真实 AI 影片中的角色转身表

来自 Flick 制作的 AI 影片的 10 个角色转身表,每个都附有生成它的提示词以及它所应用的镜头,这样你就能看到转身表如何发挥作用。

一幅墨水画,画的是同一个男孩以三种姿态站立,分别是正面、侧面和背面,背景为米白色

角色转身表在纯色背景上展示一个角色的正面、侧面和背面,以便艺术家或 AI 模型能够在每个镜头中绘制同一个人物。网上大多数转身表都是为动画绘制或作为模板出售的,你永远看不到有人使用它们后的效果。这 10 张角色转身表来自 2026 年夏季 Flick Residency 的 5 部影片,电影制作者在 187 个镜头的提示词中指定了它们,其中 120 个是视频片段,包括可能不在最终影片中的镜头。每一张都附有逐字逐句的提示词、它所提供的镜头以及指定它的镜头数量,按布局分组。

经典三视图表:正面、侧面和背面

三视图转身表在纯色背景上以相同比例展示正面、侧面和背面的全身,这足以让模型从任何角度保持面部、服装和轮廓。

完美求婚·男劫匪

一张表格,在一场抢劫的 64 个镜头中被指定

一个戴着棕色牛仔帽、红色头巾、皮背心和靴子的男人的三个视图:正面、左侧和背面,浅灰色背景

男劫匪转身表

提示词将头巾固定为"他的关键视觉标志",并要求在摄影棚背景中使用真实演员,而不是渲染图。

Create a professional full-body character reference sheet (turnaround) for the MALE robber — the man wearing the dark brown cowboy hat and dark red bandana in this photo. Show THREE views of the same man side by side in one image: front view, left side profile, and back view. Full body in frame, neutral standing A-pose, plain light gray seamless studio background, even soft studio lighting, orthographic framing. 

Keep his exact identity and wardrobe from the photo IDENTICAL across all three panels: early-30s white American man, tall and lean, long weathered face, sun-reddened skin, a few days of light stubble, deep-set eyes with a detached cold gaze. Worn dark brown cowboy hat brim tilted slightly down, a loose dark red (maroon) bandana at the neck as his key visual signature, slightly yellowed off-white western shirt, old brown leather vest, dark straight-leg jeans, wide leather belt with a metal buckle and an old leather holster holding a revolver, heavily worn dark brown pointed-toe cowboy boots.

CRITICAL: maximum photorealism — this must look like a real photograph of a real actor on a studio backdrop, NOT a CGI render or illustration. Real skin texture with pores and stubble, realistic worn fabric with genuine wear and creasing, natural film-like color and grain, true-to-life proportions. No plastic or over-smoothed skin, no artificial look. Show ONLY this one character, no other people, no scene, no props beyond his own worn gear, no text or labels.
Zhelun “Jerry” Lin 执导的《The Perfect Proposal》
含义:为戴深棕牛仔帽和暗红围巾的男劫匪制作三视图全身写实照片级角色参考表,灰色背景。
A cowboy robber kneels with a ring box before a red-haired woman in a restaurant high above a night city00:12

它所提供的镜头

劫匪跪下嘲笑求婚,这是一个 Seedance 2.0 镜头,其提示词通过指定表格将他放置在左侧。

以当前参考图求婚场景 - 角色替换版与固定构图参考。严格保持人物外形、服装、发型、体型、人物位置、桌椅结构、高级餐厅空间、落地窗、窗外夜景和电影灯光连续一致。

这是一个全程固定机位、不剪切的连续长镜头。

不切镜;
不换机位;
不推拉;
不摇移;
不跟拍;
不变焦;
不使用手持晃动;
不使用慢动作;
全部表演、烟花异常和最终爆炸都在同一个固定构图中发生。

没有音乐,没有背景音乐,没有配乐。

镜头起始状态

男劫匪Male Robber Turnaround位于画面左侧,女劫匪Female Robber, Left Leg Gun Holster Corrected位于画面右侧。

男劫匪面对女劫匪,手中拿着从Jeremy那里抢来的紫色戒指盒和戒指。他缓慢地单膝跪地,将打开的戒指盒举向女劫匪,模仿刚才Jeremy的求婚。

女劫匪站在他面前,看着他和戒指盒。

两个人都带有浓重的德州口音。

这不是认真、真诚或感人的求婚。两名劫匪都把这一切当成一场荒谬的游戏。他们正在互相调情、互相逗弄,享受自己刚刚抢劫成功、完全控制现场的得意感。

男劫匪的表演与台词

男劫匪单膝跪地,举着戒指盒,脸上带着坏笑和戏谑的神情。

他用浓重的德州口音,带着明显的调情和挑逗意味说:

“Will you marry me, Abigail?”

这句台词必须是:

flirtatious;
戏谑的;
带有恶作剧感;
像故意模仿Jeremy;
像是在调戏女劫匪;
松弛、傲慢、得意;
不是真诚告白;
不是郑重求婚;
不要感动;
不要哽咽;
不要表现得特别激动。

他的潜台词更像:

“看,我们现在连这个求婚都可以抢过来玩。”

女劫匪的表演与台词

女劫匪看着跪在地上的男劫匪。她没有被真正感动,也没有把这个求婚当回事。

她带着浓重的德州口音,用一种接住他的玩笑、继续和他调情的方式回应:

“Oh, Jeremy, God help me.”

这句台词必须是:

flirtatious;
戏谑的;
带着坏笑;
有点夸张地模仿浪漫女主角;
像是在说“好吧,我陪你继续玩”;
不是真正的惊喜;
不是真正的感动;
不要哭泣;
不要庄重;
不要爱情电影式的真诚。

她说完后,伸手从紫色戒指盒中拿出戒指,轻松、得意地将戒指戴到自己的手上。

她自己戴戒指的动作要显得:

熟练;
随意;
得意;
带有调情感;
像把抢来的战利品占为己有;
不是神圣的婚约仪式。

两个人仍然在互相看着、互相坏笑,完全没有意识到画面即将崩坏。

第一阶段:虚假的盛大浪漫仪式

就在女劫匪把戒指戴到自己手上的瞬间,窗外的烟花突然不约而同地集体绽放。

大量暖色烟花同时照亮夜空,仿佛有人为这场求婚安排了一场极其盛大的仪式。

烟花颜色以以下暖色为主:

金黄色;
橙金色;
琥珀色;
暖白色。

这一瞬间的感觉应该是:

过分盛大;
过分巧合;
荒谬地浪漫;
像宇宙突然认真配合这两个劫匪的玩笑;
带有黑色喜剧式的假仪式感;
看起来几乎完美,但完美得不自然。

男劫匪和女劫匪仍然没有特别激动。他们只是觉得这一切非常有趣,继续用得意、调情的眼神看着彼此。

第二阶段:AI画面开始失控

烟花绽放后,画面开始出现一种不符合现实物理规律的异常。

最初只是少量暖色火花从窗外飘近。

随后,这些火花没有击碎玻璃,也没有被玻璃阻挡,而是像AI生成画面出现穿模错误一样,直接穿过完整的落地窗,进入餐厅室内。

必须明确表现:

落地窗暂时仍然完整;
火花直接穿透玻璃;
火花和玻璃互相重叠;
没有正常的碰撞关系;
没有先破窗;
像系统忘记玻璃是实体;
像生成模型无法理解室内外空间。

火花进入室内后:

