Cinematic Wuxia holding breath challenge

Section 1 | 0–10s

[General Control Principles]

Cinematic realistic texture, pure ancient Chinese Xianxia aesthetics, restrained deadpan comedy performance, observational cinematography, realistic micro-expressions, body weight, silk material, delicate film grain, volumetric air depth, and realistic 3D camera displacement.

The entire world lives continuously from the first frame to the last, but must strictly execute three-track authority:

Protagonist Plot Track = 100% screenwriting authority.
Environmental Life Track = Continuous movement authority, 0% screenwriting authority.
Background Character Track = Independent life authority, 0% protagonist interaction authority.
The environment must not create, trigger, explain, interrupt, solve, change, emphasize, or sync with any character plot. Wind, clouds, mist, light, reflections, buildings, distant characters, vegetation, and water bodies must not produce special reactions due to dialogue, eye contact, comedic punchlines, or protagonist actions. Camera movement can only be driven by protagonist plot and character blocking.
However, all non-protagonist areas must never be frozen.

Before formal generation, interpret all current reference images as the world's DNA, rather than frozen pixels or static background plates. Comprehensively understand the landscape logic, architectural vocabulary, scale relationships, materials, weather, cloud systems, vegetation, reflective surfaces, traffic flow lines, primary light direction, foreground-midground-background relationships, and off-screen space extension, then reconstruct it as a truly connected 3D location. Retain the identity and logic of the reference world, but allow for the replanning of spatial arrangements and camera entry directions, without mechanically replicating the original 2D composition of any single reference image.

[Characters]

Character A | Senior Sword Immortal Sister
East Asian female, 25–30 years old, oval face, fair natural skin tone, dark almond eyes, long black hair partially tied up and secured with a white jade hairpin, tall and slender, wearing white embroidered silk Hanfu, semi-transparent layered wide sleeves, silver waist seal, jade pendant, white cloth boots.

Character B | Junior Sister
East Asian female, 20–25 years old, rounded and lively face shape, black hair braided, petite stature, wearing green linen Hanfu, dark belt, wooden hairpin, black cloth shoes.

[Living World Continuous Operation Rules]
Throughout the 10 seconds, within every approximately 2-second time window, multiple non-protagonist motion sources from different spatial layers must be visible, but these movements cannot start simultaneously, nor move synchronously at the same speed.

Extreme Foreground/Background: A whole layer of massive clouds or air layers matching the current world, which has already begun migrating at a slow, steady speed before the opening.
Deep Mid-ground: Arrange 4–6 very small environmental characters living independently: someone walking continuously along a distant path, someone going up or down stairs, someone stopping to adjust sleeves or belongings before continuing, and two others naturally passing each other. All are unrelated to the two protagonists, do not look at them, do not stop for dialogue, do not stop because the protagonists stop, and do not synchronize actions for comedic effect. No visible background character can be frozen for the entire segment.

Mid-ground Air Layer: Thin mist, low clouds, or air volumes reasonably existing in the current environment continuously move around real buildings and terrain. Mist must be occluded by physical structures, becoming temporarily invisible after entering behind a building and reappearing from the other side consistent with spatial relationships, never penetrating physical buildings.

Foreground Layer: When the camera moves realistically, let a foreground element matching the current world—mist layers, plants, banners, building edges, or other reasonable objects—briefly pass near the lens, reinforcing that the camera is realistically located inside the space.

If water, wet stone, metal, jade, or other reflective surfaces exist in the scene, reflections must change continuously with the camera position and viewing angle, not remain fixed on the background.

All environmental movements share the same weather system and uniform wind direction.

[0–5s | Wide Shot or Long Shot]

The camera starts from within the reconstructed 3D world with a realistic forward, slightly lateral tracking shot. Foreground space or air layers briefly pass the lens, and mid-ground buildings create clear, realistic parallax relative to the extreme background.

Two people walk forward side-by-side normally.

The same Senior Sword Immortal Sister suddenly says very seriously:

"Today we practice composure."

The Junior Sister turns her head slightly to look at her.

The Senior Sister continues:

"Whoever laughs first, loses."

The Junior Sister immediately retracts all expressions.

Both stop simultaneously, then turn to face each other.

All of this must stem from the Senior Sister's initiative to practice composure and must not be triggered by any background changes.

