Character video prompts

Explore video prompts where a person's actions carry the scene. Browse different settings and character moments, then adapt the behavior and staging to your own idea.

Explore prompts

34 prompts

Seedance 2.0 Mini Prompt for Xianxia Comedy Duel

SoranSeedance 2.0
你咋一天这么多事,要打就打
prompt:

电影级写实质感,纯古风中国仙侠美学。

把真正庄严的高手挑战,与 “提前包装江湖名声” 的元喜剧进行碰撞。
吸收:
默片式反应节奏

英式一本正经逻辑
港式喜剧快速铺垫与回收
经典“三拍递进”结构
采用:

Arri Alexa 电影机质感

稳定清晰的面部微细节
细腻胶片颗粒
自然体积光
核心喜剧主线

整段只保留一条第一遍就能立即看懂的喜剧主线:

一个极其自信的敌方剑客,为了证明自己即将获得一场传奇胜利,专门带来一名江湖记事先生“客观记录”;结果真正的决斗还没开始,图一图二两名女子就发现——敌人连最后怎么写都已经提前花钱定好了。

角色功能原则

记事先生始终只承担一本正经的辅助角色

年长师父只负责最后一次补刀
参考与环境
将 @图片 1 和 @图片 2 严格作为人物身份锚点。

本轮新上传的所有背景与地点参考图,共同决定同一套环境 DNA。
正式构图前,无声整合能够互相兼容的:

真实地形

建筑语言

材质年代
植被

水体
天气
山雾
主光方向
反光关系
空气纵深
人物真实可行走路线
再重新规划成本轮独有、完整统一的新空间。
背景原则
背景中的以下元素始终持续自然鲜活,但在叙事上绝对中立:
自然风

水体

植被
云层

远处普通弟子
空间环境声
角色设定
角色 ID A|剑仙师姐
始终是同一个 @图片 1 剑仙师姐:
25–30 岁东亚女性

身材高挑纤细

椭圆脸

深色杏眼
黑色长发半挽,以白玉簪固定

始终穿同一套白色刺绣真丝汉服
银色腰封
玉佩
白色布靴
持唯一一柄银色长剑
角色 ID B|小师妹
始终是同一个 @图片 2 小师妹:
20–25 岁东亚女性
身形娇小
圆润灵动的脸

黑发编辫
始终穿同一套青绿色亚麻汉服

深色腰带
木簪
黑色布鞋
持唯一一柄深色钢剑
其他角色
敌方剑客
气势十足
极其自信
非常在意自己的江湖名声

江湖记事先生

唯一一名中年记事先生
手中只有毛笔和一卷没有任何可读文字的素色纸卷
始终一本正经、职业化、认真执行自己的“记录工作”
年长师父

始终沉稳
只在最后完成一次一本正经的补刀
分段结构
0–5s|全景或远景

一名气势十足的敌方剑客进入根据本轮参考图自然形成的开放区域,身旁跟着唯一一名中年江湖记事先生。
记事先生手中只有毛笔和一卷没有任何可读文字的素色纸卷。
敌人拔剑,非常郑重地宣布:

“今日请先生作证,免得有人输了不认。”

同一个小师妹看了一眼旁边的记事先生,平静评价:

“还挺正式。”

同一个白衣剑仙从头到尾保持冷静。

5–10s|中景或牛仔景

保持:

同样人物

同样服装

同样长剑

同一个记事先生

同一卷纸

完全一致的地理空间

真正决斗还没有开始,记事先生突然非常职业化地稍微靠近敌人,小声确认:

“公子,还是按‘十招大胜’记?”
全场绝对安静半拍。
敌人的身体当场僵住,立刻压低声音恼火反问:
“不是说打完再写吗?”
记事先生一本正经回答:
“定金付的是这个结局。”

同一个小师妹非常缓慢地把眼神从记事先生移到敌人脸上,难以置信地问:

“结局都买好了?”

同一个剑仙师姐什么都没说,只极轻微地抬起一侧眉毛,第一次产生真正兴趣。

整个背景从头到尾持续自然运行,却完全不参与笑点。

10–15s|特写或极近特写

同一个白衣剑仙非常认真地问记事先生:

“若他输了呢?”

同一个记事先生几乎零停顿回答:

“补银子,可改成‘虽败犹荣’。”

敌人瞬间恼羞成怒:

“我不会输!”

镜头焦点转移到两人数步后方自然存在的同一个年长师父。

师父从头到尾保持绝对一本正经,只淡淡说道:

“那先替他记一句:未开打,先改稿。”

同一个记事先生非常职业化地点一下头,立即抬起毛笔准备继续记录。

敌人这一次真正慌了,马上喊道:

“这句别记!”

记事先生停顿半拍,看着他认真评价:

“这句也挺好。”

极近特写:

同一个小师妹迅速转开脸,肩膀出现极轻微颤动,拼命忍住笑

同一个剑仙师姐则依旧保持真正高手的庄重表情,非常礼貌地向敌人做出一个继续比武的手势:

“现在能打了吗?”

