1:"$Sreact.fragment"
3:I[54118,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1737","static/chunks/fe207bbb-c84ed91deca4f888.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6397","static/chunks/6397-b7ccd1edf8aedf12.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7979","static/chunks/7979-1799576a8e82a4f3.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","144","static/chunks/144-1a048554a0613a1b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2610","static/chunks/2610-c41f461a4ed7cf24.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","528","static/chunks/528-7ef2006e50a52549.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4077","static/chunks/4077-cba4e88bdb106467.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1924","static/chunks/1924-8c4ee98e9b0dbf83.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2530","static/chunks/2530-9eb218eca5ffc00b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7177","static/chunks/app/layout-75dbb92d3cda2f1c.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"Providers"]
4:I[98195,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1737","static/chunks/fe207bbb-c84ed91deca4f888.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6397","static/chunks/6397-b7ccd1edf8aedf12.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7979","static/chunks/7979-1799576a8e82a4f3.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","144","static/chunks/144-1a048554a0613a1b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2610","static/chunks/2610-c41f461a4ed7cf24.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","528","static/chunks/528-7ef2006e50a52549.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4077","static/chunks/4077-cba4e88bdb106467.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1924","static/chunks/1924-8c4ee98e9b0dbf83.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2530","static/chunks/2530-9eb218eca5ffc00b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7177","static/chunks/app/layout-75dbb92d3cda2f1c.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"GoogleTagManager"]
5:I[4411,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1737","static/chunks/fe207bbb-c84ed91deca4f888.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6397","static/chunks/6397-b7ccd1edf8aedf12.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7979","static/chunks/7979-1799576a8e82a4f3.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","144","static/chunks/144-1a048554a0613a1b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2610","static/chunks/2610-c41f461a4ed7cf24.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","528","static/chunks/528-7ef2006e50a52549.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4077","static/chunks/4077-cba4e88bdb106467.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1924","static/chunks/1924-8c4ee98e9b0dbf83.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2530","static/chunks/2530-9eb218eca5ffc00b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7177","static/chunks/app/layout-75dbb92d3cda2f1c.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"FilmAwardStrip"]
6:"$Sreact.suspense"
7:I[40459,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1737","static/chunks/fe207bbb-c84ed91deca4f888.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6397","static/chunks/6397-b7ccd1edf8aedf12.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7979","static/chunks/7979-1799576a8e82a4f3.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","144","static/chunks/144-1a048554a0613a1b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2610","static/chunks/2610-c41f461a4ed7cf24.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","528","static/chunks/528-7ef2006e50a52549.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4077","static/chunks/4077-cba4e88bdb106467.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1924","static/chunks/1924-8c4ee98e9b0dbf83.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2530","static/chunks/2530-9eb218eca5ffc00b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7177","static/chunks/app/layout-75dbb92d3cda2f1c.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"Navbar"]
8:I[15848,[],""]
9:I[55256,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","8039","static/chunks/app/error-332e154bd557e323.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"default"]
a:I[62142,[],""]
19:I[79302,["4219","static/chunks/app/global-error-f07d85f704128170.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"default"]
:HL["/_next/static/media/7d4881bb7e1bf84d-s.p.woff2","font",{"crossOrigin":"","type":"font/woff2"}]
:HL["/_next/static/css/c91c557421edbcc1.css?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","style"]
:HL["/_next/static/css/9c4f0c3be44ead33.css?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","style"]
2:T412,
(function () {
  window.dataLayer = window.dataLayer || [];
  function gtag() { window.dataLayer.push(arguments); }

  gtag('consent', 'default', {
    ad_storage: 'denied',
    ad_user_data: 'denied',
    ad_personalization: 'denied',
    analytics_storage: 'denied',
    functionality_storage: 'denied',
    personalization_storage: 'denied',
    security_storage: 'granted'
  });

