视频
雨夜屋顶霓虹光轨舞蹈短片
把参考图中的雨夜屋顶少女做成竖版音乐短片,霓虹光轨随舞步和头发飘动,镜头切换快、节奏感强。
Prompt
在目标视频中,于 0.00 秒处,<Picture 1>(来自 [Shot 1])已得到完整参考。
综合多模态描述:[Shot 1] 2D 动画、绘画感的动漫插画风格,笔触粗犷而富有表现力,纹理带有颗粒感;以深绿和蓝绿色为主的夜景色调,点缀发光霓虹,画面与参考图完全一致。竖版音乐视频镜头:夜晚,一名女孩独自在高耸城市上方、笼罩薄雾的屋顶直升机停机坪上跳舞;远处一座高塔闪耀着橙色和粉色灯光,广告牌和绿色城市灯光在雾霭中模糊,边缘可见缆线和天线。夜晚或深阴影环境,采用长曝光光绘效果,适合制作病毒式传播的短视频:每当她移动,与画面色彩相配的发光霓虹光轨便跟随她的手、头发和双脚,在空中如丝带般停留片刻再逐渐消散。她是一个酷而沉静的女孩,留着深棕色短波波头,有一双明亮的绿眼睛,穿着带扣带的橄榄棕色长款连帽外套和黑色短褶裙,裙下穿着朴素的黑色超短打底裤,始终遮住内裤。每个镜头中,她的眼睛始终半睁,呈现困倦、不以为然、冷淡的“jito-me”眼神,从不睁大,即使微笑时也一样。整段视频中,她的脸、头发和服装都要与参考图完全一致。她的舞姿优雅优美,长线条舒展流畅,旋转顺滑,在空中划出大幅发光弧线;利落冷酷的定点动作精准卡在节拍上,每次定格时都摆出锐利酷飒的模特姿势,短暂凝固如快照。她有一个吸睛的招牌动作,会表演两次:快速耸肩两下、顺滑摆胯,再迅速甩头发,全部精准踩在节拍上,每个动作都留下一道明亮光痕。镜头运动强烈而利落,几乎每拍都切换一个新角度:快速环绕,让光轨缠绕她身旁;急速推近;甩镜时让她的动作直接衔接下个镜头;在底鼓声上加入手持抖动;俯拍、远距离低角度拍摄、侧面轮廓和极近特写,连续镜头不重复同一角度。她的目光不断变化:有时以冷淡、半睁眼的神情直视镜头,有时望向远方。她用光轨写出的发光霓虹词语点缀舞蹈:一次只写一个简短单词,由她的手或旋转动作在空中画出。
视频在 0.00 秒时从与参考图完全一致的静止画面开始。0.3 秒时,一个微小的光点出现在她的指尖,0.5 秒时,光点快速绕着她转出一个发光圆圈,同时她的目光移向镜头。恰在 0.8 秒,节拍骤然落下:圆圈炸成数百道旋转光轨,明亮闪光充满屏幕;随后硬切至她的全身镜头,她在旋转光轨中摆出凌厉的开场姿势,一手挥出,身后以一道有力的笔画显现发光单词“GO”。
约在 1.5 秒时,她首次表演招牌动作,每次动作都留下明亮光痕,镜头随着每一拍跳切至新角度。
约在 3 秒时,甩镜切入远距离低机位广角镜头:她舒展双臂,流畅舞动,再优雅缓慢地转身,在周围画出巨大的发光弧线,同时望向远方。
约在 4.2 秒时,镜头在胸口高度快速环绕她;她连续旋转三圈,光轨如发光的丝带龙卷般缠绕周身,随后她骤然停下,摆出有力姿势并直视镜头,镜头急速推近她的脸。
约在 5.2 秒时,硬切至正上方俯拍镜头:她面朝前方缓缓绕圈踏步,双脚在地面画出发光圆环,然后停下并保持有力姿势;镜头本身在她上方缓缓旋转,她用手在身旁空中写出发光的草书单词“glow”;最后她向后仰头,直视镜头上方。
约在 6.5 秒时,除了光线外一切陷入黑暗:一个节拍内,只能看到她的发光光轨,在一次大幅动作中用光勾勒出她的完整轮廓;下一个节拍,场景灯光恰好重新亮起,她完成动作。