诡异地散落到桌面上;
落在戒指盒旁边;
飘过两名劫匪的身体;
穿过物体边缘;
在不合理的位置复制;
部分火花悬停;
部分火花运动方向错误;
部分火花忽然变大或变多。

这一阶段要让观众清楚感受到:

这不是普通烟花事故,而是AI生成的画面正在失去对空间、碰撞和物理规律的控制。

第三阶段:生成错误迅速蔓延

AI错误开始快速扩散。

暖色烟火数量异常增加,越来越多的烟花和火焰从窗外穿透玻璃进入室内。

画面逐渐出现以下崩坏感:

火花不自然地复制;
烟火数量远远超过合理范围;
室内外空间开始混合;
烟花像贴图一样重叠在人物和家具上;
光焰边界发生错位;
火焰穿透桌子、人物和墙面;
烟花运动轨迹不遵循重力;
部分烟火突然变形;
暖色光线出现过度曝光和错误扩张;
整个场景像无法继续维持稳定生成。

这种崩坏必须越来越严重,像AI系统在试图生成“盛大烟花求婚”,却不断过度生成,最终完全失去控制。

画面的感觉不是数字故障界面,也不是传统电子glitch,不要出现文字、代码或像素噪点。

核心是:

现实画面本身被错误生成的烟花吞噬,物理逻辑和空间结构彻底失效。

第四阶段:AI彻底崩坏式大爆炸

所有暖色烟花突然在落地窗外同时爆发。

金黄色、橙色、暖白色的烟火和火焰像失控洪水一样穿透整面玻璃,猛烈涌入餐厅。

这不是普通、写实、物理准确的爆炸。

它必须像一次AI generation meltdown:

烟花数量无限增加;
爆炸范围不合理地扩张;
火焰同时来自多个错误方向;
室内外彻底失去边界;
人物、桌子和物体边缘发生短暂错位;
所有画面元素仿佛同时被错误的生成力量推开;
场景完整性瞬间崩塌;
画面像无法继续维持人物和环境的一致性;
暖色火光迅速占满整个构图。

最终,失控的烟火将以下内容全部炸飞:

单膝跪地的男劫匪;
站在右侧的女劫匪;
桌子;
戒指盒;
戒指相关道具;
椅子;
画面中所有其他可见物体。

他们不是以正常动作戏方式被炸飞,而是像整个生成画面突然失去稳定性,所有元素同时被错误地掀飞、抛离、吞没。

两人被炸飞时的表演限制

在爆炸真正发生后:

男劫匪不再说任何台词;
女劫匪不再说任何台词;
两个人都不要尖叫;
不要增加滑稽喊叫;
不要互相对视;
不要做准备逃跑的动作;
不要做夸张反应;
不要增加新的情绪表演;
不要追加对白;
不要让他们主动跳起。

他们只是突然被失控的AI烟火连同整个场景一起炸飞。

爆炸本身就是最后的表演。

“YEE-HAW!”结尾音效

当大爆炸已经把两名劫匪和场景中的所有物体炸飞,暖色火光几乎完全吞没整个固定镜头时,加入一次短促、粗犷、兴奋的:

“YEE-HAW!”

这个“YEE-HAW!”不是男劫匪或女劫匪在现场喊出的台词。

它是一个:

非写实音效;
额外的牛仔喜剧声音标点;
像牛仔骑马冲出去时的欢呼;
带有浓重德州西部片气质;
充满荒谬的胜利感;
与画面毁灭形成强烈反差;
用于完成黑色喜剧节拍。

音效时机必须准确:

两个人已经被炸飞;
画面已经被失控火光吞没;
不再有任何角色台词或表演;
随后才突然出现一次 “YEE-HAW!”

“YEE-HAW!”必须:

只出现一次;
短促;
响亮;
粗犷;
欢快;
带明显牛仔骑马式感觉;
不要拖得很长;
不要重复;
不要和角色口型对应;
不要让任何人物重新进入画面喊叫。

这个声音结束后,画面立刻停止、冻结或进入结尾状态。

声音设计

全程没有音乐,没有背景音乐,没有配乐。

只保留:

男劫匪带浓重德州口音的台词:
“Will you marry me, Abigail?”
女劫匪带浓重德州口音的台词:
“Oh, Jeremy, God help me.”
戒指盒被打开和拿稳时的轻微摩擦声;
女劫匪取出戒指的细小声音;
戒指戴到手上的轻微摩擦声;
窗外烟花同时绽放的声音;
火花穿透玻璃时诡异、不符合现实的窜动声;
火花落到桌面附近的细碎声音;
AI画面逐渐失控后越来越密集、越来越混乱的烟火声;
最终吞没整个室内的巨大爆炸声;
爆炸结束节点处单独出现一次:
“YEE-HAW!”

不要加入:

音乐;
浪漫配乐;
西部片配乐;
喜剧配乐;
角色被炸飞时的额外尖叫;
爆炸后的对白;
旁白。
画面风格

整体风格是:

真实电影感;
高级夜间餐厅;
德州亡命情侣气质;
黑色幽默;
戏谑式调情;
荒谬的假求婚;
过度完美的烟花仪式;
AI生成画面逐渐失控;
最终彻底崩坏;
爆炸后以“YEE-HAW!”作为西部喜剧标点。

整个镜头的节奏是:

调情式假求婚 → 过度浪漫的烟花 → 物理规则出现错误 → AI生成失控 → 全场炸飞 → “YEE-HAW!”

严格禁止
不要切镜;
不要换机位;
不要推镜;
不要拉镜;
不要摇移;
不要变焦;
不要使用音乐;
不要把两个人演成真诚恋人;
不要把男劫匪演成认真求婚;
不要把女劫匪演成感动落泪;
不要让台词失去德州口音;
不要让烟花始终只停留在窗外;
不要让玻璃先正常破碎再让火焰进入;
必须先表现烟花直接穿透玻璃的AI穿模错误;
不要把爆炸做成完全写实、物理准确的动作片爆炸;
爆炸必须具有AI画面彻底崩坏、过度生成、空间失效和逻辑崩塌感;
人物被炸飞后不要添加任何对白或表演;
不要让角色自己喊“YEE-HAW!”;
不要让“YEE-HAW!”重复出现;
不要让爆炸结束得过于平静。
镜头核心

男劫匪单膝跪地,拿着紫色戒指盒,用浓重德州口音、带着调戏和玩笑意味说:

“Will you marry me, Abigail?”

女劫匪同样带着浓重德州口音,flirting地回应:

“Oh, Jeremy, God help me.”

她从盒中拿出戒指,自己戴到手上。

窗外暖色烟花集体绽放,像一场荒谬而盛大的求婚仪式。随后,火花开始像AI穿模错误一样直接穿透完整玻璃,散落到桌面和室内。错误迅速扩散,烟火不断复制、膨胀,室内外空间和物理规则彻底崩塌。

最终,暖色烟花从窗外爆发进整个室内,把两名劫匪以及画面中的所有物体全部炸飞。两个人被炸飞时不说话、不尖叫、不追加表演。

当爆炸和暖色火光彻底吞没画面后,突然响起一次粗犷、欢快、非写实的牛仔音效:

“YEE-HAW!”

