Prompt library
Explore a curated AI video prompt library built from publicly shared examples. Study how strong AI video prompts describe subject, motion, camera, scene, style, and timing before writing your next Seedance, Kling, or Veo prompt.
King Kong Attack Video Prompt for Seedance 2.0
[主体]: @图片1 中的巨型大猩猩, 黑色银背毛发根根分明, 肌肉轮廓极度写实, 双眼充满愤怒, 展现出电影级的五官细节1. [动作]: 巨型动物@图片从深海中缓慢站起, 水流顺着强壮的肌肉缓慢滑落. 它愤怒地发出一声巨响, 双手用力捶打大桥的钢索, 钢缆断裂弹起, 它猛力撕裂桥面, 桥上的汽车被瞬间震飞, 整个动作连贯自然, 绝不僵硬. [场景]: 暴风雨中的海港大桥, 远景是模糊的现代都市高楼. 闪电划破天空, 侧光打在金刚...
Epic Dragon Battle Video Prompt for Seedance 2.0
镜头1（3秒）：一名披着残破斗篷的资深战士骑在巨大的黑曜石鳞片龙背上，身处破碎的嶙峋悬崖边缘。巨龙双翼张开，发出雷鸣般的咆哮。它们从悬崖边向下俯冲，天空中充满了跳动的火星和旋转的飞灰。 镜头2（4秒）：战士紧贴在龙颈上，巨龙在空中进行剧烈的高速螺旋下降。下方是成千上万的中世纪步兵和重骑兵，陷入混乱而血腥的肉搏战中。尘土飞扬，钢铁交错，旌旗摇曳。摄像机采用眩晕的环绕轨道运动，紧跟巨龙的螺旋飞行路径。 镜头3（4秒）：战场地面的特写镜...
Epic Sci-Fi Action Sequence Prompt for Seedance 2.0
第一幕：工业废墟追逐战，15秒，一镜到底。 主体：领头装甲战车车顶屹立着@图2赛博朋克女侠 图片1 ——白色正装西服、机械义臂、金属颈甲、马尾发型、金色耳环，单手高举红色旗枪指向前方峡谷深处，五官细节电影级写实，面部表情全程稳定不变形。 两辆装甲厚重的沙漠越野车在狭窄富铁峡谷中疾驰，峡谷两侧散布着废弃采矿设备与悬挂的巨型起重机挂钩摆锤。 [0-3秒 · 峡谷切入] 镜头从峡谷壁之间切入，紧跟领头战车前方急速后退——车顶女侠屹立不动...
Seedance 2.0 Global Female General Armor Transformation Video Prompt
核心画面提示词 @94d74b42-17ec-45d3-ab60-487c3a1700cb12秒完成10套世界女性铠甲变装，全程不戴眼镜、全套铠甲头盔+武器、纯女性将领造型，转场均带辉光粒子特效，风格统一高级、真实、震撼，各时段画面： 0-1秒： 中国红金明光铠+鎏金战盔，持鎏金长枪，抬手抚肩吞兽，金甲流光、身姿挺拔，东方女战神气场拉满 1-2秒： 日本战国女武者铠甲+鹿角胴丸头盔，持武士刀，转身甩披风扫镜，甲片层叠精致，冷艳凌厉...
Cinematic Duel: Wolverine vs. Hulk
High-budget cinematic action sequence. A battle-worn Wolverine with sharp, reflective metal claws engages in a brutal fight against a massive, muscular Hulk in a misty pine forest. Wolverine has realistic facial featu...
Cinematic Martial Arts Duel: Boar vs. Rooster in Minecraft Style
Subject & Action: A cinematic martial arts duel between a sturdy anthropomorphic wild boar wearing a green tunic and a large brown leather backpack, and a stoic anthropomorphic rooster dressed in a white martial arts...
Seedance 2.0 Video Prompt: Classic Movie Scene Recreation (Two Segments)
第一段： 参考@视频1的人物动作和运镜手法，生成@图片1中黑衣人物在竹林将飞刀掷出的视频，视频中只有黑衣女子一个人物。起始帧的视角景别严格参照@视频1，飞刀掷出以后慢动作对焦飞刀，并虚化黑衣人物，注意刀是刀刃的方向向着镜头飞来，做到1比1还原参考视频，仅人物改为黑衣女子。仅生成打斗音效和环境的音效，不要配背景音乐 第二段： 参考@视频1的飞刀往前飞行的视角和运镜手法，生成一把飞刀向前飞，飞刀视角逐渐变焦显示@图片1中女子在林中练剑...
Seedance 2.0 Video Prompt: First-Person Knight vs. Giant
场景：第一人称视角战场 - 黄昏 / 阴天 单镜头（第一人称视角） 氛围：混乱、幽闭和紧张。泥泞、烟灰和潮湿铁金属的气味充斥着空气，被骑士沉重的头盔狭窄的开口所限制。巨大的巨人怪物的庞大身躯和不可阻挡的力量与被困在沉重盔甲中的人类视角形成鲜明对比。 镜头描述： 从骑士头盔内部持续不断的、无间断的第一人称视角拍摄，通过坚硬的金属面罩狭缝（没有玻璃）框取剧烈的身体互动。摄像机模拟出身体遭受巨大动能打击时的沉重、剧烈的物理效果，被向后抛...