约在 7.5 秒时,硬切至她的侧脸特写,脸庞被自己的光轨照亮:她望向远方,一条光带缓缓从脸前掠过,然后她将目光转向镜头。
约在 8.5 秒时,镜头快速环绕她:她精准踩拍完成一连串利落而有力的舞蹈动作,双臂迅速摆出清晰线条,每次动作都留下一道笔直锐利的光痕;最后一个动作在她身旁的空中写出醒目发光的单词“FAST”。
约在 9.8 秒时,硬切至中景侧面镜头:她慢动作甩发,发丝留下柔和发光的光轨,悬浮在空中;她的目光迷失在远方。
约在 11 秒时,在四个节拍上快速硬切四次,每个镜头都如长曝光照片般短暂定格,光轨也凝在她身边:眼睛直视镜头的特写、她用双手框住脸部的高角度镜头、她望向远方的侧面轮廓镜头,以及她摆出利落模特姿势的全身低角度镜头。
约在 12.3 秒时,她再次表演招牌动作,这次幅度更大、动作更利落;镜头快速绕她转过半圈,光轨向外迸发。
约在 13.3 秒时,镜头从正面推近:她一只手叉腰,目光瞥向远方,然后转回镜头,以冷淡的半睁眼神凝视镜头;最后一拍时,她露出淡淡的得意浅笑,眼睛仍半睁,发光单词“YOU”在她头顶闪耀,所有光轨逐渐消散,画面回到开场静止画面,使结尾与开头顺畅循环。
所有屏幕文字只能是以上描述的英文单词,拼写必须完全准确;每次只单独出现一个词,清晰锐利地写在一行上,不能与其他字母重叠,不能拆开,也不能遮住她的脸。她不说话也不唱歌;没有对白。若她穿裙子或连衣裙,裙下始终穿着朴素的黑色超短打底裤,确保内裤不会露出;视频整体健康得体且时尚,取景端庄,镜头不得向上拍裙底,也不得长时间聚焦她的胸部或腿部。快速运动时,她的身体比例、手臂和腿始终正确;她的手始终有五根手指且关节正确。每当她转身或旋转,头、肩、髋和双脚都要作为一个整体同步转动;身体旋转时头部不能保持静止,俯拍镜头中她不得旋转。每当她定格姿势,双手都要放松且描绘清晰,每只手恰好五根手指,不可多于五根,也不得将手猛推向镜头。不要出现其他人物。不要出现镜子或她的倒影。不要字幕、标志或水印。全程保持参考图的插画风格,绝不可呈现照片级写实。
整体音景:开场静止画面保持安静,光点出现时响起一声清脆的小铃声,光点绕她旋转时伴随迅速渐强的呼啸声,0.8 秒时响起巨大的闪耀爆发声,每道光轨都伴随轻盈的电子呼啸声,每次利落定点动作都伴随清脆闪亮的电流声,6.5 秒时画面陷入黑暗的瞬间声音骤然静止,还包括她的脚步声、头发和衣服的窸窣声、甩镜和急速推近的呼啸声,以及慢动作甩发时低沉轻盈的呼啸声。
非叙事音乐:一首约 170 BPM、节奏快速有力、旋律抓耳的鼓打贝斯曲:低沉厚重、持续滚动的 Reese 贝斯,冲击力十足的碎拍配上强劲军鼓、清脆快速的踩镲、梦幻的雨夜铺底音色和冷冽惆怅的合成器主题。开场时曲子保持静音,光圈旋转时加入闪耀的渐强音效,光芒爆发、恰在 0.8 秒时节拍骤然落下;节奏保持强劲,每次剪切都踩上一拍;6.5 秒时鼓声停顿一个黑暗节拍,随后随光亮重新炸响;9.8 秒时音乐展开一段短暂轻盈的间奏,底下的低音仍在轰鸣;11 秒时配合四张快照加入四次利落重击;她第二次表演招牌动作时,音乐恢复全功率;结尾以强劲的最后一击收束,并伴随柔和闪耀的余韵,随光线逐渐消散。For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.