随后画面立即结束。
Zhelun “Jerry” Lin 执导的《The Perfect Proposal》
含义:男劫匪用德州口音戏谑求婚女劫匪,窗外烟花突然爆发并穿透玻璃涌入室内,最终AI生成失控导致暖色火焰炸毁整个场景,电影质感黑色喜剧风格。

Mila & Sasa · Mila

用意大利语的两行提示词生成的 2D 转面图

蓝色波波头、红脸、黄色套装和蓝色靴子的女孩的平面 2D 转面图,四个视图

Mila 转面图

英文:"为我创建 Mila 的角色设计及其转面图。保持简约设计、眼睛、她的颜色、线条类型和阴影。"

Creami il character design di Mila, con il suo turnaround. Mantieni il design minimalista, gli occhi i suoi colori, il tipo di tratto, e ombre. 
Katrin Ann Orbeta 执导的《Mila & Sasa》
含义:米拉的角色设计转身图,保持极简风格、她的眼睛、配色、线条类型和阴影。
2D cartoon girl with a blue bob asleep under a blanket in a cream sleeping pod on a spaceship00:05

它所提供的镜头

Mila 在她的舱室中睡觉,这是一个 MiniMax H3 镜头,为她的身体命名了这个转面图,并为她闭着的眼睛使用了表情表。

Start Capsula di Mila, Vuota100%.
Mila Mila Character Design Turnaroundcon la tuta azzurra ha gli occhi chiusi Mila - Espressioni del Visodorme sotto le lenzuola che la coprono fino ai piedi.
Non sorride.
Muove leggermente la testa come se stesse sognando.
Alla fine apre lentamente gli occhi, malinconica.
Hand-drawn 2D animation, illustrated storybook, cozy retro-futuristic sci-fi, warm pastel palette, soft sepia linework, vintage children's book, gentle golden light, whimsical, nostalgic, cozy spaceship interior, cinematic parallax, subtle atmospheric animation. Negative Prompt photorealistic, 3D CGI, realistic rendering, glossy surfaces, hyper-detailed textures, anime, manga, thick black outlines, neon colors, high contrast, dark horror mood, dramatic lighting, fast camera movement, excessive motion, shaky camera, realistic modern kitchen, futuristic cyberpunk aesthetic, visual clutter.


Katrin Ann Orbeta 执导的《Mila & Sasa》
含义:米拉穿蓝色制服闭眼躺在被子下做梦,最后忧郁地睁眼,手绘2D动画复古未来温暖柔和风格。

带有面部特写的转身表

在正面、侧面和背面视图旁边的大型头部特写面板为模型提供了特写尺寸的面部,这在对话和反应镜头中最为重要,因为面部会填满画面。

Lifeguard · 成年 Benji

四个身体视图和一个头部特写,在 9 个片段中命名

一个留着金色长发、穿着灰色 T 恤和运动裤的苍白男子的角色表:正面、两个侧面、背面和一个头部特写

成年 Benji 角色表

提示词将衣服固定为两个灰色十六进制代码,并排除了所有道具,因此角色表上的内容不会泄漏到不需要它的镜头中。

Create a professional character reference sheet of this man, Adult Benji (the shaggy long blonde haired man in the source image). Show the SAME character with identical facial features, proportions, and hair in a clean turnaround layout: a full body front view, a left profile view, a right profile view, and a back view, plus one large closeup headshot panel. Neutral A-pose, even flat lighting, orthographic framing, plain seamless pure white background. Dress him in a plain blank solid grey t shirt and grey sweatpants pulled from a cool grey palette (around #545456 and #777775), no logos or patterns. Do NOT include any props, no amulet, no necklace, nothing in his hands. Show only this one character, no other people, no scenery. Keep his pale sedated look, dark under eye circles, patchy stubble, and long blonde hair falling into his eyes consistent across every panel.
Owen Doerr 执导的《Lifeguard》
含义:金发男子本吉的专业角色参考表,展示正面侧面背面全身及特写,灰色服装白色背景。
Close-up of a pale man with long blond hair in a grey t-shirt, looking aside in a dark metal room00:04

它所提供的镜头

一个短暂的特写使用角色表来定义他的特征,而单独的起始帧设置了他的头部位置和视线。

Duration: 4s. Aspect ratio: 16:9. No background music, no score. Ambient sound effects only — low sandstorm rumble outside, faint wind against the chamber.

Benji Facing Joe, Profile Turn is the start frame. Preserve it exactly, especially Benji's head position and eyeline exactly as shown — this is an exact start frame. Adult Benji Character Sheet defines benji's features, use only for consistency.

Camera: locked-off and static, no drift.

Color grade: strict palette — deep teals and cool greys #161E20, #253534, #1B2727, #3F5153, #545456, #777775, warm amber #F2A462 and burnt sienna #BE704A only where the storm-light reaches. No colors outside this palette. Cinematic photorealism, moody low-key lighting, film grain, atmospheric haze.

MOTION (0–4s):
Benji holds the head position and eyeline from @image1 exactly at first, listening — a quiet, contained annoyance on his face, brows tightening slightly, jaw a little tense, but restrained, not overt. He listens like this for a beat. Then his head turns forward, toward camera, the annoyance still sitting faintly on his face as the turn completes.

Camera stays locked throughout — all movement is in Benji's head turn and the small shift of his expression.
Owen Doerr 执导的《Lifeguard》
含义:班吉从侧脸转向正面镜头,表情微微紧绷显露克制的不悦,冷色调写实风格配沙暴环境音。

Lifeguard · 救生员

仅做了两处更改就重新生成的角色表

一位深色卷发女性穿着赤陶色泳装手持救生圈的角色表:四个视图和一个头部特写

救生员角色参考表

泳装颜色设置为十六进制代码 BE704A,重新生成提示词列出了恰好两处改动,因此她的其余部分保持完全一致。

Regenerate this exact same character reference sheet keeping the woman's identity, face, dark curly hair, olive-bronze Mesopotamian features, toned physique, and her warm terracotta hex BE704A one piece lifeguard swimsuit, and the small greyish antiqued silver Pazuzu amulet on a thin cord around her neck all identical. TWO changes only: 1) Remove ALL text and the white cross graphic from the swimsuit so it is a clean plain solid terracotta suit with no lettering or logos on any view. 2) Have her HOLD a classic circular lifeguard ring buoy float, the round life preserver ring, held naturally at her side or in front of her in the full body views. Keep the same clean turnaround layout: full body front view, left profile, right profile, back view, plus the large closeup headshot panel. Same even flat lighting, orthographic framing, plain seamless pure white background. Show only this one character.
Owen Doerr 执导的《Lifeguard》
含义:同一美索不达米亚特征女性角色的转身参考图,穿陶土色救生衣,手持圆形救生圈,包含正面、侧面、背面及特写视图。
A woman and a boy walk away from the sea toward a white lifeguard tower on a wide empty beach00:05

它所提供的镜头

镜头提示词指定了三张角色表,救生员、男孩和她的护身符,并要求完全保留两张角色表。

Duration: 5s. Aspect ratio: 16:9. No background music, no score. Ambient sound effects only — waves washing in, sand underfoot, wind building from a soft breeze to hard howling gusts by the end.

The Beach, Lifeguard Tower Distant sets the beach setting — sand, the lifeguard tower, the away-from-ocean angle. Lifeguard Character Reference Sheet is the lifeguard — preserve her character sheet exactly. Young Benji Character Sheet is Benji — preserve his character sheet exactly, keep wardrobe consistent. Pazuzu Amulet Reference Sheet sets the Pazuzu stone amulet on a cord around the lifeguard's neck.