Meanwhile, the Environmental Life Track runs independently: distant clouds continue migrating in the original direction, a distant character continues crossing a distant platform laterally, another continues moving along a road or stairs, mid-ground mist maintains its original speed, and a stable wind field continuously affects vegetation and hanging objects. The background must not suddenly start when dialogue begins, and the background must never stop when the two people stop.

[5–10s | Mid-shot Duo / Cowboy Shot]

The two face each other, about an arm-and-a-half apart.

No spells, no swords drawn.

Both strive to maintain complete lack of expression.

To break the Senior Sister's composure, the Junior Sister makes a tiny movement: one cheek puffs out very slightly for about half a second, then immediately returns to normal.

The Senior Sister almost reacts but forces herself to stay calm, only raising one eyebrow very slightly.

The corner of the Junior Sister's mouth almost twitches upward, but she immediately suppresses it.

A nearly imperceptible tremor then appears on the Senior Sister's lips.

Neither can actually laugh out loud.

The camera performs a slow, realistic small arc movement of about 15–20 degrees around the two. The camera must actually change physical position; digital zoom simulation of orbiting is forbidden. The background's near, middle, and far layers thus produce realistic parallax at different speeds.

Comedy must only come from the two's micro-expressions and restraint.

Meanwhile, the world behind them continues to live: a distant character continues walking and is naturally obscured by a real building, another background character naturally appears from behind another spatial obstruction and continues their route; air volumes continue moving between buildings; as the camera moves in an arc, the mid-ground structures and distant space maintain different parallax speeds. All background movements are out of sync with the protagonists' rhythm.

[Segment 1 End Continuation State | 9.5–10.0s]

The two protagonists' bodies gradually stabilize to facilitate the next segment's connection, but only stabilize the characters, not freeze the world.

The Senior Sister and Junior Sister still face each other, both clearly in a state of nearly being unable to hold back laughter but still forcing a serious face. Character centers of gravity, eye lines, and expressions are clear; the camera gradually stabilizes.

At this time:
Extreme distant clouds are still migrating.
At least one distant character is still in transit.
Mid-ground air volumes are still in motion.
At least one reflection or lighting state is still evolving.

This "characters stable, world still moving" state serves as the anchor point for the Segment 2 time connection.

[Segment 1 Specs]

16:9 widescreen, native synchronized Mandarin dialogue, restrained music, realistic footsteps, fabric sounds, distant footsteps, wind, and spatial environmental sounds. The screen features two main female characters, while allowing for very small, completely independent distant environmental characters. No subtitles, no modern elements.
Negative (Segment 1 Independent):
blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, subtitles, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent protagonist identity, changing clothes, face morphing, hairstyle change; static background plate, frozen scenery, frozen background people, background people standing motionless for the whole clip, living people rendered as landscape texture, duplicated background extras, background extras staring at protagonists, background extras synchronizing with protagonists, background extras reacting to dialogue, background extras reacting to comedy; flat 2D reference image, animated wallpaper, camera sliding over a photograph, fake digital zoom, fake parallax, foreground midground and background moving at identical speed, no occlusion, no object permanence, fixed cloud texture, static cloud sea, frozen mist, static reflections, painted reflections, static distant people, lifeless architecture surroundings, environment stopping during dialogue, environment stopping during close-up, environment freezing when protagonists stop, all environmental movement beginning at the same moment, all environmental movement synchronized to plot beats, dramatic wind caused by dialogue, cloud change caused by laughter, lighting change used as punchline, environment creating or solving story, camera following weather instead of protagonists, random new landmarks, impossible geometry, background replacement, teleporting extras, modern elements, glitching cuts

Segment 2 | 10–20s Independent Extended Video
[Continuation Principles]

If the current workflow allows for video continuation or video reference, prioritize using the complete Segment 1 video itself as the temporal motion reference, while using the final frame of Segment 1 as the opening visual state for Segment 2.

One must not simply re-generate a "similar-looking" location.

The same world must be allowed to continue running forward along the time established in Segment 1.

Complete inheritance of:
The same Senior Sword Immortal Sister
The same Junior Sister
Identical face, hairstyle, body proportions, clothing
Accurate positioning
The laughter-holding expression from the end of the previous segment
Eye line
Camera height
Camera axis
Lens perspective
Simultaneously, the motion phase of the living world itself must be inherited:
Background characters walking at the end of Segment 1 continue from their actual position and direction at that time, not returning to the start.
Clouds maintain the previous direction and speed.
Mist continues flowing from the actual spatial state at the end of the previous segment.
Reflections continue evolving.
The entire environment must never reset to the Segment 1 opening state.
Three-track authority continues:

Protagonist Plot Track = 100% screenwriting authority.
Environmental Life Track = Continuous physical movement authority, 0% screenwriting authority.
Background Character Track = Independent life authority, 0% protagonist interaction authority.