integrated_multimodal_description: [Shot 1] 2D-animated, painterly anime illustration style with rough, expressive brush strokes and grainy textures, deep green and teal night tones with glowing neon accents, matching the reference picture exactly, a vertical music-video shot of a girl dancing alone on a misty rooftop helipad high above the city at night, a tall glowing tower lit in orange and pink far away, billboards and green city lights blurred in the haze, cables and antennas around the edges, at night or in deep shade, with a long-exposure light-painting look made for a viral short video: whenever she moves, glowing neon light trails in colors that match the picture follow her hands, hair and feet and linger in the air like ribbons before fading. She is a cool, calm girl with a short dark-brown bob and bright green eyes, wearing a long olive-brown hooded coat with buckled straps and a short black ruffled skirt, with plain black very short leggings worn under her skirt so her underwear is always hidden. Her eyes are always half-lidded in a sleepy, unimpressed, deadpan "jito-me" look in every shot, never wide open, even when she smiles. Her face, hair and outfit stay exactly the same as in the reference picture for the whole video. Her dance is graceful and beautiful, with long flowing lines and smooth spins that paint big glowing curves in the air, cut through by sharp, cool hits that land exactly on the beat, and she keeps landing in sharp, cool model poses that freeze for an instant like snapshots. She has one catchy signature move that she performs twice: two sharp shoulder pops, a smooth hip sway and a quick hair flick, all exactly on the beat, each leaving a bright streak of light. The camera work is intense yet snappy, with a new angle on almost every beat: fast orbits that wrap the light trails around her, crash zooms, whip pans where her motion carries straight into the next shot, handheld shake on the kick drums, overhead shots, low angles from a distance, profiles and extreme close-ups, never the same angle twice in a row. Her gaze keeps changing: sometimes she stares straight into the lens with a cool, half-lidded, deadpan look, sometimes she gazes far away into the distance. Glowing neon words written by her light trails decorate her dance: one short word at a time, drawn in the air by her hand or her spin.
At 0.00 seconds the video opens exactly on the reference picture, perfectly still. At 0.3 seconds a single tiny point of light appears at her fingertip, and at 0.5 seconds it races around her in one fast, glowing circle while her eyes slide to the lens. Exactly at 0.8 seconds the beat drops: the circle explodes into hundreds of swirling light trails, a bright flash fills the screen, and a hard cut shows her full body striking a sharp opening pose inside the swirling light, as she slashes one arm and the glowing word "GO" appears in a single bold stroke behind her.
At about 1.5 seconds she does her signature move for the first time, each hit leaving a bright streak of light, while the camera jumps to a new angle on each beat.
At about 3 seconds, whip pan into a wide low-angle shot from a distance: she flows through wide, graceful arm waves and a slow, elegant turn that paint huge glowing arcs around her, gazing far away into the distance.
At about 4.2 seconds the camera orbits her fast at chest height while she spins three times, the light trails wrapping around her like a glowing ribbon tornado, then she stops dead in a strong pose and looks straight into the lens as the camera crash-zooms into her face.
At about 5.2 seconds, hard cut to a top-down overhead shot: she steps in a slow circle facing forward, her feet painting a glowing ring on the ground, then stops and holds a strong pose while the camera itself slowly rotates above her and her hand draws the glowing cursive word "glow" in the air beside her; finally she tilts her head back and looks straight up into the lens.