Camera: locked-off within each shot, no handheld drift. Hard cuts between the three shots.

Color grade: washed-out warm bleach-bypass, cross-processed, dreamlike remembered mood — faded reds, milky whites, golden nostalgic light. Strict palette: bright aqua 
#5FC6C2 sea, dusty coral 
#E8A87C highlights and reflected light, warm sand 
#F2C896 beach, hazy sky blue 
#C4D4D0 sky and haze, bleached white 
#F0E6C8 brightest whites. No colors outside this palette. Soft grain, atmospheric haze.

SHOT 1 (0–5s):
Preserve The Beach, Lifeguard Tower Distant exactly — same framing, same beach, same tower in the distance, same angle facing away from the ocean. The lifeguard and Benji walk together away from camera, heading up the sand toward the tower in the distance, her arm around his shoulders steadying him as they go — he leans into her, moving slowly. Seen from behind. Wind light, their hair and the tower's small flag just beginning to stir.
Owen Doerr 执导的《Lifeguard》
含义:沙滩场景中救生员扶着男孩本吉走向远处瞭望塔,采用褪色暖色调和梦幻般的漂白效果,风逐渐增强。

整个角色组的七视图表

《静止时光》画布上的 13 张角色表始于一个七视图提示词,四个全身视图和三个面部特写,这样人群场景中的每个小镇居民都可以按名字选角。

《静止时光》· 米格尔

从一位男性的两张参考图生成的七个视图

一位戴灰色帽子、穿棕褐色马甲和牛仔裤的老年男性的七视图角色表:四个全身视图和三个面部特写

12 号男性角色表

为七个视图编号并以"不改变姿势。不改变服装。"结尾,确保每个画面都是同一男性的中性记录。

Create a full professional character sheet of the same person shown in the reference images (image 1 full body, image 2 facial close up).

Maintain identical facial features, hairstyle, cap, skin texture, and clothing.

Generate the following views in a clean studio setting with neutral soft lighting and a plain light background:

1. Full body front view
2. Full body left profile
3. Full body right profile
4. Full body back view
5. Close up front portrait
6. Close up left profile
7. Close up right profile

Neutral expression.
Arms relaxed at sides.
No stylization.
No dramatic lighting.
No pose changes.
No outfit changes.

Highly detailed, sharp focus, consistent proportions across all views.
Marius Jensen 执导的《Still Time》
含义:根据参考图像生成同一人物的专业角色设定表,包含全身和面部多角度视图,中性光照和背景。
Three men sit on a red pickup's tailgate outside a gun shop at night, the man in a cap and vest in the middle00:06

它所提供的镜头

镜头指定了三张角色表,并说明每位男性坐在哪个位置,这张角色表中的男性"坐在中间",位于枪店外。

REFERENCE SETUP

@Image1 = master composition and staging reference.
McClintock FullSheet= McClintock reference character sheet exactly.
McClintock Brother FullSheet= McClintock brother reference character sheet exactly.
Man 12 — Character Sheet= Miguel reference character sheet exactly.
Use @Image1 for:
- exact nighttime street location outside McClintock’s Guns & Ammo
- red pickup truck placement
- storefront lighting and practical night ambience
- seating positions of all three men
- Miguel / Man12 seated in the center
- McClintock with glasses seated on the left
- McClintock’s brother seated on the right
- rifles resting naturally with the two McClintock brothers
- beer can already in Miguel’s hand
- overall rural small-town late-night atmosphere

IMPORTANT:
Do not change the characters’ positions or the street geography.
The rifles remain completely passive props throughout the shot. Nobody lifts, aims, handles, or gestures with them.

DURATION

6 seconds.

SCENE INTENTION

A quiet, slightly humorous late-night respite.

These men have been sitting outside for a while. The conversation is casual and familiar. Miguel is telling some story or making a comment; the McClintock brother with glasses finds it genuinely funny.

Meanwhile the other McClintock brother has completely checked out and fallen asleep in his chair.

The humor should come from the contrast:
Miguel animatedly talking → McClintock laughing → Miguel casually taking a sip → the third man peacefully snoring through all of it.

Natural, understated, lived-in behavior.

SHOT DESIGN

One continuous medium-wide shot based closely on @Image1.

Very subtle cinematic push-in throughout the six seconds.

Do not cut.
Do not radically reframe.

TIMECODED ACTION

0:00 - 0:02

Miguel, seated in the center, is already speaking casually to McClintock on the left.

He makes one small conversational gesture with his free hand while holding the beer can naturally in the other.

His body language is relaxed and familiar, as if finishing a mildly amusing story.

McClintock with glasses watches Miguel, then breaks into genuine laughter.

His laughter should feel spontaneous:
- shoulders move slightly
- head dips or tilts back naturally
- broad but realistic smile
- relaxed body posture

The rifle remains resting passively across his lap.

0:02 - 0:04

Miguel smiles faintly at McClintock’s reaction.

He raises the beer can and takes one natural sip.

Do not exaggerate the drinking action.
Just a casual, ordinary sip while sitting.

McClintock’s laughter settles into a lingering grin.

0:04 - 0:06

Miguel lowers the can again.

At the same time, subtly draw attention to the McClintock brother on the right.

He remains asleep in his chair:
- head slightly dropped forward or tilted
- body heavy and completely relaxed
- slow breathing
- a soft audible snore

He does NOT wake up.

Miguel and the McClintock with glasses briefly register his snoring without making a big reaction — perhaps a tiny glance or knowing amused expression.

End on the relaxed tableau of all three men.

CAMERA

Single continuous shot.

Slow, almost imperceptible push-in.

Natural cinematic handheld stability is acceptable, but keep it restrained.

No dramatic camera movement.
No pans toward the rifles.
No close-up of weapons.

PERFORMANCE NOTES

MIGUEL / MAN12
- relaxed
- friendly
- naturally conversational
- small hand gestures only
- drinks casually
- no exaggerated drunken behavior

McCLINTOCK WITH GLASSES
- attentive to Miguel
- genuine spontaneous laugh
- warm, familiar interaction
- laughter subsides naturally into a smile

McCLINTOCK BROTHER ON RIGHT
- asleep from the beginning
- heavy relaxed posture
- subtle breathing
- soft periodic snoring
- completely unaware of the conversation
- does not wake up

AUDIO

No music.
No score.

Natural nighttime street ambience only:
- faint distant traffic
- subtle night insects if appropriate
- soft storefront hum

Miguel may speak conversationally, but keep his words indistinct and non-specific unless dialogue audio is supplied.

McClintock gives a short natural laugh.

The sleeping brother produces a soft, believable snore near the end.

No extra voices.

CONTINUITY RULES

Keep all three men seated throughout.
Keep the beer can with Miguel.
Keep both rifles resting exactly as passive objects.
Do not introduce pedestrians or additional characters.
Keep truck doors and storefront unchanged.
Maintain the same nighttime lighting and composition.