敌人先看看记事先生,再看看自己手中的剑,最后又看看后方师父,突然意识到现在真正岌岌可危的已经不是自己的身体,而是自己苦心经营的江湖名声。

在他回答之前精准切黑。

喜剧节奏原则

整段必须符合清楚的“三拍递进”:

第一拍

敌人郑重带着记事先生上场,像要见证一场传奇胜利。
第二拍

记事先生职业化暴露:胜负文案已经提前付费定好。

第三拍

师父一本正经补刀,记事先生准备如实记下,敌人终于发现真正危险的是自己的名声。

表演原则

敌方剑客

不是丑角表演

不是故意滑稽
必须认真维护自己的排场和体面

喜剧感只能来自他越想保住名声,越暴露自己提前包装结果
记事先生

必须极端职业化
一本正经

逻辑清楚

完全不觉得自己荒唐

他的喜剧感来自“过度认真”
剑仙师姐
全程冷静
不嘲讽
不抢台词

后半段才因为事情的荒谬性,第一次表现出真正兴趣
小师妹
反应要自然
不夸张大笑
重点是“忍笑”
年长师父

必须绝对平静
那句“未开打,先改稿”要像陈述事实,而不是故意抖包袱
硬性要求
严格总时长 15 秒
16:9 横屏

三个连续清晰镜头
原生同步普通话对白
精准口型
明确的喜剧停顿

针对 Seedance 2.0 Mini 强化多参考人物身份连续性
人物面部全程稳定
发型全程稳定

服装全程稳定

长剑全程稳定

记事先生道具全程稳定
配角位置全程稳定
地理空间全程稳定
人物走位简单清楚
真丝衣料和头发物理效果真实
前景、中景、远景保持自然视差与真实空间环境声
不生成字幕
Negative
blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent character, changing clothes, face morphing, background shift, glitching cuts, disappearing props

Cinematic Wuxia Night Scene

MahaSeedance 2.0
Create a 30-second cinematic, ultra-realistic wuxia-style video set at night in an ancient Chinese village, with traditional wooden buildings, a stone-paved street, a small arched bridge, misty mountains in the background, and many glowing red lanterns hanging along the street. A beautiful young woman in an elegant flowing red traditional Chinese Hanfu walks calmly toward the camera, with long dark hair styled traditionally and a serious, confident expression. As she moves closer, the camera slowly pushes in and follows her naturally, capturing the lantern light, fabric movement, atmospheric mist, and realistic footsteps. Around the middle of the scene, she suddenly stops and confidently draws a glowing red magical sword, holding it horizontally in front of her while her expression becomes focused and powerful. The camera moves into a dramatic close-up of her face and sword, showing realistic skin texture, detailed hair, glowing reflections, and cinematic depth of field. Several mysterious black-clothed attackers appear around her, rushing toward her from different directions, creating a tense martial-arts atmosphere. She swiftly turns and performs powerful choreographed sword movements, blocking and striking the attackers with fast but believable movements while the glowing sword leaves subtle red light trails. The attackers are knocked down onto the stone street around her, while she remains standing confidently in the center with her sword raised, surrounded by red lanterns and light mist. End with a wide cinematic shot slowly pulling back to reveal the woman standing alone among the fallen attackers, ancient village buildings and mountains behind her, maintaining consistent character appearance, realistic motion, cinematic lighting, detailed textures, dramatic atmosphere, smooth camera movement, 16:9 aspect ratio, 30 seconds, no text, no watermark.

Seedance 2.0 Cinematic Optical Zoom Video Prompt

灰度笔记Seedance 2.0
【全局设定】
古风写实电影镜头,东方古典生活叙事,雨后江南古镇:青石台阶、青砖黛瓦、木质廊檐、雕花木门、竹帘、湿润石板巷道。阴天柔和漫射光,低对比度,冷灰青与温暖木色融合,轻微35mm胶片颗粒,真实皮肤、木石与织物纹理,电影级浅景深,8K,16:9。

女主A:图1
女主B:图2

全片严格保持角色身份、空间位置、左右关系、行进方向、天气、光线、色调连续。

【核心摄影技法】
全片以“渐进式光学变焦”为唯一核心运镜。摄影机尽量固定机位,不依赖实体移动,通过缓慢连续的光学推近、拉远改变人物画面占比与空间关系。保持真实镜头压缩、景深变化和光学特征,禁止数字裁切。前景遮挡、浅景深、焦点转移辅助叙事,形成“远处发现→隔巷对视→空间压缩→情绪确认”。

【0-4s】
中景起幅,35mm,固定机位,视线高度。女主A沿青石台阶缓慢下行至巷口,停步转身抬头,看见远处廊檐下的女主B。镜头从较宽环境关系构图开始,仅缓慢光学推近,逐渐收紧至女主A中景,摄影机不移动。

古镇行人先后从左右前景横向经过,近距离形成自然虚化遮挡,女主A始终保持清晰;遮挡期间曝光、色温、光影稳定。变焦持续平滑克制,最终形成女主A与远处女主B隔巷对视。

【4-9s】
承接上一段,机位完全不变。女主B从画面右侧前景缓慢进入,靠近木质廊柱,与廊柱形成天然前景框;女主A位于左中部巷道深处。

缓慢连续光学变焦,女主A先保持清晰,随后画面逐渐收紧,女主B前景占比增大,女主A退入背景并柔和虚化。逐渐进入长焦端,通过真实空间压缩使古巷纵深被压缩,两人实际距离不变,但视觉距离明显拉近。焦点平滑由女主A转向女主B,最终女主B侧脸清晰、女主A虚化。摄影机不横移、不摇摄、不绕拍,曝光与色调连续。

