VdoBloom
← Back

High-Intensity Experimental Music Video

Music Videoby @viva · source ↗August 13, 2026

Full prompt

@Image1=MAINIMG @Image2=PIPIMG MAIN="MAINIMG unique subject"; MAIN_FACE="Main screen character facial identity"; MAIN_HAIR="Main screen character hairstyle base"; MAIN_BODY="Main screen character body proportions"; MAIN_C0="Main screen character original outfit base"; MAIN_ACC="Main screen character key accessory logic" PIPCHAR="PIPIMG unique subject"; PIP_FACE="Lower left character facial identity"; PIP_HAIR="Lower left character hairstyle base"; PIP_BODY="Lower left character body proportions"; PIP_C0="Lower left character original outfit base"; PIP_ACC="Lower left character key identification logic" WALL="Background wall layer"; FLOOR="Floor layer"; CHAR="Main screen character layer"; PIP="Lower left observation window layer" subject: "Read MAINIMG and PIPIMG only once and lock the two subjects respectively. MAIN only belongs to the main screen character; PIPCHAR only belongs to the lower left PIP window character. Both are independent characters, no confusion, no premature swapping, no mistaking PIPCHAR for MAIN, and no mistaking MAIN for PIPCHAR. For most of the video, Main screen = MAIN, lower left PIP = PIPCHAR, with a clear position swap only in the final high-speed transition." vocal: "Use the same original Japanese female lead vocal throughout; lyrics are not pre-written, freely improvised Japanese short phrases by the generation side; clear Japanese articulation and lip-sync; strong rhythmic hooks + short phrases + sustained notes; no quoting existing songs, no reading prompt words, no ah/oh/woo loops. No male lead vocals, no male-female switching, no changing singer due to character swaps." music: "150BPM intense Japanese dance music/Electronic House/High-speed Club; huge four-beat kick, tight snare/claps, rolling low frequency, sidechain synth bass, bright supersaw chord wall, metallic percussion, vocal chops, arpeggios, riser effects, impact. No slow intro. 0-13s each segment has clear downbeats and transition feel; 13-15s enter high-density chopped finale, and complete MAIN and PIPCHAR position swap at the end. The lead vocal remains the same female voice, no changing to male, no changing voice identity due to role swap." VOCAL_IDENTITY_LOCK: "0.00-15.00s always use the same original Japanese female lead vocal. Sound is an independent audio layer, does not change with visual character identity, appearance, gender performance, PIP content, or final role swap. Even if PIPCHAR enters the main screen, the female lead vocal must be completely continuous. No male lead, no male-female switching, no dual lead, no changing singer due to character swap." priority: "1 Main screen MAIN recognition established, 2 lower left PIPCHAR recognition established, 3 same female lead vocal throughout, 4 main screen continuous Japanese singing/dancing, 5 main screen sync 3-layer transition, 6 lower left PIP stability, 7 CHAR description change clearly visible, 8 camera, 9 final swap successful." MAIN_IDENTITY_DEFINITION: "MAIN consistency refers to character identity recognition, not identical pixel-level appearance. The goal is not