  try {
    var match = document.cookie.match(/(?:^|;\s*)gdpr_consent=([^;]*)/);
    if (!match) return;
    var saved = JSON.parse(decodeURIComponent(match[1]));
    if (!saved || typeof saved !== 'object') return;
    gtag('consent', 'update', {
      analytics_storage: saved.analytics ? 'granted' : 'denied',
      ad_storage: saved.marketing ? 'granted' : 'denied',
      ad_user_data: saved.marketing ? 'granted' : 'denied',
      ad_personalization: saved.marketing ? 'granted' : 'denied',
      functionality_storage: saved.functional ? 'granted' : 'denied'
    });
  } catch (e) {
    /* malformed cookie â€” the denied default stands */
  }
})();
0:{"P":null,"b":"vX-6P7rGIK5v7u82yfIum","p":"","c":["","pixelflows","cinematic-fashion-b-roll-videos-seedance-2-5"],"i":false,"f":[[["",{"children":["pixelflows",{"children":[["slug","cinematic-fashion-b-roll-videos-seedance-2-5","d"],{"children":["__PAGE__",{}]}]}]},"$undefined","$undefined",true],["",["$","$1","c",{"children":[[["$","link","0",{"rel":"stylesheet","href":"/_next/static/css/c91c557421edbcc1.css?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","link","1",{"rel":"stylesheet","href":"/_next/static/css/9c4f0c3be44ead33.css?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}]],["$","html",null,{"lang":"en","suppressHydrationWarning":true,"className":"__variable_8c25df","children":[["$","head",null,{"children":[["$","script",null,{"id":"consent-mode-default","dangerouslySetInnerHTML":{"__html":"$2"}}],["$","script",null,{"id":"film-award-dismissed","dangerouslySetInnerHTML":{"__html":"try{if(document.cookie.split(';').some(function(c){return c.trim()===\"film_award_2027_dismissed=1\"}))document.documentElement.setAttribute(\"data-film-award-dismissed\",'')}catch(e){}"}}],["$","link",null,{"rel":"preconnect","href":"https://images.segmind.com"}],["$","link",null,{"rel":"preconnect","href":"https://segmind-resources.s3.amazonaws.com"}],["$","link",null,{"rel":"preconnect","href":"https://www.googletagmanager.com"}],["$","link",null,{"rel":"preload","href":"/fonts/Satoshi-Variable.woff2","as":"font","type":"font/woff2","crossOrigin":"anonymous"}]]}],["$","body",null,{"className":"antialiased bg-background text-foreground","children":["$","$L3",null,{"children":[["$","$L4",null,{"gtmId":"GTM-MHW23H4Z"}],["$","div",null,{"className":"min-h-screen flex flex-col","children":[["$","$L5",null,{}],["$","$6",null,{"fallback":["$","div",null,{"className":"h-16 bg-background/95 backdrop-blur border-b border-border"}],"children":["$","$L7",null,{}]}],["$","main",null,{"className":"flex-1","children":["$","$L8",null,{"parallelRouterKey":"children","error":"$9","errorStyles":[],"errorScripts":[],"template":["$","$La",null,{}],"templateStyles":"$undefined","templateScripts":"$undefined","notFound":[["$","div",null,{"className":"min-h-screen bg-background flex items-center justify-center px-4 sm:px-6 lg:px-8","children":["$","div",null,{"className":"max-w-lg w-full text-center","children":[["$","div",null,{"className":"mb-8","children":["$","div",null,{"className":"relative","children":[["$","h1",null,{"className":"text-9xl font-bold text-muted/20 dark:text-muted/10 select-none","children":"404"}],["$","div",null,{"className":"absolute inset-0 flex items-center justify-center","children":["$","div",null,{"className":"relative","children":[["$","div",null,{"className":"absolute -top-8 -left-8 w-16 h-16 bg-brand-100 dark:bg-brand-950 rounded-full flex items-center justify-center animate-bounce","children":["$","svg",null,{"ref":"$undefined","xmlns":"http://www.w3.org/2000/svg","width":24,"height":24,"viewBox":"0 0 24 24","fill":"none","stroke":"currentColor","strokeWidth":2,"strokeLinecap":"round","strokeLinejoin":"round","className":"lucide lucide-zap w-8 h-8 text-brand-500","children":["$Lb","$undefined"]}]}],"$Lc","$Ld"]}]}]]}]}],"$Le","$Lf","$L10","$L11"]}]}],[]],"forbidden":"$undefined","unauthorized":"$undefined"}]}],"$L12"]}],"$L13","$L14"]}]}]]}]]}],{"children":["pixelflows","$L15",{"children":[["slug","cinematic-fashion-b-roll-videos-seedance-2-5","d"],"$L16",{"children":["__PAGE__","$L17",{},null,false]},null,false]},null,false]},null,false],"$L18",false]],"m":"$undefined","G":["$19",[]],"s":false,"S":true}
1a:I[75289,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6541","static/chunks/fb4e7fc9-e31ad622f5fde25d.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9761","static/chunks/998e86a0-e112630ec4db17dc.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6118","static/chunks/6118-7328b7d3f629a5ef.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","393","static/chunks/393-7cd2157fc3790475.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3985","static/chunks/3985-1f9bcb95de313d61.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4844","static/chunks/4844-011fe3465334e7d0.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","713","static/chunks/713-dcb3ca4f0233704d.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3504","static/chunks/3504-ae9bc2dc93d2ef47.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1885","static/chunks/1885-f206f5873f09b3e2.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5126","static/chunks/5126-bdde68ce894e4935.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5642","static/chunks/5642-c80cb7ee6e914a6a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6372","static/chunks/6372-3e2880fc7d769d1f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6742","static/chunks/app/pixelflows/%5Bslug%5D/page-e866b8c4ac64c944.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],""]
1b:I[4717,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1737","static/chunks/fe207bbb-c84ed91deca4f888.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6397","static/chunks/6397-b7ccd1edf8aedf12.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7979","static/chunks/7979-1799576a8e82a4f3.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","144","static/chunks/144-1a048554a0613a1b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2610","static/chunks/2610-c41f461a4ed7cf24.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","528","static/chunks/528-7ef2006e50a52549.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4077","static/chunks/4077-cba4e88bdb106467.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1924","static/chunks/1924-8c4ee98e9b0dbf83.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2530","static/chunks/2530-9eb218eca5ffc00b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7177","static/chunks/app/layout-75dbb92d3cda2f1c.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"Footer"]
1c:I[35733,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6397","static/chunks/6397-b7ccd1edf8aedf12.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","144","static/chunks/144-1a048554a0613a1b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6090","static/chunks/6090-68b9a87311fa78f8.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","8974","static/chunks/app/page-3bc9473e7ebbbb43.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"Toaster"]
1d:I[20659,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1737","static/chunks/fe207bbb-c84ed91deca4f888.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6397","static/chunks/6397-b7ccd1edf8aedf12.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7979","static/chunks/7979-1799576a8e82a4f3.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","144","static/chunks/144-1a048554a0613a1b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2610","static/chunks/2610-c41f461a4ed7cf24.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","528","static/chunks/528-7ef2006e50a52549.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4077","static/chunks/4077-cba4e88bdb106467.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1924","static/chunks/1924-8c4ee98e9b0dbf83.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","2530","static/chunks/2530-9eb218eca5ffc00b.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7177","static/chunks/app/layout-75dbb92d3cda2f1c.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"LazyLayoutExtras"]
20:I[30661,[],"OutletBoundary"]
22:I[72272,[],"AsyncMetadataOutlet"]
24:I[30661,[],"ViewportBoundary"]
26:I[30661,[],"MetadataBoundary"]
b:["$","path","1xq2db",{"d":"M4 14a1 1 0 0 1-.78-1.63l9.9-10.2a.5.5 0 0 1 .86.46l-1.92 6.02A1 1 0 0 0 13 10h7a1 1 0 0 1 .78 1.63l-9.9 10.2a.5.5 0 0 1-.86-.46l1.92-6.02A1 1 0 0 0 11 14z"}]
c:["$","div",null,{"className":"absolute -top-4 -right-12 w-12 h-12 bg-brand-secondary-100 dark:bg-brand-secondary-950 rounded-full flex items-center justify-center animate-pulse","children":["$","svg",null,{"ref":"$undefined","xmlns":"http://www.w3.org/2000/svg","width":24,"height":24,"viewBox":"0 0 24 24","fill":"none","stroke":"currentColor","strokeWidth":2,"strokeLinecap":"round","strokeLinejoin":"round","className":"lucide lucide-search w-6 h-6 text-brand-secondary-500","children":[["$","circle","4ej97u",{"cx":"11","cy":"11","r":"8"}],["$","path","1qie3q",{"d":"m21 21-4.3-4.3"}],"$undefined"]}]}]
d:["$","div",null,{"className":"absolute -bottom-6 left-4 w-10 h-10 bg-green-100 dark:bg-green-950 rounded-full flex items-center justify-center animate-bounce delay-200","children":["$","div",null,{"className":"w-3 h-3 bg-green-500 rounded-full"}]}]
e:["$","div",null,{"className":"mb-8","children":[["$","h2",null,{"className":"text-2xl font-semibold text-foreground mb-4","children":"Page Not Found"}],["$","p",null,{"className":"text-muted-foreground text-lg mb-2","children":"Oops! The page you're looking for doesn't exist."}],["$","p",null,{"className":"text-muted-foreground","children":"It might have been moved, deleted, or you entered the wrong URL."}]]}]
f:["$","div",null,{"className":"flex flex-col sm:flex-row gap-4 justify-center items-center mb-12","children":[["$","$L1a",null,{"href":"/","className":"inline-flex items-center justify-center gap-2 px-6 py-3 bg-brand-600 dark:bg-brand-500 text-white rounded-lg hover:bg-brand-700 dark:hover:bg-brand-600 transition-colors font-medium text-base w-full sm:w-auto","children":[["$","svg",null,{"ref":"$undefined","xmlns":"http://www.w3.org/2000/svg","width":24,"height":24,"viewBox":"0 0 24 24","fill":"none","stroke":"currentColor","strokeWidth":2,"strokeLinecap":"round","strokeLinejoin":"round","className":"lucide lucide-house w-5 h-5","children":[["$","path","5wwlr5",{"d":"M15 21v-8a1 1 0 0 0-1-1h-4a1 1 0 0 0-1 1v8"}],["$","path","1d0kgt",{"d":"M3 10a2 2 0 0 1 .709-1.528l7-5.999a2 2 0 0 1 2.582 0l7 5.999A2 2 0 0 1 21 10v9a2 2 0 0 1-2 2H5a2 2 0 0 1-2-2z"}],"$undefined"]}],"Go Home"]}],["$","$L1a",null,{"href":"/models","className":"inline-flex items-center justify-center gap-2 px-6 py-3 bg-surface border border-border text-foreground rounded-lg hover:bg-surface/80 transition-colors font-medium text-base w-full sm:w-auto","children":[["$","svg",null,{"ref":"$undefined","xmlns":"http://www.w3.org/2000/svg","width":24,"height":24,"viewBox":"0 0 24 24","fill":"none","stroke":"currentColor","strokeWidth":2,"strokeLinecap":"round","strokeLinejoin":"round","className":"lucide lucide-search w-5 h-5","children":[["$","circle","4ej97u",{"cx":"11","cy":"11","r":"8"}],["$","path","1qie3q",{"d":"m21 21-4.3-4.3"}],"$undefined"]}],"Browse Models"]}]]}]
10:["$","div",null,{"className":"border-t border-border pt-8","children":[["$","p",null,{"className":"text-sm text-muted-foreground mb-4","children":"Popular destinations:"}],["$","div",null,{"className":"flex flex-wrap justify-center gap-4 text-sm","children":[["$","$L1a",null,{"href":"/models/tags/image","className":"text-brand-600 dark:text-brand-400 hover:text-brand-700 dark:hover:text-brand-300 transition-colors","children":"Image Models"}],["$","span",null,{"className":"text-muted-foreground","children":"â€¢"}],["$","$L1a",null,{"href":"/models/tags/video","className":"text-brand-600 dark:text-brand-400 hover:text-brand-700 dark:hover:text-brand-300 transition-colors","children":"Video Models"}],["$","span",null,{"className":"text-muted-foreground","children":"â€¢"}],["$","$L1a",null,{"href":"/models/tags/llms","className":"text-brand-600 dark:text-brand-400 hover:text-brand-700 dark:hover:text-brand-300 transition-colors","children":"LLMs"}],["$","span",null,{"className":"text-muted-foreground","children":"â€¢"}],["$","$L1a",null,{"href":"/models","className":"text-brand-600 dark:text-brand-400 hover:text-brand-700 dark:hover:text-brand-300 transition-colors","children":"All Models"}]]}]]}]
11:["$","div",null,{"className":"mt-8","children":["$","$L1a",null,{"href":"/","className":"inline-flex items-center text-sm text-muted-foreground hover:text-foreground transition-colors","children":[["$","svg",null,{"ref":"$undefined","xmlns":"http://www.w3.org/2000/svg","width":24,"height":24,"viewBox":"0 0 24 24","fill":"none","stroke":"currentColor","strokeWidth":2,"strokeLinecap":"round","strokeLinejoin":"round","className":"lucide lucide-arrow-left w-4 h-4 mr-2","children":[["$","path","1l729n",{"d":"m12 19-7-7 7-7"}],["$","path","x3x0zl",{"d":"M19 12H5"}],"$undefined"]}],"Go back to homepage"]}]}]
12:["$","$L1b",null,{}]
13:["$","$L1c",null,{"closeButton":true,"toastOptions":{"duration":30000}}]
14:["$","$L1d",null,{}]
15:["$","$1","c",{"children":[null,["$","$L8",null,{"parallelRouterKey":"children","error":"$undefined","errorStyles":"$undefined","errorScripts":"$undefined","template":["$","$La",null,{}],"templateStyles":"$undefined","templateScripts":"$undefined","notFound":"$undefined","forbidden":"$undefined","unauthorized":"$undefined"}]]}]
16:["$","$1","c",{"children":[null,"$L1e"]}]
17:["$","$1","c",{"children":["$L1f",null,["$","$L20",null,{"children":["$L21",["$","$L22",null,{"promise":"$@23"}]]}]]}]
18:["$","$1","h",{"children":[null,[["$","$L24",null,{"children":"$L25"}],["$","meta",null,{"name":"next-size-adjust","content":""}]],["$","$L26",null,{"children":["$","div",null,{"hidden":true,"children":["$","$6",null,{"fallback":null,"children":"$L27"}]}]}]]}]
25:[["$","meta","0",{"charSet":"utf-8"}],["$","meta","1",{"name":"viewport","content":"width=device-width, initial-scale=1"}]]
21:null
28:I[25748,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4689","static/chunks/app/pixelflows/%5Bslug%5D/layout-cadf3dfbc811c3c4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"default"]
34:I[85501,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6541","static/chunks/fb4e7fc9-e31ad622f5fde25d.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9761","static/chunks/998e86a0-e112630ec4db17dc.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6118","static/chunks/6118-7328b7d3f629a5ef.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","393","static/chunks/393-7cd2157fc3790475.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3985","static/chunks/3985-1f9bcb95de313d61.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4844","static/chunks/4844-011fe3465334e7d0.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","713","static/chunks/713-dcb3ca4f0233704d.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3504","static/chunks/3504-ae9bc2dc93d2ef47.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1885","static/chunks/1885-f206f5873f09b3e2.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5126","static/chunks/5126-bdde68ce894e4935.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5642","static/chunks/5642-c80cb7ee6e914a6a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6372","static/chunks/6372-3e2880fc7d769d1f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6742","static/chunks/app/pixelflows/%5Bslug%5D/page-e866b8c4ac64c944.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"default",1]
59:I[4704,[],"IconMark"]
29:T9ab,Create a 15-second high-fashion editorial film starring the adult female character in @[char ref]. Preserve her exact face, hairstyle, body proportions, skin tone, outfit, accessories, and visual style. Use the reference as the ONLY character reference. Render the entire film in its aesthetic, with realistic skin, natural anatomy, and physically accurate hair and fabric. No anime, no cartoon, no outfit changes.