NEGATIVE PROMPT

no music
no new characters
no standing up
no weapon handling
no aiming rifles
no threatening gestures
no gunfire
no aggressive behavior
no drunken staggering
no exaggerated drinking
no slapstick comedy
no waking the sleeping man
no loud cartoon snoring
no random dialogue
no camera whip
no sudden zoom
no cuts
no lighting changes
no vehicle movement
no distorted hands
no object duplication
Marius Jensen 执导的《Still Time》
含义:三个男人深夜坐在枪店外的卡车旁,米格尔讲故事时麦克林托克大笑,另一个兄弟在椅子上打瞌睡,自然写实的夜景氛围。

《静止时光》· 玛莎

相同的七视图提示词,用于人群中的一张面孔

一位金发老年女性穿着绿色绗缝马甲和牛仔裤的七视图角色表:四个身体视图,三个特写

穿绿色马甲的女性角色表

提示词采用相同的七视图布局,完全不描述这位女性,因为参考图像已经承载了她的外观。

Create a full professional character sheet of the same person.

Maintain identical facial features, hairstyle, skin texture, and clothing.

Generate the following views in a clean studio setting with neutral soft lighting and a plain light background:

1. Full body – front view
2. Full body – left profile
3. Full body – right profile
4. Full body – back view
5. Close-up – front portrait
6. Close-up – left profile
7. Close-up – right profile

Neutral expression.
Arms relaxed at sides.
No stylization.
No dramatic lighting.
No pose changes.
No outfit changes.

Highly detailed, sharp focus, consistent proportions across all views.
Marius Jensen 执导的《Still Time》
含义:制作同一人物的专业角色设定表,包含全身和特写的多角度视图,中性表情,工作室照明,简洁背景。
A small-town crowd stands on a street at morning, a blond woman in a green vest near the front00:05

它所提供的镜头

一个人群反应镜头用简单的文字赋予这张表单一个任务:用它表现"Martha 的面容、发型、比例、绿色背心装扮、年龄"。

REFERENCE SETUP

Hf 20260818 152833 586bfd47 Cc9c 45e5 A7a7 82ec0c63b786 = master crowd composition and staging reference.
Use it for the crowd layout, street angle, storefront geography, morning lighting, and overall small-town atmosphere.

Woman 1 Character Sheet = Laura character reference.
Use it for Laura’s face, hairstyle, proportions, blue cardigan outfit, age, and expression language.

Woman in Green Vest, Character Sheet = Martha character reference.
Use it for Martha’s face, hairstyle, proportions, green vest outfit, age, and expression language.

SCENE INTENTION

This is a short reaction beat.

The crowd is already gathered and watching with expectation.
Then Laura notices something off-camera.
She reacts first.
She looks sharply toward the off-camera event and points in the same direction she is looking.

Martha immediately senses that something is wrong.
She moves closer to Laura.
Both women share a concerned reaction.

The rest of the crowd also begins reacting, not with chaos, but with a ripple of shared concern and alert attention.

The feeling is:
something important has just changed off-camera.

DURATION

4 to 5 seconds.

SHOT DESIGN

One continuous shot.

Begin closely matched to Hf 20260818 152833 586bfd47 Cc9c 45e5 A7a7 82ec0c63b786, with the crowd already standing and watching off-camera.

The camera performs a subtle cinematic push-in through the shot.
Very gentle and steady.
No abrupt movement.

The framing should keep Laura and Martha clearly readable near the front of the crowd, while preserving the rest of the townspeople around them.

TIMECODED ACTION

0:00 - 0:01
The crowd stands expectant and watchful.
They are already focused on something off-camera.
Subtle micro-movements only:
small posture shifts,
blinks,
light breathing,
tiny head adjustments.

Laura and Martha are calm but attentive.

0:01 - 0:02
Laura is the first to notice a change.
Her eyes sharpen and shift more decisively toward the off-camera point.
Her expression changes into concern.

She raises her hand and points in the same direction she is looking.
Important:
the pointing gesture should feel natural and instinctive,
not theatrical,
not exaggerated.

0:02 - 0:03
Martha reacts to Laura’s cue.
She moves a little closer toward Laura, with a concerned expression.
She does not overreact.
She simply closes the distance slightly, as if needing to see better or confirm what Laura has noticed.

Her attention locks onto the same off-camera point.

0:03 - 0:05
A ripple of reaction passes through the rest of the crowd.
Several people shift their gaze more intently.
A few lean subtly.
Some narrow their eyes, tense their faces, or slightly reposition themselves.
The group collectively becomes more concerned and alert.

Do not turn this into panic.
This is a restrained, believable crowd reaction.

END BEAT

End with:
- Laura still pointing,
- Martha closer to her,
- both women concerned,
- the crowd around them more visibly attentive and uneasy.

CAMERA DIRECTION

Single continuous shot.
Slow cinematic push-in for the whole clip.
No pan unless absolutely minimal.
No handheld shake.
No crash zoom.
No whip movement.

The camera should feel observant and grounded.

PERFORMANCE NOTES

Laura:
She is the first to notice something important.
Her reaction should be clear and readable.
She turns her attention more sharply,
her face shows concern,
and she points naturally toward the off-camera event.

Martha:
She reacts to Laura’s cue.
She moves slightly closer to Laura and looks in the same direction.
Her expression should carry concern and alertness.

Crowd:
They should not remain frozen.
They should react in a subtle wave:
head turns,
posture changes,
small tension,
concern spreading through the group.

Do not let the crowd become noisy or chaotic.
Keep it believable and controlled.

AUDIO NOTES

No music.
No score.
No dialogue.

Only natural production sound:
soft morning ambience,
faint town atmosphere,
subtle clothing rustle,
small crowd shuffling,
light footsteps or repositioning,
quiet breathing.

CONTINUITY RULES

Keep the composition grounded in @[Image 1](image_1).
Keep Laura visually consistent with @[Image 2](image_2).
Keep Martha visually consistent with @[Image 3](image_3).
Keep the morning light warm and natural.
Keep the crowd in the same general arrangement.
Do not introduce new characters.
Do not change wardrobe.
Do not change store geography.

NEGATIVE PROMPT

no dialogue
no music
no score
no panic
no screaming
no running
no exaggerated pointing
no overacting
no cartoon facial expressions
no chaotic crowd movement
no frozen mannequin crowd
no cheering
no laughing
no random smiling
no dramatic camera whip
no crash zoom
no shaky handheld
no jump cuts
no surreal behavior
Marius Jensen 执导的《Still Time》
含义:劳拉注意到画外事件后指向它,玛莎靠近她,人群中关切情绪波动,柔和晨光下小镇街景缓慢推进镜头。

正面和两侧侧面:Catch Out 的标注转身图

Catch Out 的角色转身图用第二个侧面、正面、左侧和右侧视图替换了背面视图,每个面板都带有标签,影片在整个团队(包括狗)的群体镜头中命名了它们。

Catch Out · Gumby

三个带标签的面板,在 13 个片段中被命名

一个身材修长的年轻男子的三个带标签视图,穿着橄榄色夹克、红色头巾和棕色灯芯绒裤:正面、左侧、右侧

Gumby 角色转身图

一句话就包含了整套服装,从褪色的红色头巾到磨损的白色运动鞋,并要求"在每个面板中"呈现。

Professional character turnaround reference sheet. Show the SAME man (Gumby) in 3 panels side by side: front view, left profile, right profile. Full body, neutral A-pose, plain white seamless background, even studio lighting, orthographic framing. Identical face (23, tall lanky, scruffy patchy facial hair, easy laugh lines), same outfit in every panel: faded red bandanna around neck, oversized olive thrift store jacket, baggy worn brown corduroy pants, scuffed white sneakers. Label each panel: FRONT, LEFT, RIGHT.
Dr. Dreams 执导的《Catch Out》
含义:一个瘦高男子的三视图角色参考表,穿橄榄色夹克和灯芯绒裤,白色背景工作室光效。
Three young drifters walk along a desert highway shoulder at sunset, one carrying a dog00:08