a pixel-perfect copy of the source image, but for the audience to recognize it is the same MAIN regardless of style changes." MAIN_IDENTITY_LOCK: "MAIN fixed attributes: facial bone structure and feature relative positions; eye shape and expression tone; front hair structure and hairstyle base; head-to-body ratio, physique, body proportions; sense of age and overall temperament; key accessory logic. These form the core of MAIN recognition and must be maintained." MAIN_MUTABLE_ATTRIBUTES: "MAIN attributes allowed to change drastically: drawing style and medium; line language; coloring and shading; material representation; 2D/3D/Print/Craft processing; clothing color; clothing material; clothing details and decorations; stage-style modification of the same clothing base; local accessory enhancement or weakening; lighting effects and outline emphasis; degree of abstraction. As long as fixed attributes remain, these changes are valid." MAIN_FORBIDDEN_DRIFT: "MAIN Forbidden: turning into someone else; facial/feature reconstruction; hairstyle base disappearing; physique, age, or gender drifting significantly; unfamiliar clothing completely detached from original identity; key accessory logic disappearing." PIP_IDENTITY_DEFINITION: "PIPCHAR must also maintain consistent character identity. No matter how the angle, movement, or eating state changes in the lower left PIP, it must always be seen as the same PIPCHAR." PIP_IDENTITY_LOCK: "PIPCHAR fixed attributes: facial bone structure and feature relative positions; hairstyle base; body proportions; sense of age and temperament; clothing base and key identification points. The character in the lower left PIP window must always be PIPCHAR herself." identity_repeat: "MAIN and PIPCHAR must always be clearly distinguished. 0-13s main screen = MAIN, lower left PIP = PIPCHAR. Only the final swap segment allows swapping positions. The audio layer does not participate in this swap." MAIN_SCREEN_MODEL: "The main screen consists of three synchronized layers: WALL = back wall/facade; FLOOR = floor/stage floor/ground; CHAR = MAIN character body. The main screen uses a synchronized switching logic where WALL/FLOOR/CHAR enter new segments together at each major transition point." SYNC_RULE: "The main screen uses synchronized three-layer transitions, not complex asynchronous ones. Goal is execution priority, with each segment clearly changing scene." CHAR_RULE: "Main screen CHAR only refers to MAIN body. All main screen character changes only apply to MAIN: face/head/hair/neck/torso/arms/hands/waist/legs/feet/clothing/accessories. Must still be seen as the same MAIN after change." CHAR_TRANSFORMATION_REQUIRED: "Each MAIN description change must be immediately visible. Just changing color temperature, brightness, slight filters, or weak textures does not count as a change. Each time at least two of the following must change: line language, coloring, material, medium, clothing color, or clothing details." WALL_RULE: "WALL is only responsible for background walls/screens/posters/displays/installations. Mysterious large faces, huge eyes, abstract portraits, posters, screens, graphic character impressions, and huge MAIN