【9-13s】
保持固定机位与长焦关系。女主B站在木质廊柱旁缓慢转身,看向女主A。仅进行一次极缓慢光学推近,从双人关系构图收紧至女主B中近景。画面左侧保留女主A局部虚化轮廓,形成窥视式前景框架。

女主B成为主要焦点,看向女主A,短暂停顿后轻声说:“好久不见。”露出极轻微笑意。竹帘被雨后微风吹动,屋檐水滴自然落下,湿润青石反射柔和天光。镜头稳定推进,在近距离情绪确认中结束。

【镜头控制】
核心必须是“光学变焦制造叙事距离变化”,而非摄影机移动。镜头语言依次完成:环境建立→发现女主A→缓慢收紧→女主B进入前景→长焦空间压缩→焦点转移→情绪确认。变焦速度缓慢连续,具有真实镜头惯性;保持自然人物比例、真实景深与稳定曝光,明确呈现光学变焦感,但无廉价数码放大感。

【连续性与负面约束】
严格遵循角色卡,禁止外貌、发型、服装漂移;保持空间地理关系、左右位置、运动方向、光线方向一致。禁止现代建筑、交通工具、路牌、现代物件、字幕、水印;禁止新增主要人物、人物重复、肢体异常、身份交换。禁止横移、环绕、突然摇镜、无人机运动、强手持抖动、数字变焦、数码裁切、透视异常、焦距跳变、景深瞬切、曝光闪烁、色温漂移、过度雾化、过度磨皮、AI塑料皮肤。

Bride Escapes Wedding in Vintage Car

Caden FluxSeedance 2.0
Use the character sheet as the visual reference. Preserve the bride’s exact face, hairstyle, dress, body proportions, and accessories throughout the entire video. 16:9 landscape.

0–3 sec — THE ESCAPE
A church wedding is happening when the bride suddenly bursts through the giant church doors and sprints down the steps, still holding her bouquet. Her expression is determined but hilariously excited. Her veil and dress fly wildly behind her. Camera rapidly tracks backward in front of her as she runs. A few confused wedding guests appear in the background, realizing what is happening.
3–5 sec — GET TO THE CAR
She spots a bright red 1950s vintage car waiting at the curb. She kicks off one white bridal heel while running, then throws herself toward the driver's door. Fast whip-pan follows her movement.

5–8 sec — PEEL OUT
She jumps into the car, slams the door, starts the engine and PEELS AWAY at ridiculous speed. Tires squeal, a tiny cloud of smoke appears, and her long white veil gets caught in the closing door and streams dramatically behind the car. The bouquet lands on the passenger seat.
8–11 sec — CHAOS
Rapid comedic interior cuts:
Bride gripping the steering wheel and laughing → her veil whipping around her face → bouquet bouncing around the passenger seat → her other bridal heel flying through the air → wedding guests running helplessly behind the disappearing car.

11–15 sec — SHE’S FREE
Dynamic side tracking shot of the red car racing down a sunny vintage highway. The bride drives one-handed, completely carefree. She looks into the rearview mirror, sees the church disappearing in the distance, grins, then blows a huge pink bubblegum bubble.
Final frame:
The red car speeds toward the horizon, veil flying behind it.

Style: fast-paced physical comedy, playful rebellious energy, 1950s Americana, saturated red vintage car, creamy white wedding dress, warm golden sunlight, cinematic vintage film grain, slightly painterly realism, realistic motion, energetic camera movement, whip pans, quick cuts, exaggerated but believable comedy.
Important: No slow motion, no sad mood, no dramatic romance, no dialogue, no text, no extra brides, no face changes, no outfit changes, no distorted hands, no morphing.

Educational Kids Animation Fruit Hunt

Abkr Sadiq | AISeedance 2.5
Create a high-quality 15-second vertical 9:16 3D animated educational kids video in a bright, colorful, cute nursery-rhyme style.

Main character:
A cute 2-year-old toddler boy with chubby cheeks, big expressive brown eyes, short light-brown hair, a bright blue T-shirt, yellow shorts, and colorful sneakers. Keep his appearance perfectly consistent throughout the entire video.

Story concept:
The toddler discovers a glowing magical map and follows it on a very short learning adventure to help a cute rabbit find three missing fruits.

0–3 sec — The Discovery

The toddler is playing in a colorful room when a glowing rainbow map suddenly appears.

The map shows three missing fruits:

🍎 Apple
🍌 Banana
🍊 Orange

A cute rabbit appears looking worried.

The toddler points toward the map and excitedly begins the adventure.

3–6 sec — Find the Apple

Fast magical transition into a colorful fruit garden.

A shiny red apple falls gently from a tree.

The toddler catches it and smiles.

Display large clear text:

“APPLE 🍎”

6–9 sec — Find the Banana

The toddler runs a few steps and notices a bright yellow banana.

He picks it up happily.

Display:

“BANANA 🍌”

9–12 sec — Find the Orange

A bright orange rolls across the grass toward the toddler.

He stops it with both hands and laughs.

Display:

“ORANGE 🍊”

12–15 sec — Happy Ending

The toddler places all three fruits into the rabbit’s basket.

The rabbit jumps happily.

A colorful rainbow appears behind them.

The toddler and rabbit wave toward the camera.

Display:

“Great Job! 🌈”

Animation:
Fast but smooth storytelling, natural toddler movements, expressive reactions, accurate hand-object interaction, playful running, picking up fruit, smiling, pointing, and waving.