The background cannot cause the two to laugh, the environment cannot determine the winner, and distant characters cannot help complete the punchline.

[10–15s | Duo Close-up / Restrained Arc Movement]

Start directly from the accurate expressions of the two holding back laughter at the end of Segment 1.
Both continue trying not to laugh.
The Junior Sister changes strategy. She suddenly adjusts her posture to be extremely dignified, then very seriously mimics the Senior Sister's usual cold, calm, and solemn expression.

The Senior Sister immediately sees she is being mimicked.

A very slight change occurs in her nostrils.
She holds it in.
After seeing this tiny reaction, the Junior Sister's own lips begin to tremble more noticeably.

The Senior Sister then sees the Junior Sister is desperately holding back laughter, so the two begin to generate new feedback to "the way the other looks while trying not to laugh."

Escalate only through tiny micro-movements:
A slight eyebrow raise
A suppressed breath
A lower lip quiver
A nearly invisible shoulder vibration

Exaggerated funny faces or large comedic movements are forbidden.

At approximately 14.5 seconds, the two finally break at the exact same moment, simultaneously letting out a short, real laugh.

This laugh must come 100% from the mutual feedback of their expressions.

Meanwhile, the Segment 1 world continues running at its original pace: background characters continue their routes and are occluded by buildings; another person passes through a different route in deeper space; a third background character completes a tiny life action unrelated to the protagonists and continues forward. Distant clouds continue migrating, mist continues flowing, and reflections continue changing with the slight camera angle shift.

The moment they laugh, sudden wind, sudden light, background characters turning heads, cloud acceleration, or any synchronized environmental reaction is forbidden.

[15–20s | Mid-Long Shot Conclusion]

After laughing briefly, the two immediately regain their seriousness.
The Junior Sister asks:
"A draw?"
The Senior Sister thinks for a beat:
"Again."
The Junior Sister nods seriously.

The two solemnly return their faces to an extremely formal state, then simultaneously turn back forward and continue walking forward side-by-side normally.

After taking two steps, the Junior Sister casually glances sideways at the Senior Sister.

Unexpectedly, the same Senior Sister is already looking at her sideways.

Their eyes happen to meet.

The corners of both their mouths show the slightest tendency to laugh again, but neither speaks.

Do not add a third punchline.

Maintain an observational aftertaste at the end.

Because the two start walking again, the camera transitions to a gentle three-quarter side tracking shot. The reason for camera movement can only come from character walking, but physical spatial displacement must actually occur.

A foreground element matching the reference world naturally sweeps past one side of the frame; the two protagonists are in the mid-ground; at least two autonomous background characters remain active in deeper space. One passes through a misty area, partially obscured by the air layer before continuing; another moves at a distinctly different screen speed at a different depth.

The relative displacement of massive distant landmarks is very slow. Clouds and mist completely inherit the motion direction from Segment 1 and do not restart. If reflective surfaces exist, highlights slide continuously with the lateral camera position change.

After the protagonists' final eye-contact laugh ends, the background continues to move.

[Final 0.5 Seconds]

The two protagonists continue walking forward.

The audience can still clearly see multiple independent non-protagonist motion sources.

The joke has ended, but the world has not.

[Segment 2 Specs]

16:9 widescreen, native synchronized Mandarin dialogue and real short laughter, precise lip-sync, seamlessly continuous with the Segment 1 temporal state. Realistic camera parallax, autonomously moving distant background characters, continuous air movement, physically plausible dynamic reflections, and environmental spatial sound. The screen mainly features the two female characters, while allowing for very small distant environmental characters. No subtitles, no modern elements.