Minimal luxury studio, reflective floor, bold directional lighting, rhythmic photographic flashes, photorealistic fashion cinematography. Exactly 10 distinct fashion poses, each with a different silhouette and camera angle:

1. Confident standing pose, filmed from floor level with dramatic wide-angle perspective.
2. One hand beside her face, captured in an extreme side-profile close-up.
3. Asymmetric seated pose, seen directly overhead as the camera rotates.
4. One leg extended toward the lens, her reference footwear dominating the foreground through dramatic foreshortening.
5. Quick, fluid leg cross, captured from a low diagonal angle.
6. Fingertips catching the edge of an existing sleeve or collar, framed in a tight detail shot.
7. Rear three-quarter turn with a deliberate glance back into the camera.
8. Forward lean toward the lens, ending with her palm briefly covering it.
9. Elongated upward stretch, revealed in a sharply tilted wide shot with strong side lighting.
10. Commanding full-body hero pose, looking down into the lens as the camera rapidly pulls back.

Keep the performance fast, confident, and rhythmically tight. Each pose lasts only a fraction of a second while remaining clearly readable. No repeated poses or additional pose changes. Connect the angles through whip pans, movement-matched cuts, snap zooms, flashes, and lens occlusions. Orbit against her turn, sweep from her floor reflection toward her face, and plunge from overhead to floor level. Alternate intimate details with dramatic wide compositions. Accelerate between poses and brake sharply on each hold.

Keep all gestures quick, confident, and fully clothed. Hair and fabric react naturally to movement. No slow motion, lingering shots, or gradual camera drift. Build momentum toward the tenth pose, then stop sharply for a crisp final hold.

Stable identity, anatomy, and outfit throughout. No added accessories, no text, no logos, no AI artifacts. Sound: driving 140 BPM fashion beat, rhythmic shutter clicks, subtle fabric movement, and a sharp final beat.2a:T2bbd,SYSTEM PROMPT â€” Seedance 2.5 Image-Adaptive Prompt Director

You are a cinematography prompt engineer for Seedance 2.5. You will receive two inputs in every task:

Prompt tab â€” a sample prompt written by the user. This is your structural and stylistic template: its shot grammar, its block order, its level of technical detail, its tone.
Input image tab â€” one or more reference images. These are your visual source of truth: subject identity, wardrobe, environment, palette, and mood.

Your job is not to copy the sample prompt's content and not to caption the image. Your job is to rebuild the sample prompt's structure and craft level around what the reference image actually shows, so the output is a new, fully-formed Seedance 2.5 prompt that could not have been written without either input.

CORE PRINCIPLE

You are a physics engine operator, not a mood board. Every line you write must resolve to a visible pixel or an audible sound. If a word in the sample prompt is a flourish that produces neither, drop it during adaptation rather than carrying it forward out of habit.

The reference image is read literally: count what is in frame, measure what can be measured (approximate height relationships, garment structure, material, light direction, camera height and angle, depth of field, colour temperature), and never invent details the image does not support. Where the sample prompt asks for an action or camera move the image can't confirm (because it's a still), extrapolate motion that is physically consistent with the pose, weight distribution, and implied momentum already visible in the frame â€” never a generic or clichÃ©d motion.

STEP ZERO â€” READ BOTH INPUTS BEFORE WRITING ANYTHING

A. Deconstruct the sample prompt's structure. Identify, in order:

Its block architecture (does it use headers, CRITICAL blocks, numbered assets, a geometry map, a shot list â€” or is it looser prose?)
Its register (technical/measurable vs. evocative â€” match it, but pull evocative language toward measurable wherever the reference image gives you the data to do so)
Its length and pacing (single shot vs. multi-shot, how many seconds, how many beats)
Any locked conventions it already uses (no on-screen text, no BGM, diegetic audio only, brand-name handling, etc.) â€” preserve these unless the user says otherwise

B. Deconstruct the reference image. Identify, in order:

Subject(s): count, approximate age register (role/appearance only, never assumed exact age), body build, pose, weight-bearing leg or point of balance, gaze direction, hand position, garment behaviour (taut, loose, mid-motion, static)
Wardrobe and identity markers: colour, material, fit, layering, accessories, hair colour/length/style
Environment: architecture, depth planes (foreground/midground/background), practical light sources visible in frame, surface materials, weather or atmosphere cues
Camera: apparent focal length/FOV, height relative to subject, angle (eye-level, low, high, Dutch), depth of field falloff
Light: direction, quality (hard/soft), colour temperature, practical sources vs. ambient fill
Anything in motion or implying motion (fabric, hair, dust, water, blur)

C. Confirm the target engine is Seedance 2.5. This governs two hard ceilings:

Up to 50 image references â€” use this budget deliberately. A single character can carry a front reference, a profile/detail reference, and a wardrobe reference as separate slots rather than one composite. A prop or environment plate gets its own slot rather than being folded into a character line. Do not pad the reference list with near-duplicates â€” if two references would teach the model the same thing, cut one.
Up to 30 seconds runtime â€” do not default to the ceiling. Build only the runtime the adapted action actually needs. If the sample prompt's action would run past 30 seconds when adapted, split into two prompts and say so.

If it is genuinely unclear whether the target is 2.0 or 2.5, ask once in one line before writing. Otherwise assume 2.5 per this system prompt and proceed.

THE ADAPTATION RULE

For every structural element present in the sample prompt, ask: "What does the reference image supply for this slot?"

If the image supplies real data (a pose, a garment, a light direction, a spatial layout) â€” replace the sample's placeholder content with the image's actual content, expressed at the same level of technical precision the sample used or higher.
If the image is silent on something the sample prompt specifies (e.g., a camera move, a line of dialogue, a sound cue) â€” retain the sample's original specification, adapted only enough to stay physically consistent with the image's geometry and framing.
If the image shows something the sample prompt has no slot for (a distinctive prop, a second figure, a strong light source) â€” add it as its own block or asset rather than dropping it silently. Reference images always win on visual fact.

Never let the sample prompt's original subject, wardrobe, or setting leak into the output once the reference image contradicts it. The image is the identity source; the sample prompt is the grammar source.

WRITE THE VISIBLE
Instead of	Write
she looks confident	chin level, shoulders back, weight settled evenly over both feet, gaze fixed forward
the room feels warm	one low-angled tungsten source camera-left, warm amber falloff across the near wall, cooler blue spill from a window at frame right
she moves quickly	strides at a pace that swings her coat hem clear of her calves, heel-strike audible on the hard floor
he looks taller than her	he stands roughly a head taller, her eyeline meeting his collarbone

Measurables to extract from the image and use in the prompt: approximate height relationships in cm where the image implies them Â· garment weight and how it would move (light chiffon vs. structured wool) Â· light direction and temperature in degrees/Kelvin-equivalent language Â· depth planes named explicitly (foreground/midground/background) Â· camera height and angle stated in FOV degrees and eye-level relationship Â· direction always labelled screen-relative or character-relative.

OUTPUT STRUCTURE â€” MATCH THE SAMPLE, ENFORCE THE SPINE

Use the sample prompt's own architecture as the visible shape of your output. Underneath that shape, silently enforce this priority order so nothing is missing regardless of how loose the sample was:

Header â€” shot count, runtime, cut policy, speed policy
Style prefix â€” render style, texture, skin, technical invariants carried from the image
No on-screen text (unless the image itself contains legible text/graphics â€” if so, transcribe it verbatim and describe it physically)
Critical blocks â€” anything from the image or sample that must not drift (max 4, ordered by importance)
Assets â€” one per character/prop/environment plate, each with identity + this-scene action + a fidelity assertion tying it to its reference
Geometry map â€” absolute frame position and depth planes, drawn from the image's actual composition
First frame â€” what is already happening at frame one, matching the reference image if it is the opening frame
Optics â€” lens lock in FOV degrees, matching the image's apparent focal length
Camera â€” physicality register and behaviour (handheld, locked, dolly, etc.)
Light & colour â€” direction, quality, temperature, drawn from the image
Atmosphere â€” air density, depth planes, any vapor bound to a visible source
Action timing â€” timecoded beats, physically consistent with the image's implied momentum
Physics â€” mass, deformation, rebound, lag, contact, gravity behaving normally
Acting â€” face, eyes, brow, matched to the expression actually visible in the reference
Audio â€” diegetic by default unless the sample specifies an attached track
Locks â€” positive ordered chain of what must hold across the whole shot

If the sample prompt does not use explicit block headers, keep the output in the sample's own prose style but ensure every one of the sixteen concerns above is addressed somewhere in the text â€” do not silently drop physics, geometry, or acting because the sample's register was casual.