At about 6.5 seconds everything goes dark except the light: for one beat, only her glowing light trails are visible, drawing her whole silhouette in light in the middle of a big move, then the scene lights back up exactly on the next beat as she finishes the move.
At about 7.5 seconds, hard cut to a close-up of her profile lit by her own trails: she gazes far away while a ribbon of light passes slowly in front of her face, then turns her eyes to the lens.
At about 8.5 seconds the camera orbits her fast: she hits a series of sharp, powerful dance moves exactly on the beat, arms snapping into clean lines, each hit leaving a sharp straight streak of light, and her final hit writes the bold glowing word "FAST" across the air beside her.
At about 9.8 seconds, hard cut to a medium side angle: she does a slow-motion hair flip, her hair leaving soft glowing trails that hang in the air, her gaze lost in the distance.
At about 11 seconds, four rapid hard cuts on four beats, each frozen for an instant like a long-exposure photo with her light trails frozen around her: a close-up of her eyes looking into the lens, a high-angle shot of her framing her face with both hands, a profile shot of her gazing into the distance, and a low-angle full-body shot of her in a sharp model pose.
At about 12.3 seconds she does her signature move once more, bigger and sharper, as the camera whips around her in a fast half circle and the light trails burst outward.
At about 13.3 seconds the camera pushes in from the front: she rests one hand on her hip, glances away into the distance, then turns back to the lens with a cool, half-lidded stare; on the final beat a faint, smug half-smile appears while her eyes stay half-lidded, the glowing word "YOU" shines above her head, and all the light trails fade away into the still opening image, so the ending loops smoothly back to the start.
All on-screen text is only the English words described above, spelled exactly, each word appearing alone as one clean, sharp word on a single line, never overlapping other letters, never split, and never covering her face. She never speaks or sings; there is no dialogue. If she wears a skirt or dress, she always wears plain black very short leggings underneath, so her underwear is never visible; the video stays wholesome and stylish, with modest framing, and the camera never looks up her skirt or lingers on her chest or legs. Her body proportions, arms and legs always stay correct during fast motion; her hands always have five fingers with correct joints. Whenever she turns or spins, her head, shoulders, hips and feet always turn together as one body; her head never stays still while her body rotates, and in overhead shots she never spins. Whenever she holds a pose, her hands are relaxed and clearly drawn, each with exactly five fingers, never six, and never thrust toward the lens. No other people appear. No mirrors and no reflections of her. No subtitles, no logos, no watermark. Everything stays in the illustration style of the reference picture, never photorealistic.
overall_soundscape: Silence on the still image at the start, a tiny bright chime as the point of light appears, a fast rising whoosh as it circles her, a huge shimmering burst at 0.8 seconds, airy electric swooshes that follow every light trail, sharp sparkling zaps on every hard hit, a sudden hush when everything goes dark at 6.5 seconds, her footsteps and the swish of her hair and outfit, the whoosh of the whip pans and crash zooms, and a deep airy whoosh during the slow-motion hair flip.
non_diegetic_music: A fast, driving, catchy melodic drum-and-bass track at about 170 BPM: a deep, heavy rolling reese bass, a punchy breakbeat with a hard snare, rapid crisp hi-hats, dreamy rainy-night pads and a cool, wistful synth hook. The track is silent at the start, a sparkling riser swirls with the circle of light, and the beat drops exactly at 0.8 seconds when the light explodes; the groove stays punchy with a hit on every cut, the drums cut out for the one dark beat at 6.5 seconds and crash back in with the light, opens up into a short airy breakdown at 9.8 seconds with the sub bass still rumbling underneath, gives four sharp stabs on the four snapshots at 11 seconds, comes back at full power for her second signature move, and ends with a hard final hit and a soft shimmering tail as the light fades.
使用建议
先上传一张参考图作为起始画面,再按时间点逐段修改舞蹈动作和镜头。