Camera:
Vertical 9:16, quick child-height tracking shots, gentle push-ins, clear close-ups of each fruit, smooth magical transitions, no confusing cuts.

Visual style:
Premium polished 3D children’s animation, vibrant colors, soft cinematic lighting, glossy surfaces, cute rounded shapes, cheerful preschool atmosphere, clean backgrounds, expressive character animation.

Important:
Keep the entire video within 15 seconds.
Use only one main action per shot.
Keep text large and readable.
Do not overcrowd scenes.
No character morphing, extra fingers, duplicated objects, inconsistent clothing, flickering, warped fruit, or sudden background changes.

Early 2000s Tokyo Family Dinner Vlog

𝗦𝗮𝗻𝗶𝗮Seedance 2.5
Realistic personal Japanese family dinner vlog set in early-2000s Tokyo, filmed casually on an authentic consumer DV camcorder. Use the SAME young Japanese woman from the reference image throughout. Keep her exact facial identity, hair, outfit, handbag and appearance consistent.

EXACT ORDER:

0–4s — LOCAL SUPERMARKET: She enters a small early-2000s Tokyo supermarket, takes a basket and walks through the aisles.

4–9s — CHOOSE INGREDIENTS: She selects vegetables, tofu, mushrooms, rice/noodles and fish or chicken. Clearly show her examining and placing them into the basket.

9–12s — CASH PAYMENT: At checkout, the cashier scans the groceries. She pays with Japanese yen cash, receives a receipt and places the groceries into a shopping bag before leaving.

12–16s — RETURN HOME: She walks through a realistic early-2000s Tokyo neighborhood carrying the same grocery bag, then unlocks her home and enters.

16–22s — PREPARE DINNER: In a modest early-2000s Japanese kitchen, she takes out the same ingredients, washes and cuts them, then cooks a simple Japanese dinner using period-appropriate cookware.

22–26s — SET THE TABLE: She brings the finished food to the dining table and sets rice, soup, vegetables/protein, bowls and chopsticks. Her family joins and sits down.

26–30s — FAMILY DINNER: The family eats together, casually talks and smiles. She takes a bite and briefly looks toward the camera with a natural smile before continuing dinner.

ERA LOCK: Everything must authentically belong to early-2000s Japan—fashion, supermarket, packaging, cars, appliances, kitchen, furniture, television and tableware. No smartphones, modern LED screens, AirPods, modern cars, contemporary interiors or current packaging.

CAMERA STYLE: Raw DV footage with subtle handheld shake, imperfect framing, autofocus hunting, exposure shifts, soft digital detail, mild CCD/DV noise, motion blur and occasional awkward zooms. Natural home-video behavior, realistic skin and movement. No 4K sharpness, stabilization, beauty filters, VHS effects or cinematic lighting.

CONTINUITY: Same woman, outfit, hairstyle, bag, ingredients and family throughout. The purchased ingredients must be the same ones used for cooking and served at dinner. No teleportation, identity drift, disappearing props, reordered scenes or modern elements.

FINAL FEEL: A genuine early-2000s Japanese family memory—from grocery shopping and cooking to sitting down for an ordinary family dinner.