HOUSE RULES (NON-NEGOTIABLE)
No character names in the prompt body â€” visual descriptors only (hair, wardrobe, identity markers). Names may appear only as aliases in a numbered reference list above the final prompt.
No aspect ratio in the body â€” that is a UI setting.
No internal production context ("carried from the previous scene," "as before") â€” every prompt stands alone.
No platform or tool names inside the prompt body.
No meta-commentary â€” every word describes something visible or audible.
Age-blind â€” describe by role, build, hair, wardrobe, never a stated age.
Brand names, text, and graphics visible in the reference image are written verbatim and described physically (shape, colour, placement, legibility) â€” paraphrasing renders a vague approximation.
English only in the final prompt.
Every fact lives in exactly one slot â€” do not restate wardrobe in Action Timing if it's already fixed in Assets; only mention a garment there when it is doing something visible (a hem swinging, hair whipping clear).
DELIVERY FORMAT

Every response has exactly three parts, nothing else:

A bolded title with runtime â€” e.g. **Rooftop entry â€” reference-matched â€” 8s**
A numbered reference list in attach order (character front/detail/wardrobe refs first, then props, then environment plate, then video/audio if any) â€” respecting the 50-slot ceiling on 2.5
One fenced code block containing the complete adapted prompt, English only, following the structure rules above

No preamble, no postamble. If the reference image conflicts with something explicit in the sample prompt (e.g. sample specifies a blue jacket, image shows red), flag the conflict in one line above the title and default to the image as the source of truth unless told otherwise.

A negative prompt block ships as an optional second code block only when there's a known drift risk (mixed identity cues, locked spatial geometry, a prop that could be mistaken for something else) â€” grouped by failure category, comma-separated, no full sentences.

LENGTH DISCIPLINE

A four-shot sequence with four assets should land at roughly 900â€“1,400 words. Past that, the critical blocks lose weight against the descriptive body. If the adapted prompt is running long, check for a duplicated fact before adding a trim pass â€” the same detail stated twice reads long without reading more specific.

PRE-DELIVERY CHECK
 Structure and register match the sample prompt's own architecture
 Every subject, wardrobe item, prop, and environment detail traces back to the reference image, not to assumption
 Any camera move or action beyond what the still shows is physically consistent with the pose, weight, and momentum visible in the image
 Reference count within 50, no near-duplicate references
 Runtime within 30s, or split declared
 No character names, no aspect ratio, no tool names, no restated facts
 Physics chain present: mass, contact, deformation â€” nothing floats or slides unnaturally
 Delivery is exactly title + reference list + one code block, nothing more2b:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.2c:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.2d:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.2e:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.2f:T7d8,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 di30:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.31:T642,This workflow generates a 15-second high-fashion editorial film, featuring structured fashion poses and cinematic editing techniques. It is especially beneficial for fashion brands, creative agencies, and videographers who require quick and visually appealing content for promotional needs. The key AI models powering this workflow are Seedance 2.5 for image-to-video conversion and GPT-6 Astra for text generation. The primary value offered by this workflow is the ability to automate complex video production processes, ensuring cost-effectiveness and rapid content generation.

---

## 3. How does it work
- The workflow begins with GPT-6 Astra, which interprets an input image and generates a detailed cinematographic prompt. This AI model understands complex instructions and uses reference images to maintain character consistency.
- The Seedance 2.5 model takes the generated script and transforms it into a video, providing cinematic shot transitions and synchronized audio, ensuring a seamless and professional 15-second fashion b-roll.

---

## 4. How to customize it
Users can adapt the workflow to suit different creative visions by modifying the prompts fed into GPT-6 Astra. For instance, a user might choose to adjust the scenarios or poses to reflect their brandâ€™s unique aesthetic. Alternatively, they could replace the reference image to feature different models or wardrobes, instantly shifting the videoâ€™s focal point to meet their specific campaign needs.

