Seedance 2.5 提示词
分享链接 · https://openimages.ajiang.me/p/2098244523503857822/

color_palette · vibrantcomposition · close-upcomposition · dynamic-diagonalhuman_attributes · headwearhuman_attributes · long-hairlighting · neon-light
提示词(原文,逐字保留)
Seedance 2.5 Prompt
TITLE: SELFIE DURING A DEADLY BATTLE
DURATION: 15 seconds
FORMAT: 16:9
STYLE: high-quality Japanese anime film, smooth 2D animation, extremely fluid character motion, clean detailed line art, vivid cel shading, dynamic action animation, colorful cyber-Japanese aesthetic, comedic timing, cinematic lighting.
REFERENCE CHARACTER CONSISTENCY IS EXTREMELY IMPORTANT.
CHARACTER A / FOREGROUND:
Use reference image 1 exactly.
Platinum-white straight bob hair with blunt bangs, pale blue eyes, mischievous sharp smile, rabbit-ear headphones, oversized dark purple Japanese streetwear hoodie, colorful patches and ornaments, white gloves, chunky multicolor sneakers, red-orange Japanese sword.
Preserve her exact hairstyle, face, proportions, headphones, rabbit ears, clothes, accessories, colors and sword design throughout the entire video.
She is cheerful, fearless and obsessed with taking the perfect selfie.
CHARACTER B / BACKGROUND:
Use reference image 2 exactly.
Long straight black hair with blunt bangs, pale blue eyes, colorful rabbit-ear headband, turquoise oversized Japanese-inspired outfit, red decorative armor elements, chunky turquoise-orange boots, twin Japanese swords.
Preserve her exact hairstyle, facial features, outfit, proportions, accessories, colors and swords throughout the entire video.
She is serious and desperately fighting for survival.
LOCATION:
A neon-lit futuristic Shibuya intersection at night.
Japanese signs, giant holographic advertisements, wet asphalt reflecting neon lights, scattered debris, subtle smoke, sparks and damaged cyberpunk buildings.
Anime environment, NOT photorealistic.
Colorful but readable composition.
CORE COMEDY:
Character A happily takes selfies in the foreground while Character B is fighting a gigantic combat robot for her life directly behind her.
Character A NEVER realizes or cares how dangerous the battle is.
The contrast between the cute selfie and the extreme background battle must be immediately understandable.
TIMELINE:
[0.0–3.0 sec]
Start immediately from the POV of Character A's smartphone front-facing camera.
The smartphone itself is NOT visible.
The viewer IS the selfie camera.
Character A fills approximately 45% of the foreground frame.
She leans toward the camera with a huge mischievous smile and makes a cute peace sign beside her face.
She casually adjusts her head angle like she is searching for the perfect selfie.
Behind her, clearly visible in the same frame, Character B is already engaged in an intense battle against a gigantic humanoid combat robot.
The robot suddenly swings an enormous mechanical fist toward Character B.
Character B blocks the massive punch with crossed swords.
IMPACT.
A violent burst of sparks erupts behind Character A.
Strong shockwave.
Hair, rabbit ears, hoodie ornaments and small debris blow toward the camera.
Character A barely reacts.
She simply fixes her bangs and continues smiling into the camera.
Do NOT cut away from the selfie composition.
[3.0–7.5 sec]
Character A changes pose.
She slightly tilts her head, puffs one cheek, then gives another peace sign.
Foreground motion is cute, relaxed and controlled.
At the SAME TIME the background battle becomes dramatically more chaotic.
Character B slides backward across the asphalt.
The giant robot fires a bright energy beam.
Character B rapidly ducks underneath it.
The beam crosses horizontally through the distant background and destroys a neon billboard.
Colorful fragments and sparks rain down.
Character B performs a fast spinning sword counterattack, leaps into the air, lands on the robot's arm and slashes across its armored shoulder.
Clear anime speed lines.
Powerful impact frames.
Dynamic debris.
Smooth readable choreography.
Character A remains perfectly framed for her selfie.
She notices some dust on her shoulder and casually brushes it away without looking behind her.
Comedy comes from the extreme difference in their reactions.