它所提供的镜头

团队在高速公路路肩行走的镜头,其提示词命名了四个角色转身图:三个人和他们的狗。

no background music. 
[00:00] exterior freeway Highway Shoulder, Hitchhiking. The three people patches Patches Character Turnaround, casey Casey - Turnaround Sheetand Gumby Gumby Character Turnaroundwalk along the shoulder. patches holds her thumb out to hitchhike. Casey carries thor Thor Character Turnaround. A massive semitruck thunders past them. 
[00:03] camera tracks behind them. Patches gives a middle finger to the semi truck as it speeds away down the highway. 
[00:06] close on their old, worn out sneakers, walking along the freeway shoulder. 
Dr. Dreams 执导的《Catch Out》
含义:三个年轻人和一只猎犬在高速公路边搭便车,半挂卡车驶过,特写他们破旧的运动鞋。

Catch Out · Thor

一只狗的角色转身图

一只棕白相间猎犬的三个带标签视图,有着长而下垂的耳朵和破旧的项圈:正面、左侧、右侧

Thor 角色转身图

一只动物获得了与演员相同的处理:体型、耳朵、毛色、项圈和"深情忧郁的下垂眼睛",在每个面板中保持一致。

Professional character turnaround reference sheet for a dog. Show the SAME hound dog (Thor) in 3 panels side by side: front view, left profile, right profile. Full body, standing, plain white seamless background, even studio lighting, orthographic framing. Identical dog in every panel: medium-large hound mix, very long floppy droopy ears, tan and white coat with brown patches, road-dusty fur, worn leather collar, soulful sad droopy eyes, stocky legs. Label each panel: FRONT, LEFT, RIGHT.
Dr. Dreams 执导的《Catch Out》
含义:一只猎犬的三视图角色参考表,长耳朵棕白花色,白色背景工作室光效。
A tan and white hound runs beside young people crossing railroad tracks at night under floodlights00:15

它所提供的镜头

Thor 在一个追逐镜头中与团队一起跑过铁轨,其提示词在摄像机横向跟拍时命名了所有四个角色转身图。

ive action film. no background music. night, trainyard Trainyard Edge, No Fences. [00:00] camera dolly, tracking along side the crew Patches Character TurnaroundGumby Character TurnaroundCasey - Turnaround Sheet, Buzzed HeadThor Character Turnaround as they run across the tracks [00:03] cut to close on casey, glancing over his shoulder as he runs, camera handheld
[00:06] cut to casey's POV looking back as a  second security truck suddenly pulls out onto the road  behind him, headlights and a blinding white spotlight blasting straight at him [00:09] camera tracking behind the three of them as another truck suddenly pulls out in front of them, and shining a spotlight on them. they freeze, throwing hands up against the glare, silhouetted, caught. hard tension, engine roar, no music.
[00:12] cut to medium camera orbits in front of them, they're blasted with the spotlight, squinting against the harsh light. 
Dr. Dreams 执导的《Catch Out》
含义:实景电影风格,夜晚火车调车场,镜头跟随Patches、Gumby、Casey和Thor角色奔跑穿越铁轨,保安卡车包围并用强光照射他们,无背景音乐。

当角色发生变化时创建新表

当故事中的角色发生变化,变成幽灵或换了新发型时,制作者会制作一个新的转身参考,保留面部特征并改变一处细节,然后在后续镜头中使用新角色表的名称。

完美求婚 · Jeremy

幽灵转身参考,在59个视频片段中使用

一名身穿西装的男子在黑色背景上呈现为发光的半透明蓝色幽灵的三个视角:正面、左侧和背面

Jeremy 幽灵转身参考

提示词保留了他真实的面部和西装,包括左手腕上的手表,并在真实感底图上叠加蓝色幽灵效果。

Create a character turnaround reference sheet of this man, Jeremy (referenced by images 1 and 2 for his exact face, features, hairstyle and his suit outfit with the wristwatch on his left hand), rendered as a glowing translucent blue Force ghost / Jedi spirit. ONE image, three full body views of the SAME character side by side on a plain near black background: front view, left side profile view, and back view, neutral standing A-pose, even framing, feet fading into faint blue wisps.

CRITICAL LOOK: this must look like a REAL LIVE ACTION FILM with a VFX overlay, NOT an animation or cartoon. Preserve fully photorealistic skin texture (pores, subtle imperfections), realistic hair strands, realistic fabric weave on his suit, real cinematic lighting and photographic detail. On top of that photoreal base, apply a translucent glowing blue spirit effect: a cool blue tint over his skin and clothes, semi transparency so the dark background faintly shows through his edges, a soft blue rim light and faint glowing aura outline, and subtle floating blue light particles. The realism of skin and hair must remain clearly visible through the blue glow, like a practical hologram effect graded onto real footage.

He is NOT wearing Jedi robes. He keeps his ORIGINAL suit outfit from the reference (jacket, dress shirt, wristwatch on left wrist). Identical character, suit, proportions and blue glow in all three views. Photorealistic cinematic, ARRI Alexa film still with holographic blue ghost VFX, 21:9.
Zhelun “Jerry” Lin 执导的《The Perfect Proposal》
含义:西装男子Jeremy的三视图参考表,正面侧面背面站立姿势,真人电影摄影质感加蓝色半透明绝地英灵特效光晕,保留真实皮肤纹理和西装细节。
Close-up of a young man with long dark hair glowing faintly blue against blurred city lights at night00:12

它所提供的镜头

一个80毫米特写忏悔镜头使用幽灵角色表作为他的面部参考,镜头提示词排除了画面中的其他所有人物。

使用 ARRI Alexa 35真实电影质感,80mm大特写镜头。

严格保持与前面Jeremy近景镜头相同的机位方向与世界连续性 ChatGPT Image 2026年8月5日 00 08 31,Alexa。摄影机仍位于Jeremy的侧后方偏侧面,但距离更近,形成 Jeremy Jeremy Force Ghost Turnaround的大特写。

这颗镜头中Abigail完全不入画:

不出现Abigail的肩膀;
不出现Abigail的头发;
不出现任何人物前景;
画面中只有Jeremy。

这是一个连续固定单镜头。核心动作动词是:

confess
expose his fear
seek reassurance
fail to finish the confession

Jeremy并不是继续抱怨求婚计划。他终于坦白:自己之所以执着于时机、烟花和戒指,是因为他不相信不够完美的自己也值得被Abigail选择。

【镜头与构图】
80mm大特写;
Jeremy位于画面右侧或中偏右;
几乎不留头部空间;
清楚看到Jeremy大半个侧脸;
清楚看到眼睛、眉毛、鼻梁、嘴部、下颌与呼吸变化;
不能拍成纯背影;
不能只看到后脑勺;
头发不能遮住眼睛和嘴部表演。

摄影机全程固定:

不切镜;
不推拉;
不摇移;
不变焦;
不自动重新构图;
不使用慢动作或隐藏剪辑。

背景是高度虚化、真实、安静的城市夜景。不要钢筋、横梁、瓦砾、残骸或近距离高楼。

【人物初始状态】

这颗镜头发生在Abigail说完:

“I know. You’re not ready yet.”