---

## 5. Who is it for
- Fashion Designers and Brands
- Marketing Agencies
- Visual Content Creators and Videographers1e:["$","$L28",null,{"slug":"cinematic-fashion-b-roll-videos-seedance-2-5","template":{"_id":{"$oid":"6a9fa7a00bcd3da2348df830"},"author":"photonotixmedia@gmail.com","template_name":"Cinematic Fashion B-Roll Videos with Seedance 2.5","template_data":{"nodes":[{"id":"node-1788766051328","type":"apiNode","position":{"x":740,"y":260},"data":{"type":"api","id":"node-1788766051328","meta":{"minimised":false},"parameters":{"prompt":"$29","system_prompt":"$2a","image":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"]},"template":"gpt-6-astra","apiKeyName":"gpt_6_astra_e2b1x","modelInfo":{"average_cost":0.03669338888888889,"coming_soon":false,"created_at":"2026-09-05 03:53:39.424474+00:00","default_image_output":"https://segmind-sd-models.s3.amazonaws.com/display_images/gpt-4.png","default_response":null,"description":"Complex reasoning, code generation, and image understanding; 1M-token context.","inference_provider":"OPENAI","is_depreciated":false,"is_new":true,"is_visible":true,"latency":12.22961,"parameters":{"image":{"affectsPricing":false,"component":"image","default":null,"description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","displayOrder":1,"hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"label":"Image","type":"image"},"prompt":{"affectsPricing":false,"component":"textarea","default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","displayOrder":0,"displayValue":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":true,"label":"prompt","type":"str"}},"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","examples":[{"prompt":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input."}],"properties":{"image":{"description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","format":"uri","title":"Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"order":1}},"prompt":{"default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","minLength":1,"title":"Prompt","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":0,"placeholder":"Ask a question or give an instruction"}}},"required":["prompt"],"type":"object"},"partner":"OPENAI","popularity_score":0,"search_override_score":0,"slug":"gpt-6-astra","thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/gpt-6-astra-1903.webp","title":"GPT 6 Astra","type":"textToText","objectID":"gpt-6-astra","_highlightResult":{"description":{"value":"Complex reasoning, code generation, and image understanding; 1M-token context.","matchLevel":"none","matchedWords":[]},"parameters":{"prompt":{"component":{"value":"textarea","matchLevel":"none","matchedWords":[]},"default":{"value":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","matchLevel":"none","matchedWords":[]},"description":{"value":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","matchLevel":"none","matchedWords":[]},"displayOrder":{"value":"0","matchLevel":"none","matchedWords":[]},"displayValue":{"value":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","matchLevel":"none","matchedWords":[]},"label":{"value":"prompt","matchLevel":"none","matchedWords":[]},"type":{"value":"str","matchLevel":"none","matchedWords":[]}}},"title":{"value":"GPT 6 Astra","matchLevel":"none","matchedWords":[]}}},"modelData":{"title":"GPT 6 Astra","type":"textToText","slug":"gpt-6-astra","description":"Complex reasoning, code generation, and image understanding; 1M-token context.","is_visible":true,"default_image_output":"https://segmind-sd-models.s3.amazonaws.com/display_images/gpt-4.png","default_response":null,"thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/gpt-6-astra-1903.webp","is_depreciated":false,"is_new":true,"coming_soon":false,"segmind":true,"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","type":"object","properties":{"prompt":{"type":"string","title":"Prompt","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","minLength":1,"x-ui-metadata":{"order":0,"advanced":false,"hidden":false,"placeholder":"Ask a question or give an instruction"}},"image":{"type":"string","title":"Image","description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","format":"uri","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":1,"advanced":false,"hidden":false}}},"required":["prompt"],"examples":[{"prompt":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input."}]},"partner":"OPENAI","inference_provider":"OPENAI","latency":12.22961,"average_cost":0.036693388888888887,"search_override_score":0,"created_at":"2026-09-05 03:53:39.424474+00:00"},"parameterTemplates":{"image":"{{ input_image }}"},"fanOutInstructions":"Output usage â€” your response is used directly as the following downstream model input(s):\n  â€¢ Seedance 2.5 Â· prompt  (string) â€” Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citatiâ€¦\nReturn only the value for the connected input â€” no markdown, code fences, or commentary.","outputs":{"text":"$2b","data":{"text":"$2c","metrics":{"cost":0.15141,"inference_time":72.77442,"queue_time":0.038667,"remaining_credits":101.6587128706,"total_time":72.813087},"reasoning":null,"timings":{"inference":72.77442},"output":"$2d","choices":[{"message":{"content":"$2e"}}]},"executionTime":74061.29999995232},"errors":"","status":"success"},"measured":{"width":380,"height":1003}},{"id":"node-1788766056595","type":"apiNode","position":{"x":1190,"y":260},"data":{"type":"api","id":"node-1788766056595","meta":{"minimised":false},"parameters":{"prompt":"{{ gpt_6_astra_e2b1x }}","duration":15,"reference_images":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"seed":58385081},"template":"seedance-2.5","apiKeyName":"seedance_2_5_qgfrs","modelInfo":{"average_cost":3.033373417751536,"coming_soon":false,"created_at":"2026-08-07 06:05:13.024224+00:00","default_image_output":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-example-output.mp4","default_response":null,"description":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","inference_provider":"BYTEDANCE","is_depreciated":false,"is_new":true,"is_visible":true,"latency":311.42742,"parameters":{"aspect_ratio":{"affectsPricing":false,"component":"dropdown","default":"adaptive","description":"Output ratio: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive. Set adaptive when using first_frame_url.","displayOrder":8,"displayValue":"adaptive","hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Aspect Ratio","type":"str","values":["16:9","9:16","1:1","4:3","3:4","21:9","adaptive"]},"auto_resize":{"affectsPricing":false,"component":"text_field","default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","displayOrder":15,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Auto Resize Inputs","type":"bool"},"bitrate_mode":{"affectsPricing":false,"component":"dropdown","default":"standard","description":"Encoding bitrate: standard or high. Use standard for smaller files, high for maximum fidelity.","displayOrder":13,"displayValue":"standard","hidden":false,"isAdvanced":true,"isEnum":true,"isRequired":false,"label":"Bitrate Mode","type":"str","values":["standard","high"]},"duration":{"affectsPricing":false,"component":"dropdown","default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose (required for video edits: output follows the input video).","displayOrder":6,"displayValue":5,"hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Duration","range":[4,30],"type":"int","values":[-1,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30]},"first_frame_url":{"affectsPricing":false,"component":"textbox","default":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","description":"Starting image for image-to-video; no real human faces. Leave empty for text-to-video; cannot combine with reference_images.","displayOrder":1,"displayValue":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"label":"First Frame Image","type":"str"},"generate_audio":{"affectsPricing":false,"component":"text_field","default":true,"description":"Co-generates synchronized dialogue, effects, and music. Enable for speech or ambient scenes.","displayOrder":9,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Generate Audio","type":"bool"},"last_frame_url":{"affectsPricing":false,"component":"textbox","default":"","description":"Ending frame guiding a transition; requires first_frame_url. Use for controlled start-to-end morphs.","displayOrder":4,"displayValue":"","format":"uri","hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Last Frame Image","type":"str"},"output_format":{"affectsPricing":false,"component":"dropdown","default":"mp4","description":"mp4 = best compatibility. mov = higher color fidelity; recommended for video edits.","displayOrder":14,"displayValue":"mp4","hidden":false,"isAdvanced":true,"isEnum":true,"isRequired":false,"label":"Output Format","type":"str","values":["mp4","mov"]},"prompt":{"affectsPricing":false,"component":"textarea","default":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","description":"Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citations. Detail subject, action, camera, and audio.","displayOrder":0,"displayValue":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":true,"label":"Prompt","type":"str"},"reference_audios":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 10 MP3s (2-30s) to drive audio, cited as @Audio 1. Use for voice, music, or beat-matched pacing.","displayOrder":5,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Audios","type":"array"},"reference_images":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 30 images for character or style consistency, cited as @Image 1. Not usable with first_frame_url.","displayOrder":2,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Images","type":"array"},"reference_videos":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 10 clips (2-30s) for motion or style transfer, cited as @Video 1. Billed at video-input rate.","displayOrder":3,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Videos","type":"array"},"resolution":{"affectsPricing":false,"component":"dropdown","default":"720p","description":"Output resolution, 480p, 720p or 1080p. Use 480p for drafts, 720p for final renders, 1080p for highest fidelity.","displayOrder":7,"displayValue":"720p","hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Resolution","type":"str","values":["480p","720p","1080p"]},"return_last_frame":{"affectsPricing":false,"component":"text_field","default":false,"description":"Returns the final frame as an image URL. Enable to chain clips into longer sequences.","displayOrder":11,"displayValue":false,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Return Last Frame","type":"bool"},"seed":{"affectsPricing":false,"component":"textbox","default":42,"description":"Reproducibility seed (-1 to 2147483647). Use -1 for random, fix a value to iterate a scene.","displayOrder":10,"displayValue":42,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Seed","range":[-1,2147483647],"type":"int"},"skip_moderation":{"affectsPricing":false,"component":"toggle","default":true,"description":"Bypasses the content pre-filter for false positives. Leave off unless artistic content is wrongly blocked.","displayOrder":12,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Skip Moderation","type":"bool"}},"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","examples":[{"aspect_ratio":"adaptive","bitrate_mode":"standard","duration":5,"first_frame_url":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","generate_audio":true,"last_frame_url":"","prompt":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","reference_audios":[],"reference_images":[],"reference_videos":[],"resolution":"720p","return_last_frame":false,"seed":42,"skip_moderation":false}],"properties":{"aspect_ratio":{"default":"16:9","description":"Output aspect ratio. Use 16:9 for landscape, 9:16 for vertical social video, 21:9 for ultrawide cinematic.","oneOf":[{"const":"16:9","title":"16:9"},{"const":"9:16","title":"9:16"},{"const":"1:1","title":"1:1"},{"const":"4:3","title":"4:3"},{"const":"3:4","title":"3:4"},{"const":"21:9","title":"21:9"},{"const":"adaptive","title":"adaptive"}],"title":"Aspect Ratio","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":8}},"auto_resize":{"default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","title":"Auto Resize Inputs","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":15}},"bitrate_mode":{"default":"standard","description":"Output video encoding bitrate. 'standard' gives smaller files and faster downloads; 'high' produces ~5-6x higher bitrate for better visual fidelity and fewer compression artifacts (larger files). Does not affect price.","oneOf":[{"const":"standard","title":"Standard"},{"const":"high","title":"High"}],"title":"Bitrate Mode","type":"string","x-ui-metadata":{"advanced":true,"hidden":false,"order":13}},"duration":{"default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose. Video-editing prompts require -1 (output follows the input video, which must be 4-30s). Cost scales with output duration (token-based).","oneOf":[{"const":-1,"title":"Auto (use for video edits)"},{"const":4,"title":"4s"},{"const":5,"title":"5s"},{"const":6,"title":"6s"},{"const":7,"title":"7s"},{"const":8,"title":"8s"},{"const":9,"title":"9s"},{"const":10,"title":"10s"},{"const":11,"title":"11s"},{"const":12,"title":"12s"},{"const":13,"title":"13s"},{"const":14,"title":"14s"},{"const":15,"title":"15s"},{"const":16,"title":"16s"},{"const":17,"title":"17s"},{"const":18,"title":"18s"},{"const":19,"title":"19s"},{"const":20,"title":"20s"},{"const":21,"title":"21s"},{"const":22,"title":"22s"},{"const":23,"title":"23s"},{"const":24,"title":"24s"},{"const":25,"title":"25s"},{"const":26,"title":"26s"},{"const":27,"title":"27s"},{"const":28,"title":"28s"},{"const":29,"title":"29s"},{"const":30,"title":"30s"}],"title":"Duration","type":"integer","x-ui-metadata":{"advanced":false,"hidden":false,"order":6}},"first_frame_url":{"description":"Starting frame image URL for image-to-video. Animates outward from this reference. Note: images with real human faces are blocked by ByteDance content policy. Cannot be used together with reference_images.","format":"uri","title":"First Frame Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"First & last frame","order":1}},"generate_audio":{"default":true,"description":"Co-generate synchronized audio (dialogue, SFX, ambient, music) in the same pass. Best for scenes with speech, nature sounds, or music.","title":"Generate Audio","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":9}},"last_frame_url":{"description":"Ending frame; requires first_frame_url. Guides transitions between two frames.","format":"uri","title":"Last Frame Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"First & last frame","order":2}},"output_format":{"default":"mp4","description":"mp4 = best compatibility. mov (H.264 + yuv444p + PCM) = higher color fidelity; recommended for video editing/extension round-trips.","enum":["mp4","mov"],"title":"Output Format","type":"string","x-ui-metadata":{"advanced":true,"hidden":false,"order":14}},"prompt":{"description":"Text describing the video. Use Shot 1:, Shot 2: for multi-shot sequences. Reference uploaded assets as '@Image 1', '@Video 1', '@Audio 1' in your prompt.","title":"Prompt","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":0}},"reference_audios":{"description":"Up to 10 reference audio files (MP3, 2-30 seconds each, 30 seconds combined, max 15MB each). Cite as '@Audio 1' in your prompt. Audio-only reference is supported on seedance-2.5.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["audio/mpeg"]}},"maxItems":10,"title":"Reference Audios","type":"array","x-ui-metadata":{"advanced":true,"hidden":false,"mode-group":"Reference media","order":5}},"reference_images":{"description":"Up to 30 reference images (max 30MB each) for character/style consistency. Cite as '@Image 1', '@Image 2' in your prompt. Cannot be used together with first_frame_url.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["image/*"]}},"maxItems":30,"title":"Reference Images","type":"array","x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"Reference media","order":3}},"reference_videos":{"description":"Up to 10 reference videos (2-30 seconds each, 30 seconds combined, mp4/mov, max 200MB each) for motion/style transfer. Cite as '@Video 1' in your prompt. Requests with reference videos are billed at the video-input token rate.