Negative (Segment 2 Independent):
blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, subtitles, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent protagonist identity, changing clothes, face morphing, hairstyle change; background reset between clips, background motion restarting from zero, background extras returning to starting positions, frozen background extras, static distant people, living people rendered as scenery texture, duplicated extras, disappearing extras without occlusion, background extras watching protagonists, background extras reacting to laughter, background extras laughing with protagonists, synchronized background choreography; flat background plate, static reference image, animated wallpaper, fixed cloud sea, frozen cloud structure, frozen mist, static reflections, painted reflections, fake parallax, digital zoom instead of camera translation, foreground midground background moving at identical speed, environment freezing in close-up, environment freezing during dialogue, environment freezing when protagonists laugh, environment freezing after punchline, environmental event causing laughter, wind gust synchronized with laugh, cloud burst synchronized with joke, sunlight burst at punchline, mist revealing something at story beat, background solving the contest, camera following environmental motion instead of protagonists, new random architecture, scenery replacement, impossible geometry, extra foreground protagonists, modern elements, glitching cuts"}

第1段｜0–10s

【总控原则】

电影级写实质感，纯古风中国仙侠美学，克制冷面喜剧表演，观察式电影摄影，真实微表情、身体重量、真丝材质、细腻胶片颗粒、体积空气纵深与真实三维摄影机位移。

整个世界从第一帧到最后一帧持续生活，但必须严格执行三轨权限：

主角剧情轨 = 100% 编剧权。
环境生命轨 = 持续运动权，0% 编剧权。
背景人物轨 = 独立生活权，0% 主角互动权。
环境不得制造、触发、解释、打断、解决、改变、强调或卡点任何人物剧情。风、云、雾、光线、反射、建筑、远景人物、植被和水体，不得因为对白、眼神、喜剧包袱或主角动作产生特殊反应。摄影机运动只能由主角剧情与人物调度驱动。
但所有非主角区域绝不能冻结。

正式生成前，将当前全部参考图理解为世界 DNA，而不是被冻结的像素或静态背景板。综合理解其中的地貌逻辑、建筑语汇、尺度关系、材质、天气、云层体系、植被、反光表面、通行动线、主光方向、前景—中景—远景关系和画外空间延伸，然后重新构建为一个真实连通的三维地点。保留参考世界身份与逻辑，但允许重新规划空间排列和摄影机进入方向，不机械复制任何单张参考图原有二维构图。

【角色】

角色 A｜剑仙师姐
25–30 岁东亚女性，椭圆脸，白皙自然肤色，深色杏眼，黑色长发半挽，以白玉簪固定，高挑纤细，穿白色刺绣真丝汉服、半透明分层宽袖、银色腰封、玉佩、白色布靴。

角色 B｜小师妹
20–25 岁东亚女性，圆润灵动脸型，黑发编辫，身形娇小，穿青绿色亚麻汉服、深色腰带、木簪、黑色布鞋。

【活世界持续运行规则】
整个 10 秒内，每约 2 秒的时间窗口，都必须能看到多个来自不同空间层级的非主角运动源，但这些运动不能同时开始，也不能以相同速度同步移动。

极远景： 一整层符合当前世界的巨大云层或空气层，在开场之前就已开始以缓慢稳定速度持续迁移。
深中景： 安排 4–6 名尺寸很小的环境人物独立生活：有人沿远处真实通道连续行走，有人沿台阶上行或下行，有人停下整理衣袖或携带物后继续前进，另外两人可自然擦肩而过。所有人都与两位主角无关，不看主角，不因对白停步，不因主角停止而停止，不为笑点同步动作。任何可见背景人物都不能整段冻结。

中景空气层： 薄雾、低云或当前环境中合理存在的空气体积持续绕过真实建筑与地形运动。雾气必须被实体结构遮挡，进入建筑后方后暂时不可见，再从符合空间关系的另一侧重新出现，绝不能穿透实体建筑。

近景层： 摄影机真实移动时，让一个符合当前世界的近景元素——雾层、植物、布幡、建筑边缘或其他合理物体——短暂经过镜头附近，强化摄影机真实位于空间内部。

若场景存在水、湿润石材、金属、玉石或其他反光表面，反射必须随着摄影机位置和观看角度连续变化，不能固定在背景上。

所有环境运动共享同一天气体系与统一风向。

【0–5s｜全景或远景】

摄影机从重新构建后的三维世界内部开始一次真实的向前、略带横向的推轨。近景空间或空气层短暂经过镜头，中景建筑相对于极远背景产生清楚真实视差。

两个人并肩向前走。

同一个剑仙师姐突然非常认真地说道：

“今日练定心。”

小师妹略微转头看她。

剑仙师姐继续：

“谁先笑，谁输。”

小师妹立刻收起所有表情。

两个人同时停下，随后转身面对彼此。

这一切只能来自剑仙师姐主动提出练习定力，绝不能由任何背景变化触发。

与此同时，环境生命轨独立运行：远处云层继续原方向迁移，一名远景人物继续横向穿过远处平台，另一人继续沿道路或台阶移动，中景雾气维持原有速度，稳定风场持续影响植被与悬挂物。对白开始时背景不能突然启动，两个人停下时背景也绝不能跟着停止。