close-ups are allowed as wall art, but only as wall images, not independent characters." FLOOR_RULE: "FLOOR is only for the ground. Allows huge facial images, huge eyes, abstract portrait LED floors, and fire/water/grass/neon grid/liquid metal/flower field/ice/prism/pixel ground/metaverse data platforms. These must stay on the ground and not become independent characters or change the MAIN body surface." BACKGROUND_EXTRA_RULE: "WALL and FLOOR can absorb two environmental languages: 1 Metaverse/virtual city/holographic data architecture/transparent UI space; 2 Pixel art/8-bit/16-bit/blocky game landscapes. These only apply to the environment layer and must not change MAIN identity." goal: "15-second high-density experimental MV. 0-13s main screen is a synchronized 3-layer MV of MAIN, lower left PIP is a strange additional video of PIPCHAR eating a burger on a black background. 13-15s final 2 seconds enter high-speed finale transition, finally completing the position swap between MAIN and PIPCHAR characters, but the lead vocal always remains the same female voice." dance: "MAIN performs high-energy club choreography at 150BPM: heel-toe fast steps, fast footwork, side shifts, hip beats, shoulder hits, chest pops, arm waves, locking, body rolls, diagonal moves, half-turns, rotations, bounces, high/low position changes, short jumps, rhythmic head turns. MAIN continues clear Japanese singing and dancing from 0.00-15.00s, never standing still, staying center stage until the final swap." dance_timeline: "0-1.5 heel-toe advance+shoulder hit+arm wave | 1.5-3 diagonal move+chest pop+half-turn bounce | 3-4.5 low side step+upward arm sweep+small spin | 4.5-6 fast two-step+torso twist+backward bounce | 6-7.5 side step+arm lock+forward rhythm | 7.5-9 high/low change+fast turn+hip beat | 9-10.5 running step+controlled spin+arm sweep | 10.5-12 reverse step+body roll+jump landing | 12-13 fast cross step+270 turn+move end | 13-13.5 high-speed A | 13.5-14 high-speed B | 14-14.5 high-speed C | 14.5-15 ultimate burst" MAIN_AUTONOMOUS_STYLE: "When MAIN description changes, the generation side must autonomously invent a new visual style distinctly different from the current one. Don't follow a fixed list. Freely change line language, color systems, textures, 2D/3D, print methods, animation media, craft feel, abstraction, clothing colors, and stage details. Goal is for the audience to see ‘the same MAIN described by another visual culture’." MAIN_STYLE_RANGE: "Directions include but not limited to: comic line drawing, high-contrast American graphic, pop art, retro hand-drawn animation, sticker-style, collectible model, plush, clay, paper-cut, collage, picture book, fashion illustration, 3D cel-shaded, paper pulp craft, pixelated, low-poly, mosaic, stained glass, neon, holographic, blueprint, graffiti, Art Deco, off-register print, silkscreen, etc." COSTUME_RULE: "MAIN clothing uses ‘fixed base, variable expression’ principle. Must keep original silhouette, core identification structure, and wearing logic. Changes allowed: colors, materials, decorations, stage enhancements, pattern density, sheen/reflection, layer additions/subtractions. Forbidden: completely switching