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["video/mp4","video/quicktime"]}},"maxItems":10,"title":"Reference Videos","type":"array","x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"Reference media","order":4}},"resolution":{"default":"720p","description":"Output resolution. seedance-2.5 supports 480p, 720p and 1080p (24fps). Use 480p for drafts and fast iteration, 720p for final renders, 1080p for the highest fidelity (10-bit colour). Higher resolution costs more (token-based).","oneOf":[{"const":"480p","title":"480p"},{"const":"720p","title":"720p"},{"const":"1080p","title":"1080p"}],"title":"Resolution","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":7}},"return_last_frame":{"default":false,"description":"Returns the final video frame as a separate image, delivered in the response as `video.last_frame_url`. Chain shots by feeding that URL into the next generation's `first_frame_url`. Note this is an output: it is distinct from the `last_frame_url` input parameter, which is an image you supply to steer how the clip ends.","title":"Return Last Frame","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":11}},"seed":{"default":42,"description":"Reproducibility seed. Use -1 for random output, set a fixed value to iterate on the same scene.","maximum":2147483647,"minimum":-1,"title":"Seed","type":"integer","x-ui-metadata":{"advanced":true,"hidden":false,"order":10}},"skip_moderation":{"default":true,"description":"Bypass BytePlus content moderation pre-filter. Useful when generating artistic content or faces that may trigger false positives. Baseline BytePlus safety policies still apply.","title":"Skip Moderation","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":12}}},"required":["prompt"],"type":"object"},"partner":"BYTEDANCE","popularity_score":0,"search_override_score":0,"slug":"seedance-2.5","thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-5954.webp","title":"Seedance 2.5","type":"imageToVideo","objectID":"seedance-2.5","_highlightResult":{"description":{"value":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","matchLevel":"none","matchedWords":[]},"parameters":{"prompt":{"component":{"value":"textarea","matchLevel":"none","matchedWords":[]},"default":{"value":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","matchLevel":"none","matchedWords":[]},"description":{"value":"Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citations. Detail subject, action, camera, and audio.","matchLevel":"none","matchedWords":[]},"displayOrder":{"value":"0","matchLevel":"none","matchedWords":[]},"displayValue":{"value":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","matchLevel":"none","matchedWords":[]},"label":{"value":"Prompt","matchLevel":"none","matchedWords":[]},"type":{"value":"str","matchLevel":"none","matchedWords":[]}}},"title":{"value":"<em>Seedance</em> 2.5","matchLevel":"full","fullyHighlighted":false,"matchedWords":["seedance"]}}},"modelData":{"title":"Seedance 2.5","type":"imageToVideo","slug":"seedance-2.5","description":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","is_visible":true,"default_image_output":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-example-output.mp4","default_response":null,"thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-5954.webp","is_depreciated":false,"is_new":true,"coming_soon":false,"segmind":true,"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","type":"object","properties":{"prompt":{"type":"string","title":"Prompt","description":"Text describing the video. Use Shot 1:, Shot 2: for multi-shot sequences. Reference uploaded assets as '@Image 1', '@Video 1', '@Audio 1' in your prompt.","x-ui-metadata":{"order":0,"advanced":false,"hidden":false}},"first_frame_url":{"type":"string","title":"First Frame Image","format":"uri","description":"Starting frame image URL for image-to-video. Animates outward from this reference. Note: images with real human faces are blocked by ByteDance content policy. Cannot be used together with reference_images.","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":1,"advanced":false,"hidden":false,"mode-group":"First & last frame"}},"last_frame_url":{"type":"string","title":"Last Frame Image","format":"uri","description":"Ending frame; requires first_frame_url. Guides transitions between two frames.","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":2,"advanced":false,"hidden":false,"mode-group":"First & last frame"}},"reference_images":{"type":"array","title":"Reference Images","description":"Up to 30 reference images (max 30MB each) for character/style consistency. Cite as '@Image 1', '@Image 2' in your prompt. Cannot be used together with first_frame_url.","maxItems":30,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["image/*"]}},"x-ui-metadata":{"order":3,"advanced":false,"hidden":false,"mode-group":"Reference media"}},"reference_videos":{"type":"array","title":"Reference Videos","description":"Up to 10 reference videos (2-30 seconds each, 30 seconds combined, mp4/mov, max 200MB each) for motion/style transfer. Cite as '@Video 1' in your prompt. Requests with reference videos are billed at the video-input token rate.","maxItems":10,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["video/mp4","video/quicktime"]}},"x-ui-metadata":{"order":4,"advanced":false,"hidden":false,"mode-group":"Reference media"}},"reference_audios":{"type":"array","title":"Reference Audios","description":"Up to 10 reference audio files (MP3, 2-30 seconds each, 30 seconds combined, max 15MB each). Cite as '@Audio 1' in your prompt. Audio-only reference is supported on seedance-2.5.","maxItems":10,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["audio/mpeg"]}},"x-ui-metadata":{"order":5,"advanced":true,"hidden":false,"mode-group":"Reference media"}},"duration":{"type":"integer","title":"Duration","default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose. Video-editing prompts require -1 (output follows the input video, which must be 4-30s). Cost scales with output duration (token-based).","x-ui-metadata":{"order":6,"advanced":false,"hidden":false},"oneOf":[{"const":-1,"title":"Auto (use for video edits)"},{"const":4,"title":"4s"},{"const":5,"title":"5s"},{"const":6,"title":"6s"},{"const":7,"title":"7s"},{"const":8,"title":"8s"},{"const":9,"title":"9s"},{"const":10,"title":"10s"},{"const":11,"title":"11s"},{"const":12,"title":"12s"},{"const":13,"title":"13s"},{"const":14,"title":"14s"},{"const":15,"title":"15s"},{"const":16,"title":"16s"},{"const":17,"title":"17s"},{"const":18,"title":"18s"},{"const":19,"title":"19s"},{"const":20,"title":"20s"},{"const":21,"title":"21s"},{"const":22,"title":"22s"},{"const":23,"title":"23s"},{"const":24,"title":"24s"},{"const":25,"title":"25s"},{"const":26,"title":"26s"},{"const":27,"title":"27s"},{"const":28,"title":"28s"},{"const":29,"title":"29s"},{"const":30,"title":"30s"}]},"resolution":{"type":"string","title":"Resolution","default":"720p","description":"Output resolution. seedance-2.5 supports 480p, 720p and 1080p (24fps). Use 480p for drafts and fast iteration, 720p for final renders, 1080p for the highest fidelity (10-bit colour). Higher resolution costs more (token-based).","oneOf":[{"const":"480p","title":"480p"},{"const":"720p","title":"720p"},{"const":"1080p","title":"1080p"}],"x-ui-metadata":{"order":7,"advanced":false,"hidden":false}},"aspect_ratio":{"type":"string","title":"Aspect Ratio","default":"16:9","description":"Output aspect ratio. Use 16:9 for landscape, 9:16 for vertical social video, 21:9 for ultrawide cinematic.","oneOf":[{"const":"16:9","title":"16:9"},{"const":"9:16","title":"9:16"},{"const":"1:1","title":"1:1"},{"const":"4:3","title":"4:3"},{"const":"3:4","title":"3:4"},{"const":"21:9","title":"21:9"},{"const":"adaptive","title":"adaptive"}],"x-ui-metadata":{"order":8,"advanced":false,"hidden":false}},"generate_audio":{"type":"boolean","title":"Generate Audio","default":true,"description":"Co-generate synchronized audio (dialogue, SFX, ambient, music) in the same pass. Best for scenes with speech, nature sounds, or music.","x-ui-metadata":{"order":9,"advanced":true,"hidden":false}},"seed":{"type":"integer","title":"Seed","default":42,"minimum":-1,"maximum":2147483647,"description":"Reproducibility seed. Use -1 for random output, set a fixed value to iterate on the same scene.","x-ui-metadata":{"order":10,"advanced":true,"hidden":false}},"return_last_frame":{"type":"boolean","title":"Return Last Frame","default":false,"description":"Returns the final video frame as a separate image, delivered in the response as `video.last_frame_url`. Chain shots by feeding that URL into the next generation's `first_frame_url`. Note this is an output: it is distinct from the `last_frame_url` input parameter, which is an image you supply to steer how the clip ends.","x-ui-metadata":{"order":11,"advanced":true,"hidden":false}},"skip_moderation":{"type":"boolean","title":"Skip Moderation","default":true,"description":"Bypass BytePlus content moderation pre-filter. Useful when generating artistic content or faces that may trigger false positives. Baseline BytePlus safety policies still apply.","x-ui-metadata":{"order":12,"advanced":true,"hidden":false}},"bitrate_mode":{"type":"string","title":"Bitrate Mode","default":"standard","description":"Output video encoding bitrate. 'standard' gives smaller files and faster downloads; 'high' produces ~5-6x higher bitrate for better visual fidelity and fewer compression artifacts (larger files). Does not affect price.","oneOf":[{"const":"standard","title":"Standard"},{"const":"high","title":"High"}],"x-ui-metadata":{"order":13,"advanced":true,"hidden":false}},"output_format":{"type":"string","title":"Output Format","default":"mp4","enum":["mp4","mov"],"description":"mp4 = best compatibility. mov (H.264 + yuv444p + PCM) = higher color fidelity; recommended for video editing/extension round-trips.","x-ui-metadata":{"order":14,"advanced":true,"hidden":false}},"auto_resize":{"type":"boolean","title":"Auto Resize Inputs","default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","x-ui-metadata":{"order":15,"advanced":true,"hidden":false}}},"required":["prompt"],"examples":[{"prompt":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","first_frame_url":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","last_frame_url":"","reference_images":[],"reference_videos":[],"reference_audios":[],"duration":5,"resolution":"720p","aspect_ratio":"adaptive","generate_audio":true,"seed":42,"return_last_frame":false,"skip_moderation":false,"bitrate_mode":"standard"}]},"partner":"BYTEDANCE","inference_provider":"BYTEDANCE","latency":311.42742,"average_cost":3.033373417751536,"search_override_score":0,"created_at":"2026-08-07 06:05:13.024224+00:00"},"parameterTemplates":{"prompt":"{{ gpt_6_astra_e2b1x }}","reference_images":"{{ input_image }}"},"outputs":{"video":["https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4"],"url":"https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4","data":{"video":["https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4"],"metrics":{"cost":3.564153,"inference_time":848.369815,"queue_time":0.041331,"remaining_credits":101.5073028706,"total_time":848.411146},"output":"https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4","prompt":"$2f","seed":"58385081","timings":{"inference":848.369815},"url":"https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4"},"executionTime":850981.6999999881,"promptByUrl":{"https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4":"$30"}},"errors":"","status":"success"},"measured":{"width":380,"height":1420}},{"id":"node-1788766571473","type":"inputNode","position":{"x":330,"y":260},"data":{"type":"input","id":"node-1788766571473","template":"file","mode":"any","value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"apiKeyName":"input_image","keyname":"input_image","useInApi":true,"status":"success","outputs":{"value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"data":{"value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"]},"executionTime":0.09999996423721313},"errors":""},"measured":{"width":320,"height":582},"selected":false},{"id":"node-1788768863220","type":"outputNode","position":{"x":1640,"y":260},"data":{"type":"output","id":"node-1788768863220","value":[],"template":"auto","apiKeyName":"Output_Video","keyname":"Output_Video","useInApi":true,"meta":{"sizeMode":"screen","nativeSize":false,"width":1106,"prevWidth":380}},"width":1106,"measured":{"width":1106,"height":700},"selected":false}],"edges":[{"source":"node-1788766571473","sourceHandle":"string","target":"node-1788766051328","targetHandle":"param-image","id":"xy-edge__node-1788766571473string-node-1788766051328param-image"},{"source":"node-1788766051328","sourceHandle":"text","target":"node-1788766056595","targetHandle":"param-prompt","id":"xy-edge__node-1788766051328text-node-1788766056595param-prompt","selected":false},{"source":"node-1788766571473","sourceHandle":"string","target":"node-1788766056595","targetHandle":"param-reference_images","id":"xy-edge__node-1788766571473string-node-1788766056595param-reference_images"},{"source":"node-1788766056595","sourceHandle":"video","target":"node-1788768863220","targetHandle":"output","id":"xy-edge__node-1788766056595video-node-1788768863220output"}]},"tags":["fashion","film-making"],"image":"https://images.segmind.com/outputs/c3578c95-8cb6-4d52-886b-884b3ea1bac6.png","name":"Amritesh K","graph_id":"6a9e674d4f888976c2dd97e3","template_slug":"cinematic-fashion-b-roll-videos-seedance-2-5","seo_description":"$31","meta_description":"Create Captivating 15-Second Fashion B-Roll Videos with Seedance 2.5 and Cinematic Posing.","show_pixelflow":true,"is_private":false,"api_id":"6a9fa7a00bcd3da2348df831","api_slug":"6a9fa7a00bcd3da2348df830-v1","api_version":"v2","enable_api":true,"avg_latency":341.89,"avg_cost":3.1432},"hasApi":true,"children":"$L32"}]
33:T6ce,{"@context":"https://schema.org","@graph":[{"@type":"SoftwareApplication","@id":"https://www.segmind.com/pixelflows/cinematic-fashion-b-roll-videos-seedance-2-5#software","applicationCategory":"Workflow Template","name":"Cinematic Fashion B-Roll Videos with Seedance 2.5","description":"Create Captivating 15-Second Fashion B-Roll Videos with Seedance 2.5 and Cinematic Posing.","url":"https://www.segmind.com/pixelflows/cinematic-fashion-b-roll-videos-seedance-2-5","operatingSystem":"Any","image":"https://images.segmind.com/outputs/c3578c95-8cb6-4d52-886b-884b3ea1bac6.png","author":{"@type":"Person","name":"Segmind Community"},"offers":{"@type":"Offer","price":"3.1432","priceCurrency":"USD"},"provider":{"@type":"Organization","@id":"https://www.segmind.com/#organization","name":"Segmind","url":"https://www.segmind.com","logo":"https://www.segmind.com/favicon.svg"}},{"@type":"WebApplication","@id":"https://www.segmind.com/pixelflows/cinematic-fashion-b-roll-videos-seedance-2-5#web","applicationCategory":"Workflow Template","browserRequirements":"Requires JavaScript","name":"Cinematic Fashion B-Roll Videos with Seedance 2.5","description":"Create Captivating 15-Second Fashion B-Roll Videos with Seedance 2.5 and Cinematic Posing.","url":"https://www.segmind.com/pixelflows/cinematic-fashion-b-roll-videos-seedance-2-5","operatingSystem":"Any","image":"https://images.segmind.com/outputs/c3578c95-8cb6-4d52-886b-884b3ea1bac6.png","author":{"@type":"Person","name":"Segmind Community"},"offers":{"@type":"Offer","price":"3.1432","priceCurrency":"USD"},"provider":{"@type":"Organization","@id":"https://www.segmind.com/#organization","name":"Segmind","url":"https://www.segmind.com","logo":"https://www.segmind.com/favicon.svg"}}]}35:T9ab,Create a 15-second high-fashion editorial film starring the adult female character in @[char ref]. Preserve her exact face, hairstyle, body proportions, skin tone, outfit, accessories, and visual style. Use the reference as the ONLY character reference. Render the entire film in its aesthetic, with realistic skin, natural anatomy, and physically accurate hair and fabric. No anime, no cartoon, no outfit changes.