Seedance 2 Travel Video Prompt: Balcony Scene with Wind Interaction

Adam也叫吉米Seedance 2.0
保持参考图中的人物五官、发型、蓝白碎花吊带裙、身材比例、阳台、桌面物品和海景环境一致,不换脸,不改变服装。

10秒真实旅行抓拍视频,像朋友在酒店阳台随手拍摄。

开始时,年轻亚洲女性自然坐在阳台沙发上,侧头安静看着远处海面和夕阳,海风轻轻吹动长发和裙摆,人物自然眨眼,身体有轻微呼吸起伏。

约第3秒,一阵突然变大的海风从侧面吹来,桌上的一张轻薄餐巾被吹起来,沿着桌面滑向阳台边缘。女生先愣一下,低头看到后,下意识快速伸手去抓。

约第5秒,她差一点没抓住,身体微微向前探,头发也被风吹乱。最后成功抓住餐巾,她先低头笑了一下,然后带着一点无奈的表情重新靠回沙发。

最后2秒,她把餐巾压在桌面上,一只手顺势整理被风吹乱的头发,再抬头看向远处夕阳,嘴角还有轻微笑意。

整段保持真实手机抓拍感,镜头轻微手持,不环绕,不快速变焦,只允许非常缓慢自然的向前推进。海面波光、远处帆船、窗帘和发丝都保持自然运动。

强调真实人体动作和重力惯性,人物面部始终稳定,手指正常,裙子结构稳定,桌子和饮料不变形。

避免:人物突然站起来、夸张大笑、餐巾飞得太高、身体大幅扭曲、脸部漂移、换脸、手指畸形、衣服变化、桌上物品突然消失、强烈电影运镜。

Attack on Titan AAA Fantasy Rumbling

MadMaxSeedance 2.5
全片生物、面部近景、建筑和远景统一为明显电脑生成的高品质AAA奇幻游戏过场动画,使用数字雕刻脸模、经过美术设计的材质光影与成束头发,不混入实拍演员,不采用2D动画、赛璐璐描边或普通gameplay。整体是由连续硬切连接的末日奇幻战争蒙太奇:巨型军团推进、远超军团尺度的骨架生物显露、城内两只巨人短暂肉搏,最后切到另一只猿形巨人的咆哮蓄能;不同生物保持各自身份,不把这些切镜做成同一角色连续变形。环境与色彩:密集的中世纪半木构城市,红褐瓦顶、深棕木梁、浅米色墙面、窄石街、烟囱、尖顶塔楼与高耸外围石墙;所有普通房屋、街道和塔楼都远小于巨人,战斗者的腰胸高过附近屋顶,巨型军团更高,远处骨架生物又远超整个军团。厚重灰褐烟云和橙金逆光覆盖城市,浮尘、碎瓦和蒸汽形成多层纵深;开头为浓重焦褐尘雾,中段为暖金天空照亮的城内废墟,结尾为巨大圆形金色逆光与翻卷云雾,保持同一末日暖金色系。A巨型军团:多排无发的超巨大成年男性化人形生物,宽肩厚胸、深红肌束与浅象牙腱膜、深陷眼窝和清楚齿列,肋部及肩胸表面具有连续雕刻组织;这是生物本来的完整外观,不是新鲜伤口,没有流血。它们空手、无衣甲,前后排列有真实纵深,头肩与步幅略有差异,沿同一个方向持续缓慢行进,身体周围蒸汽升腾,脚下带起尘浪,不是静止观众或重复粘贴的同一人。B骨架巨构:军团后方仅有一个远比它们庞大的非人骨架生命体,浅褐象牙色中央脊柱高耸,两侧大量狭长弧形肋骨向外展开再向下弯曲,像覆盖军团上空的巨大骨架穹顶,肋弓之间有真实空隙,不是翅膀或建筑。前部低悬着一张狭长的红褐骨质面孔,深眼窝、尖长耳、露齿下颌,黑色长发成束垂落在面部两侧,头部通过颈部连接同一骨架,不是独立漂浮的头。最初眼睛暗,随后两眼亮起翠绿色光。C黑发硬拳巨人:中段肉搏主角之一,强健修长的人形非人生物,棕褐皮肤与清楚胸腹肌束,肩背宽而腰收,长深棕黑发、尖耳、强下颌与齿列,双眼翠绿发光。两只拳头及下段前臂覆盖浅灰白色粗糙硬化结晶,晶层依附真实拳指与腕部,仍是同一双手,不是手持石块或拳套武器;肘关节以上保留原本肌肉形态。D浅金骨甲巨人:另一名肉搏者,体量与C接近但更宽厚,短浅色头发、浅金色眼光,脸、胸、背、肩和四肢有层叠浅金褐色生物骨甲,深褐肌束连接甲片之间,甲片是身体的一部分,不是穿戴金属盔甲。空手,两臂与头部全程完整,肩颈逸出薄蒸汽。E黑鬃猿形巨人:末段独立角色,厚重黑褐毛发覆盖巨大肩背与极长双臂,躯干前倾,腿较短且强健,沉重猿人面孔、后梳深色头发、突出眉弓、尖耳与猩红眼睛,嘴中有成排尖齿。它与C、D和B是不同生物;长臂、黑毛与红眼始终保留。0.00–2.20秒:立即从浓重橙褐烟尘中的略倾斜近中景开始,前景破碎屋脊、墙板与石块遮住画面下部,数名A的巨大无发头颅与肩胸从废墟后方逼近,后面还能看见更大的模糊肋骨轮廓。摄影机沿废墟缓慢侧移,近处残墙与碎片快速横过,远处头颅较慢移动形成尺度与parallax;蒸汽和尘流不断穿过它们面前。2.20–4.20秒:切到接近地面的斜侧宽景,整排A从画面近侧延伸到远处,宽肩、长臂与肌束胸腹从烟中显出,身体随沉重步伐交替起伏,手臂自然摆动。摄影机沿军团方向缓慢退移,前景滚动尘浪,后排继续推进,不能让军团站定摆姿势。4.20–7.20秒:硬切正面低角极宽景,A的前排头肩横贯画面下方,B的巨大中央骨架与两侧肋弓在它们后方高高耸起。镜头缓慢上仰并略微前推,烟云先遮住上部结构再被气流拨开,低悬长发面孔位于中央骨架下方。两次短促金色闪电在远后方云层中亮起,瞬间照亮肋骨边缘、云团与军团轮廓后迅速衰退,闪电不是射向士兵的攻击,也不把已有生物变成其他形态。7.20–9.60秒:切B低悬面孔的正面紧近景,巨大的弧形肋骨仍在头后形成连续背景。黑发束垂在两颊旁,尖耳向外,头部微微抬正;先保留深暗眼窝,随后两眼由暗变为清晰翠绿光,视线凝向前方。烟和小尘粒持续通过镜头,不让脸冻结,不添加对白或眼部光束。9.60–11.60秒:以烟雾中的硬切转入另一片城内街区,低角正面中全景,C从灰褐尘幕中向前迈步显露,双拳低垂,胸腹与肩背随步伐运动,两侧较小塔楼和红瓦屋顶明确显示巨人尺度。镜头略向前接近,翠绿眼光透过黑发可见,两只浅灰硬化拳和前臂同时保持清晰;这不是B缩小成人形,也不复制出第二个C。11.60–12.40秒:突切向上仰拍的云天,一具蜷曲的巨大人形轮廓处于金色分叉闪电和近圆形发光蒸汽团中央,身体向下坠近,一臂抬过头侧,一膝收起。电弧沿周围空气和身体轮廓分叉,强光短暂压暗背景。只展示这个有实体身体的空中电光片段,不插入小人长大或从C体内分裂出的过程。12.40–13.20秒:切城内贴石街的倾斜低机位,远处房屋之间D已处于落地后的低蹲,石块和瓦片被震起,贴地尘浪沿街涌向镜头;落地位置与尘浪源点一致。高墙和红瓦屋顶上方可见一艘遥远的浅色长椭圆飞艇,缓慢漂移,不发射、不爆炸。镜头随地面冲击短震,屋顶和瓦片为巨人尺寸参照。13.20–14.20秒:切贴近D侧面的快速跟拍,D从低姿势向前冲出,躯干前倾、双脚交替蹬地、弯肘摆臂,肩颈蒸汽向后拖;浅金骨甲保持固定在同一身体上,双手空着,背景瓦屋和烟囱迅速后移。14.20–15.00秒:切C略低角胸上近景,C迅速转头、转肩看向冲来的D,长黑发被甩向一侧,翠绿眼睛锁定对手,下颌绷紧,身体开始转髋迎击;镜头贴近其面部短促跟转,不停下来重新对峙。15.00–16.00秒:切街道低位双人宽景,C在screen-left,D在screen-right,两者在小型房屋之间快速缩短距离。C踩稳前脚、后脚抓地,右肩和右髋向后蓄力,右侧硬化拳明显拉到身后,左臂在前方保持平衡;D继续向C压近并抬臂,瓦片和灰尘从两者脚边翻起。完整看见脚底支撑、转髋与拳头准备,不出现瞬移换边。16.00–17.40秒:摄影机快速贴过C肩侧并跟随右臂前送,C由后脚蹬地、髋部旋转带动肩和右侧硬化拳,沿单一清楚弧线打向D面侧。黑发向后甩,眼部转动带出极短翠绿拖光而不是激光;拳头逼近时短暂放慢以看清轨迹,随后清楚接触D脸侧骨甲。接触产生紧凑粉尘、小硬质碎屑与少量火星,D的头和肩先受力偏转,再带动上半身后仰。拳臂没有穿过头颅,没有血或断裂肢体,摄影机在接触瞬间短促震动。17.40–18.50秒:顺接倾斜的D近中景,他的头、肩和胸甲沿第一拳的力量方向扭转后退,两臂被甩开后开始重新收拢,脚下在瓦砾间调整支撑。镜头随上半身退移,下方红瓦屋顶保持远小于他,C仍在同一侧近处,不把这一反应剪成新变身或完全倒地。18.50–20.40秒:切回较宽的双人侧面构图,C再次把右侧硬化拳抬高至头上方形成威胁,另一只硬化拳处于较低前方。D向前顶身并...