[7.5–10.5 sec]
Character A prepares to take the photo.
She raises two fingers beside her cheek.
Her expression becomes extremely cute and confident.
She happily says:
「はい、チーズ!」
At exactly this moment, the giant robot prepares another massive attack behind her.
Character B braces herself, exhausted but ready to fight.
Suddenly—
The robot hears “cheese.”
It completely stops attacking.
Very short comedic pause.
Character B freezes in confusion.
The robot slowly turns its enormous head toward the selfie camera.
Character A continues smiling, unaware.
[10.5–13.0 sec]
The enormous battle robot unexpectedly crouches down and pushes its head into the background of Character A's selfie composition.
Its intimidating red eyes instantly change into cute friendly expressions.
The robot raises two gigantic mechanical fingers.
PEACE SIGN.
Character A becomes even happier.
She leans slightly toward the robot as if this is completely normal.
Character B stands several meters behind them, covered in dust, holding both swords, completely stunned.
Her expression clearly says:
「え?」
Hold the comedic reaction for a moment.
[13.0–15.0 sec]
Character A presses the virtual shutter button toward camera.
CAMERA FLASH.
Very quick white flash.
Freeze-frame on the final selfie photograph:
Foreground:
Character A smiling beautifully with a peace sign.
Background:
The gigantic robot happily making a peace sign directly beside her.
Farther background:
Character B frozen mid-battle with both swords raised, clothes dusty, eyes wide open, completely speechless and furious.
The final photograph should look hilariously perfect.
Add a tiny camera shutter bounce.
End on the frozen selfie for approximately 1 second.
CAMERA DIRECTION:
Maintain smartphone front-camera / selfie POV for almost the entire video.
Do not randomly switch to third-person shots.
Keep Character A large in the foreground while the action remains clearly readable behind her.
Use subtle handheld smartphone movement caused by battle shockwaves.
Do NOT use excessive camera shake.
Foreground face must remain sharp and recognizable.
ANIMATION PRIORITY:
1. Exact preservation of both reference character designs
2. Clear foreground/background storytelling
3. Readable sword combat
4. Strong physical impacts and debris
5. Clear comedic pause before the robot joins the selfie
6. Perfect final freeze-frame punchline
MOTION:
high frame-rate anime motion,
natural secondary hair motion,
cloth physics,
rabbit-ear movement,
sword follow-through,
robot mechanical inertia,
strong impact anticipation,
clear acceleration and deceleration,
smooth facial acting.
DO NOT:
do not merge the two characters,
do not swap hairstyles,
do not change costumes,
do not change hair color,
do not remove rabbit ears or headphones,
do not duplicate characters,
do not give Character A Character B's swords,
do not make Character B take the selfie,
do not change anime style into photorealism,
do not make the robot attack Character A,
do not obscure Character A's face,
do not place a visible smartphone in front of her face,
do not create extra fingers,
do not deform hands,
do not teleport characters,
do not randomly change location,
do not add unnecessary characters,
do not end before the robot makes the peace sign,
do not skip the final reaction from Character B.
MOST IMPORTANT FINAL BEAT:
The audience must first believe the giant robot is a deadly enemy.
Then the robot abruptly stops fighting ONLY because it wants to appear in the selfie.
The final image must clearly show:
Character A = completely happy,
giant robot = happily posing,
Character B = exhausted, angry and totally confused.