【5–10s｜中景双人 / 牛仔景】

两个人面对面，相距约一个半手臂。

没有法术，不拔剑。

双方努力保持完全没有表情。

小师妹为了破坏师姐定力，只做一个极小动作：一侧脸颊非常轻微地鼓起约半秒，随后立刻恢复正常。

剑仙师姐差一点产生反应，但强行保持镇定，只极小幅度挑高一侧眉毛。

小师妹嘴角差一点上扬，立刻压下。

剑仙师姐的嘴唇随后出现极难察觉的轻微颤动。

两个人都不能真正笑出来。

摄影机围绕两人进行约 15–20 度缓慢、真实的小弧形移动。摄影机必须真正改变物理位置，禁止用数字变焦模拟环绕。背景近中远层因此产生不同速度的真实视差。

喜剧只能来自两个人的微表情与克制。

与此同时，后方世界继续生活：一名远景人物继续行走并被真实建筑自然遮住，另一名背景人物从另一处空间遮挡后自然出现并继续自己的路线；空气体积继续在建筑之间运动；摄影机弧形移动时，中景结构和远景空间保持不同视差速度。所有背景运动都与主角节奏不同步。

【第1段结尾续接状态｜9.5–10.0s】

两名主角身体逐渐稳定，方便下一段续接，但只能稳定人物，不能冻结世界。

剑仙师姐与小师妹仍面对彼此，都处于明显快要憋不住笑、但仍强行保持严肃的状态。人物重心、视线与表情清楚，摄影机逐渐稳定。

此时：

极远云层仍持续迁移
至少一名远景人物仍在行走途中
中景空气体积仍在运动
至少一个反光或光影状态仍在持续演化

这个**“人物稳定、世界仍然运动”**的状态，作为第2段时间续接锚点。

【第1段规格】

16:9 横屏，原生同步普通话对白，配乐克制，真实脚步声、衣料声、远景脚步、风声与空间环境声。画面有两名主要女性角色，同时允许尺寸很小、完全独立生活的远景环境人物。不生成字幕，不出现现代元素。
Negative（第1段独立）：
blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, subtitles, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent protagonist identity, changing clothes, face morphing, hairstyle change; static background plate, frozen scenery, frozen background people, background people standing motionless for the whole clip, living people rendered as landscape texture, duplicated background extras, background extras staring at protagonists, background extras synchronizing with protagonists, background extras reacting to dialogue, background extras reacting to comedy; flat 2D reference image, animated wallpaper, camera sliding over a photograph, fake digital zoom, fake parallax, foreground midground and background moving at identical speed, no occlusion, no object permanence, fixed cloud texture, static cloud sea, frozen mist, static reflections, painted reflections, static distant people, lifeless architecture surroundings, environment stopping during dialogue, environment stopping during close-up, environment freezing when protagonists stop, all environmental movement beginning at the same moment, all environmental movement synchronized to plot beats, dramatic wind caused by dialogue, cloud change caused by laughter, lighting change used as punchline, environment creating or solving story, camera following weather instead of protagonists, random new landmarks, impossible geometry, background replacement, teleporting extras, modern elements, glitching cuts
第2段｜10–20s 独立扩展视频
【续接总则】

若当前 Seedance 工作流允许视频续接或视频参考，优先使用第1段完整视频本身作为时间运动参考，同时使用第1段最终一帧作为第2段开场视觉状态。

绝不能只重新生成一个“看起来相似”的地点。

必须让同一个世界沿着第1段已经建立的时间继续向前运行。

完整继承：
同一个剑仙师姐
同一个小师妹
完全一致的面部、发型、身体比例、服装
准确站位
上一段结尾的憋笑表情
视线
摄影机高度
摄影机轴线
镜头透视
同时必须继承活世界本身的运动相位：
第1段结尾正在行走的背景人物，从当时实际位置和方向继续走，不返回起点
云层维持之前的方向与速度继续迁移
雾从上一段结束时的实际空间状态继续流动
反光继续演化
整个环境绝不能重新恢复成第1段开场状态
三轨权限继续保持：