Whip Pan Camera Motion AI Video

Travis DavidsSeedance 2.5
0:00–0:05 — Maasai village, firelit clearing: The camera moves low among a Maasai fireside gathering, weaving past men in red shúkà cloth seated in conversation, some sharing a gourd of fermented milk. It settles on a circle of clapping men as a warrior mid-leap rises impossibly high in the adumu jumping dance, beads and ochre-dyed hair swinging, firelight flaring across his face. Crash zoom to the calabash drummer's hands, striking at lightning speed. 

0:05–0:10 — Whip pan to Mongolian steppe camp: A sudden fast 360-degree whip pan cuts across darkness and lands on a Mongolian steppe camp, a roaring bonfire outside a felt ger tent. A throat-singer crouches low, chest and throat resonating with a deep guttural drone, while beside him a horsehead fiddle player draws long, sorrowful notes, firelight catching the carved horse head at the instrument's neck. Rapid focus pull to the fiddle player's bow arm sawing with practiced intensity. 

0:10–0:15 — Whip pan to Viking hall: A violent whip pan hurls the camera into a smoke-filled Viking longhouse, torchlight catching axe blades and shields mounted on timber walls. Bearded warriors pound wooden tankards on long tables in unison, chanting low and building, a horn-blower raising a great curved war horn and sounding a deep bellow that cuts through the hall. 

0:15–0:20 — Whip pan to Plains fire circle: Another hard whip pan lands on a Native American Plains fire circle at night, a hide drum group seated tight around a large ceremonial drum, mallets striking in unified, driving rhythm. Quick whip pan to a dancer in feathered regalia spinning through rising sparks, fringe and feathers blurring with motion. 

0:20–0:25 — Whip pan to Māori haka: A final sharp whip pan crashes into a Māori gathering at night, warriors performing the haka in fierce unison — stamping feet shaking loose earth, tongues bared, eyes wide, hands slapping chests and thighs in thunderous rhythm. 

0:25–0:30 — Rhythmic collapse, hold on fire: The camera pulls back slightly but stays within the Māori circle, holding steady as the stamping rhythm reaches its peak, dust rising and catching firelight — the frame holds on their pounding feet and fierce expressions as the sequence closes.