中文译文
Seedance 2.5 提示词
标题: 致命战斗中的自拍
时长: 15 秒
画幅: 16:9
风格: 高品质日本动画电影,流畅的 2D 动画,极其流畅的角色动作,干净精细的线条,鲜明的赛璐璐上色,动态动作动画,色彩丰富的赛博日式美学,喜剧节奏,电影感打光。
参考角色一致性极其重要。
角色 A / 前景:
请严格按照参考图 1 使用。
铂白色直发波波头配齐刘海,浅蓝色眼睛,调皮锐利的笑容,兔耳耳机,超大深紫色日式街头风卫衣,彩色补丁与饰品,白色手套,厚底多彩运动鞋,红橙色日式刀。
整段视频中请严格保留她完全一致的发型、脸型、身材比例、耳机、兔耳、服装、配饰、配色与刀的设计。
她性格开朗、无所畏惧,并且执着于拍出完美的自拍。
角色 B / 背景:
请严格按照参考图 2 使用。
黑色长直发配齐刘海,浅蓝色眼睛,彩色兔耳发箍,青绿色超大日式风格服装,红色装饰性护甲元素,厚底青绿橙配色靴子,双日式刀。
整段视频中请严格保留她完全一致的发型、面部特征、服装、身材比例、配饰、配色与刀。
她严肃认真,正在拼死求活。
场景:
夜晚霓虹闪烁的未来涩谷十字路口。
日式招牌、巨型全息广告、被霓虹反射的湿润柏油路面、散落的碎片、淡淡的烟雾、火花与受损的赛博朋克建筑。
动画环境,非照片写实。
色彩丰富但构图清晰可读。
核心喜剧:
角色 A 在前景开心地自拍,而角色 B 正在她正后方与一台巨型战斗机器人殊死搏斗。
角色 A 从未意识到,也毫不在乎战斗有多危险。
可爱的自拍与极度激烈的背景战斗之间的反差必须一眼就能看懂。
时间线:
[0.0–3.0 秒]
直接从角色 A 智能手机前置摄像头的视角开始。
智能手机本身不可见。
观众就是这颗自拍镜头。
角色 A 占前景画面约 45%。
她带着巨大的调皮笑容朝镜头凑近,并在脸旁比出一个可爱的剪刀手。
她随意地调整头部角度,仿佛在寻找完美的自拍角度。
在她身后,清晰可见于同一画面中,角色 B 已经与一台巨型人形战斗机器人展开激烈战斗。
机器人突然挥出一只巨大的机械拳朝角色 B 打去。
角色 B 用双刀交叉格挡这记重拳。
撞击。
一团猛烈的火花在角色 A 身后爆开。
强烈的冲击波。
头发、兔耳、卫衣饰品以及细小碎片朝镜头方向吹来。
角色 A 几乎毫无反应。
她只是拨了拨刘海,继续对镜头微笑。
请勿切离自拍构图。
[3.0–7.5 秒]
角色 A 变换姿势。
她轻轻歪头,鼓起一边的脸颊,然后再次比出剪刀手。
前景动作可爱、放松且有控制。
与此同时,背景战斗变得更加混乱激烈。
角色 B 在柏油路面上向后滑退。
巨型机器人发射一道明亮的能量光束。
角色 B 迅速俯身躲过。
光束横向穿过远处背景,摧毁一块霓虹广告牌。
彩色碎片与火花如雨般落下。
角色 B 施展快速旋转的刀剑反击,跃向空中,落在机器人手臂上,横斩其装甲肩部。
清晰的动画速度线。
有力的冲击定格帧。
动态碎片。
流畅易读的动作编排。
角色 A 依然保持完美的自拍构图。
她注意到肩膀上有些灰尘,头也不回地随手轻轻掸掉。
笑点来自两人反应的极端反差。
[7.5–10.5 秒]
角色 A 准备按下快门。
她在脸颊旁竖起两根手指。
她的表情变得无比可爱又自信。
她开心地说:
「はい、チーズ!」
就在这一刻,身后的巨型机器人正酝酿又一次大规模攻击。
角色 B 强撑身体,精疲力竭却准备再战。
突然——
机器人听到了“cheese”。
它完全停止攻击。
非常短的喜剧停顿。
角色 B 困惑地僵住。
机器人缓缓将巨大的脑袋转向自拍镜头。
角色 A 继续微笑,毫无察觉。
[10.5–13.0 秒]
巨型战斗机器人出乎意料地蹲下,并把头探入角色 A 自拍构图的背景中。
它那令人生畏的红色双眼瞬间变成可爱友好的表情。
机器人抬起两根巨大的机械手指。
剪刀手。