之后。

Jeremy已经被她说中。

镜头开始时:

他没有看Abigail;
视线停在较低位置;
下颌轻微绷紧;
呼吸略微不稳定;
嘴唇张开后又停住;
他正在决定是否把真正的恐惧说出来。

他不是愤怒,也不是辩解。
他感到难堪,因为Abigail已经看穿了他,而他终于不得不承认她是对的。

【表演时序|约10–12秒】
0–2秒:寻找坦白的入口

Jeremy轻微吸气。

他的眼神先在较低位置停留,然后缓慢向一侧移动,短暂扫过周围已经消失的餐厅空间。

他低声说:

“I just...”

说完后停顿一拍。

“I just”不是口头禅,也不是机械停顿。
这是他想坦白,却不知道该从哪里开始。

可见变化:

眼睛轻微左右游移;
嘴唇想继续,却暂时停住;
下颌略微收紧;
呼吸短暂卡住;
不做任何手势。
2–5秒:承认自己准备得太多

Jeremy继续说:

“I had a whole thing ready.”

说这句话时,他的视线缓慢扫过远处和周围,像是在回想那个已经彻底消失的完整计划。

他的表演不是炫耀,而是一种羞于承认的自我暴露:

我不是没有认真。
我是认真到必须控制一切。

可见变化:

眼神短暂往远处移动;
随后轻微看向镜头外Abigail所在的方向;
但还没有真正和她保持眼神接触;
嘴角出现一点无奈的收紧;
声音比开头更坦白,也更脆弱。
5–8秒:列出自己依赖的“安全凭证”

Jeremy说:

“Timing, fireworks, the ring.”

三个词之间保留短而自然的停顿。

“Timing”
眼神向一侧移动;
像是在回想自己对时间和流程的控制;
声音里有一点执着和尴尬。
“fireworks”
视线短暂抬向夜空或远处城市上方;
像是在寻找那个原本应该证明一切完美的场面;
眼神很快重新落下。
“the ring”
视线明显向下;
声音进一步放轻;
因为戒指最接近他真正想问的问题。

这些不是普通清单。
Jeremy把它们当作让自己“值得被接受”的证明。

8–12秒:真正坦白,然后无语凝噎

说完“the ring”后,Jeremy停顿一拍。

他轻微吸气,终于缓慢抬起眼睛,真正看向镜头外Abigail所在的方向。

这是这颗镜头最重要的眼神变化:

前面他一直在看周围、看夜空、看低处;
到这里,他第一次真正看向Abigail;
像是在鼓起勇气,直接把自己的恐惧交给她。

他轻声说:

“If it’s not perfect...”

这里不是理论分析。
这是一次真正的坦白。

声音比之前更轻、更私人。
眼神短暂维持在Abigail的方向,里面有询问、羞耻和对答案的恐惧。

停顿后,他继续:

“why would you...”

说到“you”时,他仍然看向Abigail,像是真的准备问出:

“Why would you marry me?”

但是他无法说出 “marry me”。

表演必须明确表现出这个没有完成的句子:

“you”之后,Jeremy的嘴唇仍然保持微微张开;
他已经准备形成下一个词,但没有声音出来;
喉咙轻微收紧;
呼吸突然停住;
他做一个非常轻微的吞咽动作;
眼神无法继续停留在Abigail身上;
眼睛先从她身上移开;
随后视线缓慢向下垂落;
头部也轻微低下;
嘴唇最终闭合,整句话彻底中断。

这不是普通的“忘词”或停顿。
是Jeremy清楚知道下一个词是 “marry me”,但他无法承受亲口问出这个问题。

他无语凝噎,但不要夸张哭泣:

不哭喊;
不抽泣;
不大幅哽咽;
不做戏剧化颤抖;
可以有非常轻微的眼眶湿润感,但不要掉泪;
重点是声音消失、呼吸卡住和视线低垂。

镜头结束时:

Jeremy低垂着眼睛;
嘴唇已经闭上;
他没有完成问题;
身体保持原位;
安静停留在坦白失败后的沉默中。
【Jeremy的表演核心】

这一段不是:

“我设计了一套失败的求婚方案。”

而是:

“我以为只有把所有事情做得完美,你才可能选择我。”

他的完整情绪过程是:

羞于承认 → 尝试解释 → 环顾失去的一切 → 短暂看向Abigail → 真正开始坦白 → 接近“marry me” → 无法说出口 → 视线低垂。

“why would you...”必须带有一种几乎在向Abigail寻求确认的感觉。
他真正想问的是:

如果求婚不完美,如果我不完美,你为什么还会嫁给我?

【声音】

没有音乐,没有背景配乐。

只保留:

Jeremy近距离、真实的声音;
轻微呼吸;
“why would you...”之后突然停止的气息;
极轻的吞咽声;
极轻衣物摩擦;
微弱风声;
远处城市环境底噪。

“why would you...”之后必须留下真实的沉默。
不要用音乐填满,不要自动补出台词。

【严格禁止】
不要让Jeremy平淡朗诵整段台词;
不要让他始终看着同一个方向;
不要省略环顾周围;
不要省略看向镜头外Abigail;
不要省略最后视线低垂;
不要让他在说“why would you...”时仍然随意看夜景;
这句话必须像真正向Abigail坦白;
不要让他说出“marry me”;
不要自动补完整句子;
不要把中断演成忘词;
不要把中断演成机械卡顿;
不要把这段演成指责Abigail;
不要把这段演成愤怒抱怨;
不要加入大幅手势;
不要夸张哭泣或抽泣;
不要让Abigail入画;
不要加入额外台词;
不要加入音乐;
不要切镜、推拉、摇移或变焦。
Zhelun “Jerry” Lin 执导的《The Perfect Proposal》
含义:用ARRI Alexa拍摄Jeremy的80毫米大特写,他坦白自己因不够完美而恐惧被拒绝,说到一半哽咽停顿,视线低垂,电影质感。

Catch Out · Casey

同一个转身参考,只改变了发型

一个瘦削年轻男子的三个标注视角,剃了平头,穿着格子法兰绒衬衫和破洞牛仔裤

Casey 转身参考,平头造型

编辑保留了他第一个转身参考的所有内容,仅改变了头发,因此篝火剪发后的镜头仍然展示同一个Casey。

This is a professional character turnaround reference sheet for a young man named Casey (the source image). Keep EVERYTHING identical: the same person, the same facial features, proportions, clothing, the same multiple views (front, side, back), the same neutral A-pose, the same plain seamless light-gray background, the same even lighting and orthographic framing. Change ONLY his hairstyle: give him a freshly buzzed/shaved head, a very short close-cropped clipper shave, scalp visible, consistent across every view in the sheet. Do not change anything else about his appearance or the layout.
Dr. Dreams 执导的《Catch Out》
含义:年轻男子Casey的角色转面参考图,保持所有特征和服装不变,仅将发型改为贴头皮的平头短发,多角度展示。
Four people stand on railroad tracks at night in truck headlights, a buzzed young man in plaid on the right00:15

它所提供的镜头

在保安卡车灯光的照射下,平头造型的Casey与工作人员站在一起,这个镜头使用了他的新角色表名称,而不是旧的。

Night trainyard Wide Kids Facing The Truck 2 4 1. The three kids Patches Character TurnaroundCasey - Turnaround Sheet, Buzzed HeadGumby Character Turnaround stand frozen in the lights of the security truck. [00:00] [00:00] handheld camera tracking the The Bull  he grips Patches by the wrist and marches her toward his truck . Gumby takes off running into the night. 