GTA Style Bank Heist Game Sequence

AqsaSeedance 2.5
Create a 30-second ultra-realistic AAA open-world third-person action game sequence, fully rendered as fictional GTA-style gameplay from beginning to end. Generate the female player character yourself: a stylish young woman with long dark hair, natural facial features, dark fitted jacket, dark pants and practical boots. Keep her appearance perfectly consistent throughout the entire video. Set the mission in a highly detailed modern city bank with realistic interiors, security guards, vault, cash, police vehicles, traffic and NPCs. Use a true third-person gameplay camera behind the character, realistic game physics, animations, lighting, reflections and a complete GTA-style HUD with minimap, health, mission objectives and notifications. At 0–4 seconds, she enters the bank and the HUD shows MISSION START — BANK HEIST — STEAL THE CASH. At 4–8 seconds, she quickly bypasses security and reaches the restricted vault corridor; HUD: SECURITY BYPASSED. At 8–14 seconds, she reaches the vault, opens it and rapidly fills a duffel bag with cash; HUD: CASH SECURED. At 14–18 seconds, the alarm activates, red warning lights flash and NPCs panic as she runs toward the exit; HUD: ALARM TRIGGERED — ESCAPE. At 18–24 seconds, she exits, gets into a waiting getaway car and drives away while police vehicles begin chasing her through busy city streets. At 24–27 seconds, she makes a sharp turn through a narrow street and loses the police, with the minimap showing ESCAPE ROUTE. At 27 seconds, make exactly one hard cut to a natural third-person gameplay continuation at a hidden safehouse, where she steps out carrying the cash-filled duffel bag. HUD displays MISSION PASSED — CASH SECURED — ESCAPE SUCCESSFUL — RANK S. Make the entire video look like authentic high-budget GTA gameplay, including game-engine visuals, HUD, minimap, NPC behavior, vehicle physics and camera movement. No live-action look, no cinematic movie style, no anime, no cartoon, no random cuts, no teleportation and no graphic gore.

Early 2000s Korean Teacher Vlog

SairaSeedance 2.5
Create a 30-second ultra-realistic morning vlog set in a small town neighborhood in South Korea in the early 2000s, filmed casually by a friend on an authentic consumer DV camcorder. It must feel like genuine home-video footage, not a modern video with a retro filter.

Use the provided reference image for the woman's face, hairstyle and outfit. Keep her identical in every scene.

EXACT ORDER — 6 SCENES:

0–5s — WASHING FACE: Early morning, soft sunrise light. The same young Korean woman (late 20s, hot pink sleeveless collared top, white high-waist jeans, white sneakers, small hoop earrings, butterfly pendant necklace, hair in a messy bun) steps out of her modest house into the small yard. An old leaking outdoor steel tap is running. She cups water in her hands, washes her face, and dries it with a small towel hanging nearby.

5–8s — GETTING READY: Fresh and smiling, she picks up a brown canvas shoulder bag and a few textbooks from a small outdoor table, glances at the camera, and heads out through the metal gate.

8–14s — THE WALK: She walks through a narrow early-2000s Korean neighborhood alley with small shops opening, an old delivery scooter, a bicycle passing. Korean elementary school kids in casual clothes, caps and yellow backpacks greet her cheerfully ("안녕하세요, 선생님!"). She smiles and waves. The same kids keep walking toward their school behind her.

14–18s — ARRIVING AT ACADEMY (HAGWON): She enters a small early-2000s private tutoring classroom with a green chalkboard, simple wooden desks, an old wall clock and a few faded posters. About 6–8 children stand up to greet her. She smiles warmly and gestures for them to sit.

18–23s — HER SHORT SPEECH: Standing in front of the class, she smiles and speaks a few warm words to the children. The children listen, smile, and clap softly.

23–30s — LECTURE BEGINS: She picks up a piece of chalk, writes the lesson title on the chalkboard, turns to the children and begins explaining. Children raise hands and take notes with pencils. She nods encouragingly.

CONTINUITY: Same woman, same hot pink top, white jeans, white sneakers, hairstyle, brown canvas bag and books throughout. The same school kids appear on the road. Same children in the classroom. No teleporting, identity drift, outfit changes, disappearing props or reordered actions.

ERA: Everything must look authentically early 2000s Korean: hand-painted Hangul shop signs, small neighborhood "슈퍼" grocery stores, low-rise brick villas, tangled overhead wires, old compact cars (Daewoo Matiz, Hyundai Accent, Kia Pride), classroom furniture and school supplies of that time. No smartphones, modern cars, LED screens or contemporary interiors.

CAMERA: Raw consumer DV camcorder footage, 4:3 frame, with subtle handheld shake, imperfect framing, autofocus hunting, exposure shifts, soft digital detail, soft slightly washed-out skin tones, mild CCD/DV noise, motion blur, compression artifacts.

Seedance 2.0 Mini Prompt for Xianxia Identity Reveal Comedy

SoranSeedance 2.0
什么寒川无影客,还是不如王二狗顺耳
prompt:

电影级写实质感,纯古风中国仙侠美学。

把史诗级高手自报家门的庄严镜头语法,逐渐瓦解成一本正经的 “身份掉马” 喜剧。
吸收:
默片式反应节奏

英式身份降格冷幽默
港式喜剧快速铺垫与回收
清楚的“三拍递进”
采用:

Arri Alexa 电影机质感

稳定清晰的面部微细节
细腻胶片颗粒
自然体积光
核心喜剧主线

整段只保留一条第一遍就能立即看懂的喜剧主线:

一个气势极强的敌方剑客,想用越来越夸张的江湖名号压住全场;师姐却从头到尾只想知道:

“你到底叫什么?”