to a stranger's clothing system or unrelated identity." COSTUME_CHANGE_REQUIRED: "MAIN clothing must have immediately visible stage changes. After 3s, each stage must clearly change color system, material, or details; can't just rely on lighting/temp/filters. Even if visual medium changes, the clothing itself must independently change." costume_vector: 'C0=original; C1=keep silhouette, change primary colors to bright stage colors, add highlight fabrics and decoration; C2=keep base, change to strong graphic two/three-tone blocks, add geometric borders; C3=keep base, change to deep black/dark main + high contrast bright borders, sharpen decorations; C4=keep base, change to pearlescent/iridescent/translucent future materials, different colors from C3; C5=clear white-gold finale outfit.' COSTUME_VISIBILITY_RULE: 'C1, C2, C3, C4, C5 must be clearly different. Failure to judge stage by clothing is considered a clothing change failure.' costume_timeline: '0-3 C0 | 3-6 C1 | 6-9 C2 | 9-12 C3 | 12-13 C4 | 13-15 C5 finale leaning' MAIN_STAGE_TRANSITIONS: 'Main screen 3-layer sync transitions advance by segment. WALL/FLOOR/CHAR enter a new overall segment together at each switch.' main_stage_timeline: '0-1.5 Stage1: WALL=pop art wall+abstract graphics+high contrast colors; FLOOR=huge stranger face LED floor; CHAR=close to original high-detail, light stage mod; Mood=Opening grab. 1.5-3 Stage2: WALL=poster collage/ad visual/torn edges; FLOOR=huge eyes/iris dynamic floor; CHAR=clear comic line or high contrast graphic; Mood=First major style change. 3-4.5 Stage3: WALL=folding screen style/gold ground/traditional painting; FLOOR=fire surface/heatwaves/cracks; CHAR=retro hand-drawn animation or picture book style; Mood=Traditional+Heat. 4.5-6 Stage4: WALL=large display wall/CRT matrix/projection; FLOOR=transparent water/ripples/mirror; CHAR=sticker or graphic poster style; Mood=Digital+Fluid. 6-7.5 Stage5: WALL=MAIN huge close-up face art; FLOOR=real grassland/blowing grass; CHAR=collectible model or 3D cel-shaded; Mood=Strongest MAIN identity emphasis. 7.5-9 Stage6: WALL=metaverse city/holographic buildings/UI/data arch; FLOOR=metaverse electronic platform/grid; CHAR=holographic/neon/future stage; Mood=Tech Nightclub. 9-10.5 Stage7: WALL=large pixel art wall/8-bit city; FLOOR=pixel ground/bricks/water; CHAR=pixel-leaning or low-poly but clearly MAIN; Mood=Gaming electronic feel. 10.5-12 Stage8: WALL=mysterious giant face/abstract portrait; FLOOR=liquid metal surface; CHAR=mosaic/stained glass/craft; Mood=Heterogeneous art. 12-13 Stage9: WALL=graffiti/misprint/neon sign mix; FLOOR=flower field to ice prism; CHAR=mixed media, finale outfit push; Mood=Finale build-up. 13-13.5 Burst A: high-frequency sync switch, compress space, enhance speed. 13.5-14 Burst B: continue sync switch, prepare for PIP intrusion. 14-14.5 Burst C: boundary destruction, screen erosion, swap starts. 