角色 A 变得更加开心。
她微微朝机器人靠近,仿佛这一切完全正常。
角色 B 站在他们身后数米处,浑身是尘,手握双刀,完全愣住了。
她的表情清楚地表达着:
「え?」
让喜剧反应停留片刻。
[13.0–15.0 sec]
角色 A 朝镜头按下虚拟快门按钮。
闪光灯。
极短的白光闪烁。
定格在最终的自拍照片:
前景:
角色 A 带着美丽的笑容比出剪刀手。
背景:
巨型机器人在她身旁开心地比着剪刀手。
更远的背景:
角色 B 在战斗中被定格,双刀高举,衣服满是灰尘,双眼圆睁,完全哑口无言且怒火中烧。
最终照片应看起来滑稽而完美。
加上一次极轻微的快门反弹。
在定格自拍上停留约 1 秒。
镜头指导:
几乎整段视频都保持智能手机前置/自拍视角。
不要随意切换到第三人称镜头。
保持角色 A 在前景占据较大画幅,同时确保背景动作清晰可读。
使用由战斗冲击波造成的轻微手持手机式移动。
不要使用过度的镜头抖动。
前景面部必须保持清晰可辨。
动画优先级:
1. 严格保留两个参考角色的造型
2. 清晰的前景/背景叙事
3. 易读的刀剑战斗
4. 强烈的物理冲击与碎片
5. 机器人加入自拍前清晰的喜剧停顿
6. 完美的最终定格画面收尾
动作:
高帧率动画动作,
自然的次级头发运动,
布料物理,
兔耳运动,
刀剑随势,
机器人的机械惯性,
强烈冲击预期,
清晰的加速与减速,
流畅的面部表演。
禁止事项:
不要将两个角色合并,
不要互换发型,
不要更换服装,
不要更改发色,
不要移除兔耳或耳机,
不要复制角色,
不要给角色 A 配上角色 B 的刀,
不要让角色 B 拍自拍,
不要将动画风格改为照片写实,
不要让机器人攻击角色 A,
不要遮挡角色 A 的脸,
不要在她脸前放置可见的智能手机,
不要出现多指,
不要让手部变形,
不要瞬移角色,
不要随意更换场景,
不要添加多余角色,
不要在机器人比剪刀手之前结束,
不要跳过角色 B 的最终反应。
最重要的最终节拍:
观众必须首先相信巨型机器人是致命敌人。
随后机器人突然停止战斗,仅仅因为它想出现在自拍照中。
最终画面必须清晰呈现:
角色 A = 完全开心,
巨型机器人 = 开心摆 pose,
角色 B = 疲惫、愤怒且完全困惑。
来源与署名
原文由 @tokyo_Valentine 发布在 X:查看原推文。 本页逐字保留原文并提供机器翻译的中文解读;版权归原作者所有。
同工具的更多提示词
- 角色:@[image1]。视觉风格:暗黑幽默的哥特式故事书插画,与参考图保持一致——风化的石头、苔藓般的绿色、风暴般的灰色天空、绘画般的质感,以及诡异而温暖…@airina_xyz
- AI 陷阱 👇 Seedance 2.5:1 次渲染约 5.55 美元 给动作命名,拗不过模型最常看到的样子。 我写了"单脚障碍滑"。它却用双脚滑过每一个锥筒…@markh_app
- 以图片 #1 作为男人、狗狗以及整体美术方向的唯一视觉参考。保留男人极度瘦高的剪影、苍白狭长的脸庞、凌乱的黑色头发、黑色外套以及笨拙而忧郁的肢体语言。保留那只…@aimikoda
- 制作一个30秒的超写实电影质感音乐视频,三位年轻成年歌手在夜晚霓虹闪烁的城市中演唱一首情感丰富的现代歌曲。视频应呈现高预算专业音乐视频的质感,具备真实的人物、…@ChillaiKalan__
- 角色:@[image1] 视觉风格:异想天开的手绘墨线故事书插画,带有水彩纹理、柔和的青蓝沼泽色调、温暖的琥珀色灯笼光,比例略有夸张。角色:一位身材敦实、疲惫…@airina_xyz
- SEEDANCE 2.5 — 动作捕捉 / 角色替换提示词 创建一个 11 秒、4:3 的动作捕捉视频,要求口型精准同步。 核心目标 仅将 @vid1 中位于…@Korgamiai