold fluorescent lighting, cinematic 35mm texture, subtle film grain, natural train movement, realistic reactions, and grounded American action-comedy tone. Audio only: train rumble, wheel squeals, footsteps, clothing movement, seat creaks, exertion, dialogue, and door chimes. No music, subtitles, narration, blood, graphic injuries, weapons, superhuman moves, slow motion, speed ramps, or exaggerated impacts. Keep passengers safe and subway layout consistent.
使用建议
使用建议:准备两名主要角色的外观与服装设定,确保整段保持身份一致;按开场对峙、连续近身动作、21 秒制止、最后三秒回到对照构图的节奏生成,并保留地铁布局、乘客反应、自然环境声和对白。使用支持 30 秒、24 fps、16:9、连续动作与对白控制的视频模型;动作应由专业特技团队设计或在生成后人工检查。