14.5-15 Finale: complete MAIN and PIPCHAR swap, main screen subject becomes PIPCHAR, MAIN enters lower left PIP.' camera_lock: 'Main screen camera must move significantly in 3D space, not fixed station panning/tilt/digital zoom. Each segment must change X/Y/Z coordinates with visible parallax. No stationary shaking to fake orbiting, no digital zoom only.' orbit_rule: 'Each orbit must complete at least 90-degree change, focus segments 120-180. Clockwise/Counter-clockwise must be clear. Vertical orbits must rise from low to high over the subject.' frame_rotation: 'Screen rotation must be continuous 30-90 degrees or more, overlapping with orbit/push/pull, not random shaking.' camera_timeline: '0-1.5 front-left low clockwise 120 to right, knee to chest height | 1.5-3 continue clockwise 90 to right-back, high speed graze past shoulder to re-catch face | 3-4.5 back-low vertical orbit, over head to high-angle then descend forward | 4.5-6 front-right low counter-clockwise 150 to left-back, radius shrinks then expands | 6-7.5 high-angle counter-clockwise fast descend to ground-level graze | 7.5-9 ground sprint: dash from far to MAIN, sharp pull up at feet, over shoulder then 90 degree reverse orbit | 9-10.5 giant spiral: 120 orbit + rise + radius contraction + 60 roll | 10.5-12 super-high dive past shoulder to back then 180 flip back to front | 12-13 ultra-close to ultra-wide reverse flyover + huge WALL/FLOOR then high-speed push-in | 13-13.5 fast clockwise 90 orbit + 45 roll | 13.5-14 fast counter-clockwise 120 orbit + dive | 14-14.5 spiral 120 + 90 screen roll + screen tear | 14.5-15 complete swap from broken screen and relock new protagonist' PIP_RULE: 'Add a small PIP window in the lower left. Not a split screen, but an independent small PiP, approx 1/16 area (width/height 22-25%). Keep it small and clear, don't interfere with main screen. Fixed position, can have white line/glow border.' PIP_BACKGROUND: 'PIP environment fixed to pure black or dark void. No complex background or floor patterns. Goal is clear view of PIPCHAR eating a burger.' PIP_SUBJECT: 'Display PIPCHAR in the PIP window, not MAIN. This is an independent curious video. For 0-13s, PIP character must be PIPCHAR.' PIP_PURPOSE: 'Show PIPCHAR eating a burger from various angles on a black background. Atmosphere slightly comical/mysterious, like a random insert, but visually clear and cute.' PIP_ACTION_CORE: 'PIPCHAR holds a burger or thick meat patty and clearly eats: looking at it, lifting, biting, chewing, swallowing, biting again. Must clearly see eating, not just holding.' PIP_HAMBURG_RULE: 'Burger must be clearly visible (bun, patty, toppings). Key point is PIPCHAR seriously eating. No other food, no eating air, no just posing.' PIP_ANGLE_RULE: 'PIP shows PIPCHAR with eating action. Can slowly turn, rotate, or small continuous changes while keeping eating visible. Show front, 45, side, back-turn, half-body close-up, etc.' PIP_MOTION_STYLE: 'Minor movements only: turning, head tilting, lifting burger, biting, chewing. No major dancing. Focus is multi-angle continuous eating.' PIP_AUDIO_RULE: 'PIPCHAR does not sing or provide vocals. Does not take over song. Performs silent or low-presence eating. No impact on the female lead vocal.' PIP_TIMELINE: '0.8-2.8 front and left 45 eating, first clear bite | 2.8-4.8 left side and left-back 45, chewing while turning | 4.8-6.8 back and right-back 45, looking back while holding burger | 6.8-8.8 right side and right 45, another