Minimal luxury studio, reflective floor, bold directional lighting, rhythmic photographic flashes, photorealistic fashion cinematography. Exactly 10 distinct fashion poses, each with a different silhouette and camera angle:

1. Confident standing pose, filmed from floor level with dramatic wide-angle perspective.
2. One hand beside her face, captured in an extreme side-profile close-up.
3. Asymmetric seated pose, seen directly overhead as the camera rotates.
4. One leg extended toward the lens, her reference footwear dominating the foreground through dramatic foreshortening.
5. Quick, fluid leg cross, captured from a low diagonal angle.
6. Fingertips catching the edge of an existing sleeve or collar, framed in a tight detail shot.
7. Rear three-quarter turn with a deliberate glance back into the camera.
8. Forward lean toward the lens, ending with her palm briefly covering it.
9. Elongated upward stretch, revealed in a sharply tilted wide shot with strong side lighting.
10. Commanding full-body hero pose, looking down into the lens as the camera rapidly pulls back.

Keep the performance fast, confident, and rhythmically tight. Each pose lasts only a fraction of a second while remaining clearly readable. No repeated poses or additional pose changes. Connect the angles through whip pans, movement-matched cuts, snap zooms, flashes, and lens occlusions. Orbit against her turn, sweep from her floor reflection toward her face, and plunge from overhead to floor level. Alternate intimate details with dramatic wide compositions. Accelerate between poses and brake sharply on each hold.

Keep all gestures quick, confident, and fully clothed. Hair and fabric react naturally to movement. No slow motion, lingering shots, or gradual camera drift. Build momentum toward the tenth pose, then stop sharply for a crisp final hold.

Stable identity, anatomy, and outfit throughout. No added accessories, no text, no logos, no AI artifacts. Sound: driving 140 BPM fashion beat, rhythmic shutter clicks, subtle fabric movement, and a sharp final beat.36:T2bbd,SYSTEM PROMPT â€” Seedance 2.5 Image-Adaptive Prompt Director

You are a cinematography prompt engineer for Seedance 2.5. You will receive two inputs in every task:

Prompt tab â€” a sample prompt written by the user. This is your structural and stylistic template: its shot grammar, its block order, its level of technical detail, its tone.
Input image tab â€” one or more reference images. These are your visual source of truth: subject identity, wardrobe, environment, palette, and mood.

Your job is not to copy the sample prompt's content and not to caption the image. Your job is to rebuild the sample prompt's structure and craft level around what the reference image actually shows, so the output is a new, fully-formed Seedance 2.5 prompt that could not have been written without either input.

CORE PRINCIPLE

You are a physics engine operator, not a mood board. Every line you write must resolve to a visible pixel or an audible sound. If a word in the sample prompt is a flourish that produces neither, drop it during adaptation rather than carrying it forward out of habit.

The reference image is read literally: count what is in frame, measure what can be measured (approximate height relationships, garment structure, material, light direction, camera height and angle, depth of field, colour temperature), and never invent details the image does not support. Where the sample prompt asks for an action or camera move the image can't confirm (because it's a still), extrapolate motion that is physically consistent with the pose, weight distribution, and implied momentum already visible in the frame â€” never a generic or clichÃ©d motion.

STEP ZERO â€” READ BOTH INPUTS BEFORE WRITING ANYTHING

A. Deconstruct the sample prompt's structure. Identify, in order:

Its block architecture (does it use headers, CRITICAL blocks, numbered assets, a geometry map, a shot list â€” or is it looser prose?)
Its register (technical/measurable vs. evocative â€” match it, but pull evocative language toward measurable wherever the reference image gives you the data to do so)
Its length and pacing (single shot vs. multi-shot, how many seconds, how many beats)
Any locked conventions it already uses (no on-screen text, no BGM, diegetic audio only, brand-name handling, etc.) â€” preserve these unless the user says otherwise

B. Deconstruct the reference image. Identify, in order:

Subject(s): count, approximate age register (role/appearance only, never assumed exact age), body build, pose, weight-bearing leg or point of balance, gaze direction, hand position, garment behaviour (taut, loose, mid-motion, static)
Wardrobe and identity markers: colour, material, fit, layering, accessories, hair colour/length/style
Environment: architecture, depth planes (foreground/midground/background), practical light sources visible in frame, surface materials, weather or atmosphere cues
Camera: apparent focal length/FOV, height relative to subject, angle (eye-level, low, high, Dutch), depth of field falloff
Light: direction, quality (hard/soft), colour temperature, practical sources vs. ambient fill
Anything in motion or implying motion (fabric, hair, dust, water, blur)

C. Confirm the target engine is Seedance 2.5. This governs two hard ceilings:

Up to 50 image references â€” use this budget deliberately. A single character can carry a front reference, a profile/detail reference, and a wardrobe reference as separate slots rather than one composite. A prop or environment plate gets its own slot rather than being folded into a character line. Do not pad the reference list with near-duplicates â€” if two references would teach the model the same thing, cut one.
Up to 30 seconds runtime â€” do not default to the ceiling. Build only the runtime the adapted action actually needs. If the sample prompt's action would run past 30 seconds when adapted, split into two prompts and say so.

If it is genuinely unclear whether the target is 2.0 or 2.5, ask once in one line before writing. Otherwise assume 2.5 per this system prompt and proceed.

THE ADAPTATION RULE

For every structural element present in the sample prompt, ask: "What does the reference image supply for this slot?"

If the image supplies real data (a pose, a garment, a light direction, a spatial layout) â€” replace the sample's placeholder content with the image's actual content, expressed at the same level of technical precision the sample used or higher.
If the image is silent on something the sample prompt specifies (e.g., a camera move, a line of dialogue, a sound cue) â€” retain the sample's original specification, adapted only enough to stay physically consistent with the image's geometry and framing.
If the image shows something the sample prompt has no slot for (a distinctive prop, a second figure, a strong light source) â€” add it as its own block or asset rather than dropping it silently. Reference images always win on visual fact.

Never let the sample prompt's original subject, wardrobe, or setting leak into the output once the reference image contradicts it. The image is the identity source; the sample prompt is the grammar source.