Spring Outing Vlog Prompt for Seedance 2.0
[主体]: 一位年轻女士@【@图1】，面部五官清晰，黑色波浪长发，皮肤质感真实，身穿浅绿色露肩连衣裙，神情喜悦。 [动作]: 她双手轻握一小束油菜花，身体缓慢侧转看向镜头，露出自然甜美的微笑，微风吹动她的发丝和裙摆，动作连贯流畅，不僵硬。 [场景]: 广阔无垠的盛开油菜花田，明亮的黄色花海，背景是纯净的清澈蓝天，色彩对比鲜明。 [光影]: 柔和的侧逆光，金色的丁达尔效应穿过花丛，光影在面部产生自然过渡，营造出电影感的春天氛围。 [...
Dark Knight Transformation: FPV Cinematic Style
【核心主题】 写实科幻特摄，战场觉醒，暗烟法球碎裂，黑铁机甲暗骑士，电影级黑暗神性 【人物基础设定】 100%真实韩国男性面孔，面部有刀伤与淤青，嘴角渗血，眼神中带着极度疲惫与绝望后的燃烧意志。黑色破损战斗服，布料被战火灼焦边缘，沾满泥土与暗红血迹。身形壮硕，单膝跪地喘息，右手紧握一颗散发翻涌暗紫黑烟的"暗域法球"——球体内部旋转着深红与暗紫的能量漩涡，如同被封印的黑洞。 【氛围与画质】 视觉基调：电影质感，IMAX胶片摄影机，P...
Seedance 2.0 Cinematic Suspense Prompt with Detailed Timeline
格式： 15秒 / 一镜到底 / 电影级写实 / 无剪辑、压抑且沉重的悬疑感、战后余烬环境音效设计 场景描述： 战火刚刚熄灭的战场，能见度低，空气中弥漫着呛人的浓烟和未散去的焦味。地面被鲜血和战火浸透，满是破碎的旗帜、折断的兵器和层叠的战殁将士尸骨。一名身穿残破盔甲的将士，独自跪在一片狼藉之中，长枪插在血土中支撑着身体，盔甲上沾满血迹和烟灰，他的身躯仿佛随时会因寒冷和疲惫而倒下。观众应感受到强烈的凄凉感、窒息感以及仿佛被黑暗笼罩的...
Wuxia Combat Short Video Prompt (Seedance 2.0)
港式武俠+好萊塢動作混合，電影感強烈，時長15秒， 主角1 图片1 主角2 图片2 暴雨古代屋頂瓦片，遠處雷電城市天際線，主角1，濕發貼臉，手持雙匕首 [00:00-00:03] 低角度仰拍，主角1從屋簷躍下，慢鏡捕捉雨滴滑過匕首，落地瞬間瓦片碎裂，快速推鏡到中景 [00:03-00:10] 快速跟拍主角1主角2交手，匕首與長劍激烈碰撞，火花四濺，360度環繞鏡頭捕捉高速移動，雨水甩出弧線 [00:10-00:12] 主角1迴旋踢...
Pick the prompt that already matches the pacing, shot style, or scene logic you want. Then rewrite the subject, setting, and tone instead of building from zero.
Paste an example into AI and ask it to keep the camera logic and motion pattern while changing the character, product, environment, or story beat.
Give AI both the example prompt and a short note about what you want changed. In most cases that gets you to a usable draft faster than asking for a new prompt from scratch.
AI video prompts are easier to write when you can see how real prompts handle motion, camera direction, pacing, scene setup, and constraints in practice.
Use this library to find prompt patterns that already work, then rewrite them for your own character, product, story, or visual style instead of starting from a blank page.
The best AI video prompts tell the model what should happen over time, not just what should appear in a single frame. They make the action readable from start to finish.
They also give the model camera intent. Shot distance, movement, framing, and timing usually matter more than stacking abstract style words.
Strong prompts stay focused. Subject, environment, lighting, motion, and mood should all support the same idea so the result feels coherent instead of random.
Finally, good prompts are reusable. When the logic is clear, you can swap the character, product, setting, or tone and still get a useful new draft.
Make the action explicit
Say what changes, when it changes, and what the viewer is meant to follow.
Describe the shot, not just the scene
Useful prompts explain camera distance, movement, angle, and progression.
Keep one visual direction
Motion, lighting, atmosphere, and pacing should reinforce the same output.
Write prompts that can be rewritten
A strong structure is easier to adapt for a new idea than a pile of style keywords.
Common questions about using AI video prompts, studying prompt examples, and adapting them across video models.
An AI video prompt is a structured instruction used to guide a video generation model. A strong AI video prompt defines the subject, action, environment, camera behavior, visual style, and constraints that shape the final clip.
AI video prompt examples make structure easier to see. They show how creators describe motion, timing, shot transitions, and scene setup, which is often harder than describing a single image.
Treat them as starting points. Find the example closest to your goal, keep the structure that controls camera, motion, and pacing, and then rewrite the content for your own idea.
Keep the prompt logic and replace the content. Reuse the sequence, shot behavior, and constraint style, then swap in your own subject, product, setting, mood, or action.
An image prompt mainly describes one frame. An AI video prompt also needs to describe temporal change, movement, camera direction, pacing, and continuity across the clip.
Right now, most examples on this page are Seedance-based. The library structure is built to support more AI video prompt sources and models over time.
Yes. This page currently focuses on AI video prompts first. Over time it can expand into more models, more scene categories, and additional prompt collections such as image prompt references.