主角剧情轨 = 100% 编剧权。
环境生命轨 = 持续物理运动权，0% 编剧权。
背景人物轨 = 独立生活权，0% 主角互动权。

背景不能导致两人发笑，环境不能决定输赢，远景人物不能帮助完成包袱。
【10–15s｜双人特写 / 克制弧形运动】
直接从第1段结尾两个人正在憋笑的准确表情开始。
双方继续努力不笑。
小师妹改变策略。她突然把站姿调整得极其端庄，然后非常认真地模仿剑仙师姐平时那种高冷、平静、一本正经的表情。

剑仙师姐立即看出她在模仿自己。

鼻翼出现一次极轻微变化。
她忍住。
小师妹看到这一细小反应以后，自己的嘴唇开始更加明显地轻颤。

剑仙师姐又看到小师妹正在拼命忍笑，于是两个人开始对“对方努力不笑的样子”产生新的反馈。

只通过极小微动作升级：

一次轻微挑眉
一次压住呼吸
一次下唇轻颤
一次几乎看不见的肩膀震动

禁止夸张扮丑，禁止大幅喜剧动作。

大约 14.5 秒，两个人终于在完全相同的瞬间破功，同时发出一声短促、真实的笑。

这个笑必须百分之百来自两个人相互反馈的表情。

与此同时，第1段世界继续按原有时间运行：正在走的背景人物继续自己的路线并被建筑遮挡；另一名人物在更深空间沿不同路线经过；第三名背景人物完成一个与主角无关的微小生活动作后继续前进。远景云层持续迁移，雾继续流动，反光随着摄影机角度微变而继续变化。

两个人笑出来的瞬间，禁止突然起风、突然亮光、背景人物转头、云雾加速或任何同步环境反应。

【15–20s｜中远景收尾】

短暂笑完以后，两个人马上重新恢复严肃。
小师妹问：
“平局？”
剑仙师姐思考半拍：

“重来。”

小师妹认真点头。

两个人一本正经地把脸重新恢复成极其正式的状态，随后同时转回前方，继续并肩正常向前走。

走出两步以后，小师妹非常随意地侧眼偷看师姐。

没想到同一个剑仙师姐此刻已经在侧眼看她。

两人视线恰好撞上。

双方嘴角再次出现极细微的要笑趋势，但谁都不再说话。

不要追加第三个包袱。

结尾保持观察式余味。

摄影机因为两个人重新开始走路，转换为柔和的四分之三侧向跟拍。摄影机运动原因只能来自人物行走，但必须真实发生空间位移。

一个符合参考世界的近景元素自然从画面一侧掠过；两名主角位于中景；更深空间中至少仍有两名自主背景人物活动。其中一人穿过薄雾区域，被空气层部分遮住后继续前进；另一人在不同深度以明显不同的画面速度移动。

极远巨大景观相对位移非常缓慢。云与雾完全继承第1段运动方向，不重新开始。若存在反光表面，亮部随着摄影机横向位置变化持续滑动。

主角最后一次眼神笑点结束以后，背景仍继续运动。

【最终0.5秒】

两名主角仍然向前走。

观众仍能清楚看到多个彼此独立的非主角运动源。

笑点结束了，但世界没有结束。

【第2段规格】

16:9 横屏，原生同步普通话对白与真实短笑，精准口型，与第1段时间状态无缝连续。真实摄影机视差，自主运动的远景背景人物，持续空气运动，物理合理的动态反光和环境空间声。画面以两名主要女性角色为主，同时允许尺寸很小的远景环境人物存在。不生成字幕，不出现现代元素。

Negative（第2段独立）：
blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, subtitles, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent protagonist identity, changing clothes, face morphing, hairstyle change; background reset between clips, background motion restarting from zero, background extras returning to starting positions, frozen background extras, static distant people, living people rendered as scenery texture, duplicated extras, disappearing extras without occlusion, background extras watching protagonists, background extras reacting to laughter, background extras laughing with protagonists, synchronized background choreography; flat background plate, static reference image, animated wallpaper, fixed cloud sea, frozen cloud structure, frozen mist, static reflections, painted reflections, fake parallax, digital zoom instead of camera translation, foreground midground background moving at identical speed, environment freezing in close-up, environment freezing during dialogue, environment freezing when protagonists laugh, environment freezing after punchline, environmental event causing laughter, wind gust synchronized with laugh, cloud burst synchronized with joke, sunlight burst at punchline, mist revealing something at story beat, background solving the contest, camera following environmental motion instead of protagonists, new random architecture, scenery replacement, impossible geometry, extra foreground protagonists, modern elements, glitching cuts