Cinematic Wuxia Suspense Scene Prompt

核心风格
电影级写实质感，采用参考图驱动的活世界设计。整体从古典武侠悬疑自然过渡到克制情境喜剧，强调宏大中国山水尺度、可触摸的天气、自然体积空气、细腻胶片颗粒，以及由环境自身因果关系推动剧情。
环境规划

正式构图前，先执行一轮无声环境规划。
全部参考图共同决定：

地形逻辑
建筑语言

空间尺度

材质年代

植被生态
水系走向

天气状态
云层结构
雾气规律
主光方向
整体色彩
反光表面
空气纵深
将这些信息重组为一个全新、统一、物理自洽、此前未出现过的景观。
环境必须持续活着：
瀑布和水渠始终流动
山雾顺地势游走
植被服从统一风向
云层阴影在湿石上缓慢移动
倒影持续变化

远处建筑在雾中时隐时现
前中后景始终具备真实视差

角色锁定
角色 A
同一个来自 @图片1 的 25–30 岁东亚剑仙女子。
同一张脸，黑色长发，白色刺绣真丝汉服，玉石饰品，纤细身材比例，白色布靴，银色长剑。
角色 B
同一个来自 @图片2 的 20–25 岁东亚小师妹。
同一张脸，黑色编发，青绿色汉服，娇小身材比例，深色布鞋，深色长剑。
硬约束：
身份稳定
面部稳定
服装稳定

长剑连续性稳定
全程只出现两名可见角色

分镜

镜头 1｜0–5s
全景或远景。
两人从鞋到头完整站在一条雨后湿润的悬崖古道上，下方是层叠梯田与轰鸣峡谷。

画面内持续存在：

瀑布灌入山间水渠
竹林随风以不同幅度弯曲
山雾沿低洼地势流动
湿润石面反射移动云层的光
远处瓦顶建筑在雾中若隐若现
峡谷深处突然传来巨大而规律的“咚——咚——”声，像军阵战鼓。
两人瞬间同时转头锁定声音方向。
镜头 2｜5–10s

中景或牛仔景。
保持同样两名女子、同样地理结构、同样服装和道具。
两人谨慎沿石路向下移动。
每一次“咚”声都引发真实环境反馈：

路边积水产生震纹

落叶轻跳
悬挂木牌轻颤
桥索细微共振
摄影机横向跟拍，并从近景竹叶之间穿过，用自然视差揭示新的空间纵深，同时保持世界布局不变。
对白：

白衣剑仙低声说：“有人擂鼓。”
小师妹拔出长剑，屏息倾听

镜头 3｜10–15s

特写或极近特写揭示。
摄影机穿过最后一层山雾，揭露真正声源：
同一条瀑布推动一座巨大的古代木制水轮。
水轮通过木制连杆规律抬起三根巨大的舂米木杵，依次砸入石臼，在梯田旁发出与先前完全相同的震耳“战鼓声”。

现场并无敌人。
对白：

小师妹缓慢放低长剑，愣着说：“师姐…… 他们在舂米。”
白衣剑仙仍极其严肃地观察水轮，平静回答：“粮草如此充足，确实难打。”
停留在两人一本正经的表情上，最后再落下一记巨大的舂米“咚——”作为收尾。
技术规格

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与环境声音
人物身份、地理关系、道具全程稳定

保持真实自然视差
水流、山雾、植被、倒影和水轮机械始终持续运动

木材、石材、真丝衣料和水雾飞溅符合统一物理逻辑
不生成字幕

始终只出现两名可见角色

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

1. Core Style

Masterpiece cinematic realism, reference-driven living-world design, classical wuxia suspense that naturally resolves into restrained situational comedy, monumental Chinese landscape scale, tactile weather, natural volumetric depth, environmental cause-and-effect, subtle film grain, realistic silk, wood, stone, water, fog, and foliage physics. Built around Seedance 2.0’s official image, audio, and video reference workflow with director-level control.

2. Environment Planning Pass

Before composing, perform one silent environment-planning pass.

All supplied reference images jointly define:

terrain logic
architectural language
spatial scale
material age
vegetation ecology
water routes
weather
cloud structure
fog behavior
dominant light direction
overall palette
reflective surfaces
atmospheric depth

Recombine these reference signals into one fresh, physically coherent, previously unseen landscape. Keep it alive and continuously running. The world must never feel like a frozen backdrop.

Environmental motion must remain active throughout:

waterfalls and channels continuously flowing
fog traveling according to terrain
foliage bending with shared wind logic
cloud shadows moving across wet stone
reflections changing with water and light
distant structures appearing and disappearing through mist
strong natural parallax across foreground, midground, and background
3. Character Lock
Subject ID A

The same 25–30-year-old East Asian sword-immortal woman from @Image1.
Exact reference face, long black hair, white embroidered silk hanfu, jade accessories, slender proportions, white cloth boots, silver sword.
Subject ID B

The same 20–25-year-old East Asian junior woman from @Image2.
Exact reference face, braided black hair, teal linen hanfu, petite proportions, dark cloth shoes, dark sword.

Hard lock:
identity stable

face stable

costume stable
sword continuity stable
no extra visible characters
4. Story Setup
Classical wuxia suspense opens the scene. The comedy payoff comes entirely from the environment’s real cause-and-effect. The location itself becomes the dramatic engine.
5. Shot Design

Shot 1｜0–5s

Wide or long shot
Two women stand full-body on a wet cliff path after rain, above terraced fields and a roaring gorge.

Environment details must be active and readable:

waterfalls feeding mountain channels

bamboo bending unevenly in wind

mist following low terrain
wet stone reflecting moving cloud light
distant tiled structures appearing through fog
A huge rhythmic BOOM… BOOM… echoes up from deep in the gorge like war drums.
Both women instantly turn their heads toward the sound source.
Goal of this shot:

establish the living world
establish scale

establish suspense

establish the sound direction
Shot 2｜5–10s
Medium or cowboy shot
Same two women, same geography, same costumes, same props.
They descend cautiously along the same stone path.

Each heavy BOOM causes clear physical response in the environment:

puddles vibrate

loose leaves jump lightly

hanging wooden signs tremble

bridge ropes resonate slightly
The camera tracks laterally through foreground bamboo, revealing new depth through natural parallax while keeping the same world layout intact.
Dialogue:
Woman in white whispers: “有人擂鼓。”

The junior draws her sword and listens

Goal of this shot:

preserve geography
increase tension

deepen the sound mystery

keep the environment physically responsive
Shot 3｜10–15s
Close-up or extreme close-up with reveal
The camera clears the final layer of mist and reveals the true source of the sound:
The same waterfall drives a gigantic ancient wooden waterwheel.
The waterwheel uses wooden linkages to lift and drop three large wooden pestles in sequence, pounding grain beside the terraces.
Each strike produces the exact same thunderous rhythm previously mistaken for war drums.

There is no enemy.

Dialogue:

Junior Sister slowly lowers her sword and says: “师姐…… 他们在舂米。”
The same woman in white studies the relentless mechanism with full seriousness and replies: “粮草如此充足，确实难打。”
Hold on their serious faces while another massive grain-pounding BOOM lands precisely as the ending beat.

Goal of this shot:

reveal environmental truth

keep dead-serious performance
land the restrained situational comedy

end on a strong physical sound beat

6. Output Specs

exactly 15 seconds
16:9 landscape
three coherent shots
native synchronized Mandarin dialogue and environmental sound
stable identities, stable geography, stable props
realistic parallax
continuously moving water, fog, foliage, reflections, and machinery
realistic wood, stone, silk, spray, and mist physics
no subtitles
exactly two visible characters
7. 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