[00:04] Casey watches and yells "Traitor!". [00:06] close on patches being walked to the truck. she looks back to casey and yells "Go! Run!"[00:09] Cut to close on Casey. He does not run. Jaw set.[00:012] Cut to Thor planting his feet on ground, baying loud at the Bull.[00:15] A second set of headlights swings into the yard behind them — backup arriving.
Dr. Dreams 执导的《Catch Out》
含义:夜晚火车站场三个孩子被保安车灯照到其中一个被抓另一个逃跑狗狂吠援军到达呈现手持摄影风格

什么是优秀的角色转身表?

一个好的转身参考角色表应在纯色背景上以单一比例展示一个角色,采用中性姿势和均匀光照,至少包含正面、侧面和背面视角,并且不包含任何镜头不应复制的元素。
本页面上的10个转身参考角色表及其提示词中使用它们的镜头(包括拍摄次数),统计于2026年10月
表格影片视图命名它的镜头
男性劫匪完美求婚正面、侧面、背面64(21个影片)
Jeremy 鬼魂完美求婚正面、侧面、背面66(59个影片)
MilaMila & Sasa四个全身视图4(2个影片)
成年 Benji救生员四个视图和一个头部特写9(9个影片)
救生员救生员四个视图和一个头部特写7(6 个影片)
男子 12Still Time七个视图17(11 个影片)
绿色背心女子Still Time七个视图15(7 个影片)
GumbyCatch Out正面、左侧、右侧13(13 个影片)
ThorCatch Out正面、左侧、右侧11(11 个影片)
Casey,寸头Catch Out正面、左侧、右侧8(8 个影片)
  1. 1. 保持背景简洁

    所有 10 张图片都放置在纯白色、灰色或黑色背景上,这样模型会复制角色而不是房间。

    plain seamless pure white background
    Owen Doerr 执导的《Lifeguard》
    含义:纯白色无缝背景
  2. 2. 让角色摆成中性的 A 字姿势

    A 字姿势,双臂稍微远离身体,正交框架,平面无透视,可以在每个面板中展示真实的比例。

    Full body, neutral A-pose, plain white seamless background, even studio lighting, orthographic framing.
    Dr. Dreams 执导的《Catch Out》
    含义:全身中性A字站姿人物纯白背景均匀影棚光正交构图
  3. 3. 在每个镜头提示词中说明设定表的作用

    为设定表命名并说明其用途,这样模型才能知道这是一个角色而不是要复制的图片。

    preserve her character sheet exactly
    Owen Doerr 执导的《Lifeguard》
    含义:完全保留原角色设定表
  4. 4. 告诉模型不要复制布局

    设定表包含标签和多个姿势,镜头可能会复制它们,除非提示词明确排除。

    Do not reproduce reference sheets, text, or multiple poses.
    Katrin Ann Orbeta 执导的《Mila & Sasa》
    含义:不生成设定表文字或多个姿势
  5. 5. 每次只改变一件事

    要更新角色时,从现有的设定表开始,只说明单个改变,这样面部和服装才不会偏移。

    Change ONLY his hairstyle
    Dr. Dreams 执导的《Catch Out》
    含义:仅改变他的发型

根据这些表格构建的角色转身提示词模板

填写括号内容并删除不需要的行。

Character turnaround reference sheet of [name], from the reference image.
Views side by side, same scale: front, [left profile], [right profile], back.
Optional: one large close-up headshot panel.
Full body, neutral A-pose, orthographic framing.
Plain seamless [white / light grey] background, even studio lighting.
Identical in every panel: [face], [hair], [costume, with colours], [signature detail].
Show only this one character. No props, no scene, no text.
含义:角色转身设定表包含正面侧面背面特写全身A字站姿纯白背景影棚光

如何使用 AI 制作角色转身表?

从你喜欢的设计开始,让图像模型生成纯色背景上的正面、侧面和背面视图,对照设计检查每个面板,然后将设定表附加到每个镜头并在提示词中命名。
  1. 1从设计开始

    先生成或上传角色,因为这里 10 个设定表中有 8 个是从角色图像制作的,而不是仅从文字描述制作的。

  2. 2生成视图

    让 GPT Image 2、Nano Banana Pro 或 Nano Banana 2 在纯色背景上并排生成视图;这三个模型都可以在 Free 计划中使用。

  3. 3检查每个面板

    对比各面板中的面部、头发、服装和标志性细节,当某个面板偏移时,使用一个明确的改变重新生成。

  4. 4在每个镜头中指定角色

    将角色表附加到每个视频提示词,例如 Standard 套餐的 Seedance 2.0,并说明是哪个角色。

角色转身问题

什么是角色转身参考表?

角色转身参考表是一张图像,展示角色从多个固定角度的视图,通常包括正面、侧面和背面,在纯色背景上以相同比例呈现。动画师、游戏美术师和 AI 影片创作者使用它来绘制或生成每个镜头中的同一角色。

角色转身参考表应该包含哪些视角?

转身参考表应至少包含正面、一侧侧面和背面的中性姿势。许多参考表还会添加四分之三视角、另一侧侧面或用于对话镜头的大型面部特写。

转身参考表与模型表或角色参考表是一回事吗?

这些术语有重叠。转身参考表是一组角度视图;角色参考表或模型表是整个页面,可以在转身视图旁边添加面部特写、表情、道具和注释。

如何用 AI 制作角色转身参考表?

从角色图像开始,让图像模型在纯色背景上以中性 A 字姿势并排生成正面、侧面和背面视图,并说明每个面板中的面部和服装必须完全相同。

如何使用转身参考表保持 AI 角色的一致性?

将参考表作为参考图像附加到每个图像和视频提示词中,在提示词中指定它并说明其用途,例如"完全保留她的角色表"。添加一行说明,告诉模型不要复制参考表的标签或姿势。

角色转身参考表应该有标签吗?

诸如 FRONT、LEFT 和 RIGHT 等标签可以帮助人们阅读参考表,一些影片创作者会要求添加它们。其他人则要求不添加文本,因为模型可能会将标签复制到镜头中;如果保留标签,请在镜头提示词中说明不要复制文本。

我可以为动物或 2D 角色制作转身参考表吗?

可以。同样的布局适用于狗、猫或平面2D卡通角色:用文字描述标记或配色方案,保持背景简洁,并要求每个视角保持一致。

在 Flick 中制作角色转身图是免费的吗?

是的,您可以免费开始:新账户在注册时获得100积分,加入 Flick Discord 后再获得200积分,而制作转身图的图像模型在 Free 套餐下即可运行。使用参考图像的视频模型(如 Seedance 2.0)需要 Standard 套餐。

更多关于 Flick 中一致角色的内容

合集中的影片

制作您的角色转身图

在一个画布上生成正面、侧面和背面视角,然后将该图表附加到每个镜头上,使角色保持一致。

在 Flick 中制作转身表

免费试用,无需信用卡。

让 AI 解读本页:ChatGPTPerplexityClaude