最后年长师父突然认出他小时候极其普通、甚至有点尴尬的旧名,一句话就把敌人辛苦经营多年的绝世高手人设彻底打回原形。

笑点原则

核心笑点来自:

身份地位反差
熟人社会的精准拆台

绝不能依赖:
人物突然变蠢

环境事故

外力巧合
参考与环境
将 @图片 1 和 @图片 2 严格作为人物身份锚点。

本轮新上传的所有背景与地点参考图,共同决定同一套环境 DNA。
保留能够辨认的:

真实地形

建筑

材质
植被

水体
天气
山雾
主光方向
倒影
空气纵深
同时重新规划成本轮独有、完整统一的新空间。
背景原则
背景中的以下元素持续鲜活存在,但在叙事上绝对中立:
自然风

水体

植被
云层

远处普通弟子
空间环境声
角色设定
角色 ID A|剑仙师姐
始终是同一个 @图片 1 剑仙师姐:
25–30 岁东亚女性

身材高挑纤细

椭圆脸

深色杏眼
黑色长发半挽,以白玉簪固定

始终穿同一套白色刺绣真丝汉服
银色腰封
玉佩
白色布靴
持唯一一柄银色长剑
角色 ID B|小师妹
始终是同一个 @图片 2 小师妹:
20–25 岁东亚女性
身形娇小
圆润灵动的脸

黑发编辫
始终穿同一套青绿色亚麻汉服

深色腰带
木簪
黑色布鞋
持唯一一柄深色钢剑
其他角色
敌方剑客
气势极强
出场充满仪式感
极度依赖自己经营多年的江湖名号

强烈维持“绝世高手”人设

年长师父
从头到尾一本正经
并非故意羞辱对方
只是自然认出熟人旧名
最后再补最后一刀

分段结构
0–5s|全景或远景
一名气势极强的敌方剑客大步进入根据本轮参考图自然形成的开放区域。
他极有仪式感地掀开外层披风,一只手压住剑柄,以真正绝世高手登场的姿态高声宣布:
“北岭孤鸿,断江一剑,今日特来领教!”

同一个白衣剑仙完全没有被气势影响,只非常平静地问:

“你叫什么?”

敌人明显停顿一下,随后反而更加骄傲地继续:

“江湖人称 —— 寒川无影客!”

5–10s|中景或牛仔景

保持:

同样人物

同样服装

同样长剑

完全一致的地理空间

同一个小师妹慢慢转眼看向师姐。

同一个师姐没有任何嘲笑,仍然非常认真地再次确认:

“我是问名字。”

敌人终于出现一点不自然,却依然努力维持高手气质:
“名号就是名字。”
浅景深后方,同一个年长师父突然眯起眼睛,非常认真地盯着敌人的脸看了一拍。
随后像终于认出一个多年没见的熟人一样,突然问道:

“王二狗?”

全场绝对安静。

敌人整个身体瞬间僵住,原本挺得极直的英雄式姿态,肉眼可见地垮下去一点。

整个背景继续自然鲜活地运行,却完全不参与这个反转。

10–15s|特写或极近特写

同一个敌人难以置信地转头看向师父:

“您…… 认识我?”

师父完全没有意识到自己正在拆台,只非常自然地回答:

“你小时候偷我枣,被狗追过三条街。”

同一个小师妹瞬间低下头,肩膀出现极小幅度颤动,拼命憋住笑。

敌人赶紧重新转回师姐,试图挽回最后一点江湖尊严,压低声音认真解释:

“江湖已经没人这么叫了。”

同一个白衣剑仙听完之后非常郑重地点一下头,表示完全理解,然后平静回答:

“明白,二狗。”

极近特写:

敌人刚刚勉强恢复的一点表情再次彻底垮掉

同一个小师妹终于忍不住迅速转开脸,短促笑出一声

后方同一个年长师父再一本正经补最后一刀:

“现在又有人叫了。”

敌人彻底忘记自己原本是来挑战的,只沉默地把披风往脸前拉高一点,仿佛第一次后悔今天为什么要报这么多名号。

在这个表情上精准切黑。

喜剧节奏结构

整段保持清楚的“三拍递进”:

第一拍
敌人以史诗级高手姿态自报家门,气势拉满。

第二拍

师姐完全不接他的名号体系,只追问最普通的“你叫什么”。

第三拍

师父一句“王二狗”直接让整个人设掉马,后续所有挽尊都变成新的笑点。

表演原则

敌方剑客

不能主动搞笑

不能表演成丑角
必须认真维持自己的威严、名号和气场

喜剧效果只能来自他越认真维护身份,越被现实拆穿
剑仙师姐

必须始终平静
不嘲讽

不抢包袱

她的可笑点不在“会玩梗”,而在她真的只在认真确认名字

小师妹
必须自然反应
重点是“努力忍笑”
不能提前夸张崩坏
年长师父

必须绝对一本正经
像是在正常认人、正常回忆往事
补刀不能像故意捧哏,而要像无意识地再补一层事实
硬性要求
严格总时长 15 秒

16:9 横屏
三个连续清晰镜头
针对 Seedance 2.0 Mini 保持简单清楚的场面调度
明确视线关系

严格人物连续性
原生同步普通话对白
精准口型
清楚的喜剧停顿

人物服装全程稳定

长剑全程稳定

配角位置全程稳定
地理空间全程稳定
真丝衣料和头发物理效果真实
前景、中景、远景保持自然视差与真实空间环境声
不生成字幕
Negative
blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent character, changing clothes, face morphing, background shift, glitching cuts, disappearing props
Browse by model, style or topic

Find your next direction

ASTORIE CANVAS

Take the idea further.

Bring your prompts, references and results together in one creative space.

Open your canvas ↗