clear bite | 8.8-10.8 upper body close-up turn, highlight face and burger, clear chewing | 10.8-12.8 small continuous turntable-style observations, different angles continuous eating | 12.8-13.8 faster eating, sensing main screen abnormality | 13.8-15.0 PIP boundary sucked into main screen and expanded, PIPCHAR bursts out of lower left window to become main screen character.' PIP_STYLE_RELATION: 'Even if main screen changes are intense, prioritize PIPCHAR identification and eating visibility. Style can follow finale atmosphere but black background and burger are core.' PIP_MAIN_SEPARATION: '0-13.8s MAIN continues dancing/singing in main screen. PIP is just an additional layer. Both exist simultaneously without PIP interfering with MAIN facial visibility.' SWAP_RULE: 'Final swap is the core experimental event. 13.8-15.0 a clear position swap must occur: PIPCHAR breaks PIP boundary, expands via high-speed transition to main center; MAIN is compressed, shrunk, pulled into the lower left PIP window. After swap, main subject = PIPCHAR, lower left PIP = MAIN. Vocal identity remains unchanged.' SWAP_VISUAL_GRAMMAR: 'Use screen tearing, mirror flips, PiP expansion, window burst, layer penetration, character through-screen, rotate/scale, glitch, etc. Swap is not a simple cut but a visual spectacle.' SWAP_FINAL_STATE: 'By 14.7-15.0, new main subject must be PIPCHAR (optionally with burger); MAIN must be in the PIP window. Swap must be completed. Female lead vocal remains unchanged.' PIP_FAILURE: 'Failed if PIP window is too large, not black background, not PIPCHAR, no eating, or no final swap.' typo: 'Minimal dynamic text, single kanji or short English only. Candidates: Dance, Light, Sound, Instant / GO, UP'

Original prompt

@Image1=MAINIMG @Image2=PIPIMG MAIN="MAINIMG唯一主体";MAIN_FACE="主画面人物脸部身份";MAIN_HAIR="主画面人物发型母体";MAIN_BODY="主画面人物身体比例";MAIN_C0="主画面人物原始服装母体";MAIN_ACC="主画面人物关键饰品逻辑" PIPCHAR="PIPIMG唯一主体";PIP_FACE="左下人物脸部身份";PIP_HAIR="左下人物发型母体";PIP_BODY="左下人物身体比例";PIP_C0="左下人物原始服装母体";PIP_ACC="左下人物关键识别逻辑" WALL="背景墙层";FLOOR="地面层";CHAR="主画面角色层";PIP="左下观察窗层" subject:"仅读取MAINIMG与PIPIMG各一次并分别锁定两个主体.MAIN只属于主画面人物;PIPCHAR只属于左下PIP窗口人物.两者都是独立人物,禁止混淆,禁止提前互换,禁止把PIPCHAR误当MAIN,也禁止把MAIN误当PIPCHAR.全片绝大部分时间主画面人物=MAIN,左下PIP人物=PIPCHAR,仅在最后高速转场中执行明确位置交换." vocal:"全片使用同一个原创日语女性主唱声线;歌词不预写,由生成侧自由即兴原创日语短句;清晰日语咬字与口型同步;强节奏钩子+短句+延音;禁止引用现有歌曲,禁止读提示词,禁止ah/oh/woo循环.禁止男性主唱,禁止男女切换,禁止因画面角色交换而改变歌手." music:"150BPM极强日系舞曲/电子浩室/高速俱乐部;巨大四拍踢鼓,紧凑军鼓/拍手,滚动低频,侧链合成贝斯,明亮超锯齿和弦墙,金属打击,人声切片,琶音,上升音效,冲击.无慢前奏.0-13s每段有明确重拍与切换感;13-15s进入高密度切碎终章,并在最后完成MAIN与PIPCHAR位置交换.全片主唱保持同一个女性声线,不得变男性,不得因角色交换改变声线身份." VOCAL_IDENTITY_LOCK:"0.00-15.00s始终使用同一个原创日语女性主唱声线.声音是独立音频层,不随画面人物身份,外观,性别表现,PIP内容或最终角色交换而改变.即使PIPCHAR进入主画面,女性主唱声线也必须完全连续.禁止男性主唱,禁止男女声切换,禁止双主唱化,禁止角色交换导致歌手身份变化." priority:"1主画面MAIN识别成立 2左下PIPCHAR识别成立 3全片女性主唱固定 4主画面持续日语唱跳 5主画面同步三层转场 6左下PIP稳定存在 7CHAR描写变化清晰可见 8镜头 9最终交换成功." MAIN_IDENTITY_DEFINITION:"MAIN的一致性指人物识别一致,不是外观逐项完全一致.目标不是像素级复制原图,而是无论主画面风格如何变化,观众始终一眼认出是同一个MAIN." MAIN_IDENTITY_LOCK:"MAIN固定属性:脸部骨相与五官相对关系;眼睛形状与神情基调;前发结构与发型母体;头身比,体型,身体比例;年龄感与整体气质;关键识别饰品存在逻辑.这些构成MAIN识别核心,必须持续成立." MAIN_MUTABLE_ATTRIBUTES:"MAIN允许大胆变化的属性:描画风格与媒介;线条语言;上色方式与阴影方式;材质表现;二维/三维/印刷/工艺化处理;服装颜色;服装材质;服装细节与装饰;同一服装母体上的舞台化改造;局部配饰强化或弱化;光效与轮廓强调方式;抽象程度.只要MAIN固定属性仍成立,以上变化全部合法." MAIN_FORBIDDEN_DRIFT:"MAIN禁止:变成另一个人;脸型与五官关系明显重构;发型母体消失;体型,年龄感,性别感大幅漂移;完全脱离原角色识别性的陌生服装体系;关键识别饰品逻辑彻底消失." PIP_IDENTITY_DEFINITION:"PIPCHAR也必须保持同一人物识别.左下PIP里无论角度,动作,吃东西状态如何变化,都必须始终被看作同一个PIPCHAR." PIP_IDENTITY_LOCK:"PIPCHAR固定属性:脸部骨相与五官相对关系;发型母体;身体比例;年龄感与整体气质;服装母体与关键识别点.左下PIP窗口中的人物必须始终是PIPCHAR本人." identity_repeat:"MAIN与PIPCHAR必须始终区分清楚.0-13s绝大部分时间主画面=MAIN,左下PIP=PIPCHAR.只有最终交换段才允许彼此替换位置.声音层不参与这个交换." MAIN_SCREEN_MODEL:"主画面由三个同步变化层组成:WALL=背景墙/立面/后景装置;FLOOR=地面/舞台地板/地表;CHAR=MAIN人物本体.主画面采用同步切换逻辑,每个主要切换点WALL/FLOOR/CHAR一起进入新段落,形成清晰高执行率的整体转场." SYNC_RULE:"主画面采用同步三层转场,不追求复杂异步.目标是执行率优先,每段整体变景清楚有力." CHAR_RULE:"主画面CHAR只指MAIN本体.所有主画面角色变化只作用于MAIN:脸/头/发/颈/躯干/手臂/手/腰/腿/脚/服装/饰品.变化后仍必须被看作同一个MAIN." CHAR_TRANSFORMATION_REQUIRED:"每次MAIN描写变化都必须肉眼立即可辨.只改变色温,亮度,轻微滤镜或弱纹理不算变化.每次至少明显改变以下项目中的两项:线条语言;上色方式;材质表现;媒介表现;服装颜色或服装细节." WALL_RULE:"WALL只负责巨大背景墙面/屏风/海报墙/显示墙/装置艺术墙.允许神秘人物大脸,巨大眼睛,抽象肖像,人物海报,屏幕中的脸,图形化人物印象,以及MAIN自身超大特写脸作为墙面艺术,但都只是墙面图像,不是独立角色." FLOOR_RULE:"FLOOR只负责地面层.允许巨大人物脸部影像,巨大眼睛显示,抽象肖像LED地板,以及火焰/水面/草原/霓虹网格/液态金属/花田/冰面/棱镜/像素地面/元宇宙数据地台等地表.这些只能停留在地面,不得立起成角色,不得改变MAIN身体或服装表面." BACKGROUND_EXTRA_RULE:"WALL与FLOOR可吸收两类环境语言:1元宇宙/虚拟都市/全息数据建筑/透明UI空间;2像素艺术/8-bit/16-bit/块状游戏地景.这两类只作用于环境层,不得改变MAIN身份." goal:"15秒高密度实验MV.0-13s主画面是MAIN的同步型三层MV,左下PIP是PIPCHAR黑背景吃汉堡的奇妙附加视频.13-15s最后2秒进入高速终章转场,最终完成主画面与左下PIP人物位置交换,但主唱始终保持同一个女性声线." dance:"MAIN在主画面执行150BPM高能俱乐部编舞:脚跟脚尖快速步,快速脚步,侧移,髋部重拍,肩击,胸部弹动,手臂波浪,锁舞,身体滚动,斜向移动,半转,旋转,反弹,高低位变化,短跳,节奏性转头.MAIN从0.00-15.00s持续明确日语唱跳,绝不站定展示,直到最终交换前仍是主舞台中心." dance_timeline:"0-1.5脚跟脚尖前进+肩击+手臂波浪|1.5-3斜向移动+胸部弹动+半转反弹|3-4.5低位侧步+上扬扫臂+小旋转|4.5-6快速两步+躯干扭转+后退反弹|6-7.5侧向脚步+手臂锁定+前进律动|7.5-9高低位变化+快速转向+髋部重拍|9-10.5跑步式舞步+控制旋转+扫臂|10.5-12反向步+身体滚动+跳跃落地|12-13快速交叉步+270°转身+移动收束|13-13.5高速段A|13.5-14高速段B|14-14.5高速段C|14.5-15终极爆发" MAIN_AUTONOMOUS_STYLE:"每次MAIN描写变化时,生成侧必须自主发明一种与当前描写明显不同的新视觉表现方式.不要按固定列表死板执行.可自由改变线条语言,色彩系统,材质感,二维/三维表现,印刷方式,动画媒介,实体工艺感,抽象程度,服装配色与舞台化细节.目标是让观众明确看到‘同一个MAIN被另一种视觉文化重新描写’." MAIN_STYLE_RANGE:"可联想到但不必固定命名的方向包括:漫画线描,高对比美式平面,流行平面艺术,复古手绘动画,贴纸化,收藏模型化,毛绒化,黏土化,剪纸化,拼贴化,绘本化,时装插画化,三维赛璐璐化,纸浆工艺化,像素化,低模几何化,马赛克化,彩窗化,霓虹化,全息化,蓝图化,喷绘涂鸦化,装饰艺术化,错版印刷化,丝网印刷化等." COSTUME_RULE:"MAIN服装采用‘母体固定,表现可变’原则.必须保留原服装大轮廓,核心识别结构与整体穿着逻辑.允许变化:配色;材质;局部装饰;舞台化强化;图案密度;光泽与反射;层次增减.禁止变化:完全换成陌生角色服装体系;与原角色无关的全新身份服." COSTUME_CHANGE_REQUIRED:"MAIN服装必须发生肉眼立即可辨的阶段变化.3s以后每个阶段必须明显改变颜色系统,材质或服装细节,不能只靠灯光,色温,滤镜或整体风格冒充换装.即使MAIN视觉媒介同时变化,服装本身也必须独立发生可见变化." costume_vector:"C0=原始状态;C1=保留原服装轮廓,主配色明显改为鲜艳舞台色,加入高光面料与明确装饰强化;C2=保留母体,明显改为强烈图形双色/三色色块,增加几何边饰;C3=保留母体,明显改为深黑或深色主体+高对比亮色边线,装饰锐利化;C4=保留母体,明显改为珠光/虹彩/半透明未来材质,配色与C3明显不同;C5=明确白金配色终章服装." COSTUME_VISIBILITY_RULE:"C1,C2,C3,C4,C5彼此必须肉眼明显不同.如果观众无法仅看服装判断已经进入下一阶段,视为换装失败." costume_timeline:"0-3 C0|3-6 C1|6-9 C2|9-12 C3|12-13 C4|13-15 C5终章倾向" MAIN_STAGE_TRANSITIONS:"主画面同步三层转场按段落推进.每次切换时,WALL/FLOOR/CHAR共同进入新的整体段落." main_stage_timeline:"0-1.5 Stage1:WALL=流行平面艺术墙+抽象图形+高对比色块;FLOOR=巨大陌生人物脸部特写LED地板;CHAR=接近原始高细节风格,轻度舞台化;情绪=开场抓取 1.5-3 Stage2:WALL=海报拼贴墙/广告视觉/撕裂边缘;FLOOR=巨大眼睛与虹膜动态显示地板;CHAR=明显漫画线描或高对比平面化;情绪=第一次明显风格变化 3-4.5 Stage3:WALL=屏风绘式墙面/金地/折屏结构/传统绘画印象;FLOOR=火焰地表/热浪/燃烧裂隙;CHAR=复古手绘动画感或绘本化;情绪=传统+热烈 4.5-6 Stage4:WALL=大型显示墙/CRT矩阵/投影墙/数字屏;FLOOR=透明水面/涟漪/镜面反射;CHAR=贴纸化或图形海报化;情绪=数字+流体 6-7.5 Stage5:WALL=MAIN自身超大特写脸墙面艺术,放大眼部/唇部/局部发丝;FLOOR=真实草原/风吹草浪;CHAR=收藏模型化或三维赛璐璐化;情绪=MAIN身份最强强调段 7.5-9 Stage6:WALL=元宇宙虚拟都市墙景/全息高楼/透明UI/数据拱门;FLOOR=元宇宙电子地台/高亮网格/透明数据格;CHAR=全息化/霓虹化/未来舞台化;情绪=科技夜店 9-10.5 Stage7:WALL=大型像素艺术背景墙/8-bit与16-bit混合城市;FLOOR=像素地面/8-bit砖块/像素草地/像素水面;CHAR=像素化倾向或低模几何化,但仍清楚是MAIN;情绪=游戏化电子感 10.5-12 Stage8:WALL=神秘巨大人脸墙面/抽象肖像墙/波形屏;FLOOR=液态金属波面;CHAR=马赛克/彩窗/装饰工艺感处理;情绪=异质艺术感 12-13 Stage9:WALL=喷绘涂鸦/错版印刷/霓虹标牌混合墙;FLOOR=花田向冰面棱镜渐变;CHAR=混合媒介,服装终章化前冲;情绪=终章累积 13-13.5爆发A:主画面高频同步切入新状态,压缩空间,强化速度 13.5-14爆发B:主画面继续同步切换,准备与PIP发生位置入侵 14-14.5爆发C:主画面与PIP发生边界破坏,画面互相侵蚀,开始交换 14.5-15终章:完成MAIN与PIPCHAR位置交换,主画面主体变为PIPCHAR,MAIN进入左下PIP" camera_lock:"主画面镜头必须在真实三维空间中大幅移动,不是固定机位平移/俯仰/数字变焦/抖动.每个镜头段必须改变世界坐标X/Y/Z并产生明显视差.禁止原地晃动冒充环绕,禁止仅数字变焦冒充推进." orbit_rule:"每次环绕至少完成90°明显方位变化,重点段120-180°.顺时针与逆时针必须肉眼可辨.垂直环绕必须从低位沿主体真实升高,越过头顶到高位俯拍." frame_rotation:"整画面旋转必须是明确连续镜头滚转30-90°或更大,并与环绕/推进/升降叠加,不是随机抖动." camera_timeline:"0-1.5左前低位顺时针120°环绕到右侧,膝高升至胸高|1.5-3继续顺时针90°到右后方,高速贴肩掠过后重捕脸|3-4.5背后低位启动垂直环绕,越过头顶俯拍后向前下降|4.5-6右前低位逆时针150°环绕到左后方,半径先缩小再扩大|6-7.5高位逆时针环绕同时快速下降至贴地并掠过|7.5-9贴地冲刺:远处贴地冲向MAIN,脚边急拉升,越肩后反向环绕90°|9-10.5巨大螺旋:120°环绕+明显升高+半径收缩+60°滚转|10.5-12超高位俯冲擦过肩侧,穿到背后后立刻180°翻向回正面|12-13极近脸开始极速倒飞至超广角全身+巨大WALL/FLOOR,随后高速推近|13-13.5顺时针快速环绕90°+45°滚转|13.5-14逆时针快速环绕120°+高空俯冲|14-14.5螺旋120°+90°整画面滚转+画面撕裂|14.5-15从主画面破裂中完成位置交换后重新锁定新主角" PIP_RULE:"在主画面左下加入小型观察PIP窗口.不是四分屏,不是大分割,而是独立小型画中画.面积约全画面1/16,建议宽22-25%,高22-25%,保持小而清楚,不要过大,不要明显干扰主画面.位置固定左下,可有细白线或淡发光边框." PIP_BACKGROUND:"PIP内部环境固定纯黑空间或极暗黑色虚空.不要复杂背景,不要环境变化,不要墙面艺术,不要地面花样,不要额外特效堆满.目标是清楚观看PIPCHAR吃汉堡." PIP_SUBJECT:"PIP窗口中显示PIPCHAR,不是MAIN.这是独立的奇妙附加视频.0-13s大部分时间左下PIP人物都必须是PIPCHAR." PIP_PURPOSE:"PIP窗口展示PIPCHAR在黑背景中从多个角度吃汉堡.气氛略微滑稽,略微神秘,像莫名其妙插入的无关短视频,但视觉上必须清楚,可爱,令人困惑." PIP_ACTION_CORE:"PIPCHAR手持或拿近一个完整汉堡或厚实肉排汉堡,明确执行吃汉堡动作:看汉堡,举起,张口咬下,咀嚼,吞咽,再咬,轻微满足表情或面无表情继续吃.必须明显看出在吃汉堡,不是拿着不动." PIP_HAMBURG_RULE:"汉堡必须清晰可见,面包,肉排,配料可辨识.左下PIP核心笑点就是PIPCHAR认真吃汉堡.禁止变成别的食物,禁止吃空气,禁止只摆姿势." PIP_ANGLE_RULE:"PIP中的PIPCHAR以角色展示+吃汉堡结合方式呈现.可缓慢转身,转头,半身转体,轻微姿势切换,或小幅连续转台观察.同时保持吃汉堡动作可见.可展示正面,左45°,侧面,背后回头,右45°,半身近景等." PIP_MOTION_STYLE:"PIP中的PIPCHAR可轻微动作:轻转身,轻摆头,抬手拿汉堡,咬一口,咀嚼,吞咽,再次进食.不要大幅舞蹈.重点是多角度+持续吃汉堡." PIP_AUDIO_RULE:"PIPCHAR不唱歌,不提供独立主唱,不接管歌曲声线.PIPCHAR只执行无声或弱存在感的吃汉堡动作,可有咀嚼感但不形成主唱人声.左下PIP与最终位置交换不得影响主歌曲女性主唱." PIP_TIMELINE:"0.8-2.8正面与左45°吃汉堡,首次清楚咬下一口|2.8-4.8左侧面与左后45°,一边转身一边咀嚼|4.8-6.8背面与右后45°,回头露出脸并继续拿汉堡|6.8-8.8右侧面与右45°,再次明显咬一口|8.8-10.8上半身近景转体,突出脸部与汉堡,清楚咀嚼|10.8-12.8小型连续转台观察,不同角度持续吃|12.8-13.8吃汉堡动作加快,同时开始感应主画面异变|13.8-15.0PIP边界被主画面吸入并放大,PIPCHAR从左下窗冲出成为主画面角色" PIP_STYLE_RELATION:"如果主画面变化过于激烈,左下PIP仍应优先保证PIPCHAR识别清晰与吃汉堡动作清楚.PIP风格可适度跟随终章气氛,但黑背景和汉堡核心不能丢." PIP_MAIN_SEPARATION:"0-13.8s主画面MAIN继续正常日语唱跳与主镜头运动.左下PIP只是附加层.两者同时存在,但左下PIP不改变主画面背景逻辑,不抢主体,不遮挡MAIN主脸." SWAP_RULE:"最终交换是本次实验核心事件.13.8-15.0必须发生明确位置交换:PIPCHAR从左下PIP突破边界,经高速转场放大并进入主画面中心;MAIN被压缩,缩小,拉入左下PIP窗口或其原本位置;交换完成后主画面最终主体=PIPCHAR,左下PIP中的人物=MAIN;观众必须明确看出二者位置互换成功;但音频主唱身份绝不交换,全片主唱始终是同一个女性主唱声线." SWAP_VISUAL_GRAMMAR:"交换过程可使用屏幕撕裂,镜面翻转,画中画放大,窗口爆裂,层级穿透,角色穿屏,旋转缩放,颜色爆闪,故障扭曲,画面反转等高速转场手法.交换不是简单硬切,必须有明显‘主画面与PIP互换’的视觉奇观." SWAP_FINAL_STATE:"14.7-15.0时主画面中的新主体必须已经是PIPCHAR,并可带着汉堡或吃汉堡后的状态站上主舞台;同时MAIN必须已进入左下PIP,变成被观察的一方.位置交换必须完成,不要停留在半路.女性主唱声线仍保持不变." PIP_FAILURE:"若PIP窗口过大,不是黑背景,窗口里不是PIPCHAR,没有清楚吃汉堡,或最后没有与主画面完成角色位置交换,均视为失败." typo:"少量动态文字,仅单汉字或极短英字.候选:舞,光,音,瞬/GO,UP"
Generate this with Seedance 2 on VdoBloom.
Opens the image-to-video creator with this prompt and Seedance 2 pre-selected.
✨ Generate Video