WRITE THE VISIBLE
Instead of	Write
she looks confident	chin level, shoulders back, weight settled evenly over both feet, gaze fixed forward
the room feels warm	one low-angled tungsten source camera-left, warm amber falloff across the near wall, cooler blue spill from a window at frame right
she moves quickly	strides at a pace that swings her coat hem clear of her calves, heel-strike audible on the hard floor
he looks taller than her	he stands roughly a head taller, her eyeline meeting his collarbone

Measurables to extract from the image and use in the prompt: approximate height relationships in cm where the image implies them Â· garment weight and how it would move (light chiffon vs. structured wool) Â· light direction and temperature in degrees/Kelvin-equivalent language Â· depth planes named explicitly (foreground/midground/background) Â· camera height and angle stated in FOV degrees and eye-level relationship Â· direction always labelled screen-relative or character-relative.

OUTPUT STRUCTURE â€” MATCH THE SAMPLE, ENFORCE THE SPINE

Use the sample prompt's own architecture as the visible shape of your output. Underneath that shape, silently enforce this priority order so nothing is missing regardless of how loose the sample was:

Header â€” shot count, runtime, cut policy, speed policy
Style prefix â€” render style, texture, skin, technical invariants carried from the image
No on-screen text (unless the image itself contains legible text/graphics â€” if so, transcribe it verbatim and describe it physically)
Critical blocks â€” anything from the image or sample that must not drift (max 4, ordered by importance)
Assets â€” one per character/prop/environment plate, each with identity + this-scene action + a fidelity assertion tying it to its reference
Geometry map â€” absolute frame position and depth planes, drawn from the image's actual composition
First frame â€” what is already happening at frame one, matching the reference image if it is the opening frame
Optics â€” lens lock in FOV degrees, matching the image's apparent focal length
Camera â€” physicality register and behaviour (handheld, locked, dolly, etc.)
Light & colour â€” direction, quality, temperature, drawn from the image
Atmosphere â€” air density, depth planes, any vapor bound to a visible source
Action timing â€” timecoded beats, physically consistent with the image's implied momentum
Physics â€” mass, deformation, rebound, lag, contact, gravity behaving normally
Acting â€” face, eyes, brow, matched to the expression actually visible in the reference
Audio â€” diegetic by default unless the sample specifies an attached track
Locks â€” positive ordered chain of what must hold across the whole shot

If the sample prompt does not use explicit block headers, keep the output in the sample's own prose style but ensure every one of the sixteen concerns above is addressed somewhere in the text â€” do not silently drop physics, geometry, or acting because the sample's register was casual.

HOUSE RULES (NON-NEGOTIABLE)
No character names in the prompt body â€” visual descriptors only (hair, wardrobe, identity markers). Names may appear only as aliases in a numbered reference list above the final prompt.
No aspect ratio in the body â€” that is a UI setting.
No internal production context ("carried from the previous scene," "as before") â€” every prompt stands alone.
No platform or tool names inside the prompt body.
No meta-commentary â€” every word describes something visible or audible.
Age-blind â€” describe by role, build, hair, wardrobe, never a stated age.
Brand names, text, and graphics visible in the reference image are written verbatim and described physically (shape, colour, placement, legibility) â€” paraphrasing renders a vague approximation.
English only in the final prompt.
Every fact lives in exactly one slot â€” do not restate wardrobe in Action Timing if it's already fixed in Assets; only mention a garment there when it is doing something visible (a hem swinging, hair whipping clear).
DELIVERY FORMAT

Every response has exactly three parts, nothing else:

A bolded title with runtime â€” e.g. **Rooftop entry â€” reference-matched â€” 8s**
A numbered reference list in attach order (character front/detail/wardrobe refs first, then props, then environment plate, then video/audio if any) â€” respecting the 50-slot ceiling on 2.5
One fenced code block containing the complete adapted prompt, English only, following the structure rules above

No preamble, no postamble. If the reference image conflicts with something explicit in the sample prompt (e.g. sample specifies a blue jacket, image shows red), flag the conflict in one line above the title and default to the image as the source of truth unless told otherwise.

A negative prompt block ships as an optional second code block only when there's a known drift risk (mixed identity cues, locked spatial geometry, a prop that could be mistaken for something else) â€” grouped by failure category, comma-separated, no full sentences.

LENGTH DISCIPLINE

A four-shot sequence with four assets should land at roughly 900â€“1,400 words. Past that, the critical blocks lose weight against the descriptive body. If the adapted prompt is running long, check for a duplicated fact before adding a trim pass â€” the same detail stated twice reads long without reading more specific.

PRE-DELIVERY CHECK
 Structure and register match the sample prompt's own architecture
 Every subject, wardrobe item, prop, and environment detail traces back to the reference image, not to assumption
 Any camera move or action beyond what the still shows is physically consistent with the pose, weight, and momentum visible in the image
 Reference count within 50, no near-duplicate references
 Runtime within 30s, or split declared
 No character names, no aspect ratio, no tool names, no restated facts
 Physics chain present: mass, contact, deformation â€” nothing floats or slides unnaturally
 Delivery is exactly title + reference list + one code block, nothing more37:T9ab,Create a 15-second high-fashion editorial film starring the adult female character in @[char ref]. Preserve her exact face, hairstyle, body proportions, skin tone, outfit, accessories, and visual style. Use the reference as the ONLY character reference. Render the entire film in its aesthetic, with realistic skin, natural anatomy, and physically accurate hair and fabric. No anime, no cartoon, no outfit changes.

Minimal luxury studio, reflective floor, bold directional lighting, rhythmic photographic flashes, photorealistic fashion cinematography. Exactly 10 distinct fashion poses, each with a different silhouette and camera angle:

1. Confident standing pose, filmed from floor level with dramatic wide-angle perspective.
2. One hand beside her face, captured in an extreme side-profile close-up.
3. Asymmetric seated pose, seen directly overhead as the camera rotates.
4. One leg extended toward the lens, her reference footwear dominating the foreground through dramatic foreshortening.
5. Quick, fluid leg cross, captured from a low diagonal angle.
6. Fingertips catching the edge of an existing sleeve or collar, framed in a tight detail shot.
7. Rear three-quarter turn with a deliberate glance back into the camera.
8. Forward lean toward the lens, ending with her palm briefly covering it.
9. Elongated upward stretch, revealed in a sharply tilted wide shot with strong side lighting.
10. Commanding full-body hero pose, looking down into the lens as the camera rapidly pulls back.

Keep the performance fast, confident, and rhythmically tight. Each pose lasts only a fraction of a second while remaining clearly readable. No repeated poses or additional pose changes. Connect the angles through whip pans, movement-matched cuts, snap zooms, flashes, and lens occlusions. Orbit against her turn, sweep from her floor reflection toward her face, and plunge from overhead to floor level. Alternate intimate details with dramatic wide compositions. Accelerate between poses and brake sharply on each hold.

Keep all gestures quick, confident, and fully clothed. Hair and fabric react naturally to movement. No slow motion, lingering shots, or gradual camera drift. Build momentum toward the tenth pose, then stop sharply for a crisp final hold.

Stable identity, anatomy, and outfit throughout. No added accessories, no text, no logos, no AI artifacts. Sound: driving 140 BPM fashion beat, rhythmic shutter clicks, subtle fabric movement, and a sharp final beat.38:T2bbd,SYSTEM PROMPT â€” Seedance 2.5 Image-Adaptive Prompt Director

You are a cinematography prompt engineer for Seedance 2.5. You will receive two inputs in every task:

Prompt tab â€” a sample prompt written by the user. This is your structural and stylistic template: its shot grammar, its block order, its level of technical detail, its tone.
Input image tab â€” one or more reference images. These are your visual source of truth: subject identity, wardrobe, environment, palette, and mood.

Your job is not to copy the sample prompt's content and not to caption the image. Your job is to rebuild the sample prompt's structure and craft level around what the reference image actually shows, so the output is a new, fully-formed Seedance 2.5 prompt that could not have been written without either input.

CORE PRINCIPLE

You are a physics engine operator, not a mood board. Every line you write must resolve to a visible pixel or an audible sound. If a word in the sample prompt is a flourish that produces neither, drop it during adaptation rather than carrying it forward out of habit.

The reference image is read literally: count what is in frame, measure what can be measured (approximate height relationships, garment structure, material, light direction, camera height and angle, depth of field, colour temperature), and never invent details the image does not support. Where the sample prompt asks for an action or camera move the image can't confirm (because it's a still), extrapolate motion that is physically consistent with the pose, weight distribution, and implied momentum already visible in the frame â€” never a generic or clichÃ©d motion.

STEP ZERO â€” READ BOTH INPUTS BEFORE WRITING ANYTHING

A. Deconstruct the sample prompt's structure. Identify, in order:

Its block architecture (does it use headers, CRITICAL blocks, numbered assets, a geometry map, a shot list â€” or is it looser prose?)
Its register (technical/measurable vs. evocative â€” match it, but pull evocative language toward measurable wherever the reference image gives you the data to do so)
Its length and pacing (single shot vs. multi-shot, how many seconds, how many beats)
Any locked conventions it already uses (no on-screen text, no BGM, diegetic audio only, brand-name handling, etc.) â€” preserve these unless the user says otherwise

B. Deconstruct the reference image. Identify, in order:

Subject(s): count, approximate age register (role/appearance only, never assumed exact age), body build, pose, weight-bearing leg or point of balance, gaze direction, hand position, garment behaviour (taut, loose, mid-motion, static)
Wardrobe and identity markers: colour, material, fit, layering, accessories, hair colour/length/style
Environment: architecture, depth planes (foreground/midground/background), practical light sources visible in frame, surface materials, weather or atmosphere cues
Camera: apparent focal length/FOV, height relative to subject, angle (eye-level, low, high, Dutch), depth of field falloff
Light: direction, quality (hard/soft), colour temperature, practical sources vs. ambient fill
Anything in motion or implying motion (fabric, hair, dust, water, blur)

C. Confirm the target engine is Seedance 2.5. This governs two hard ceilings:

Up to 50 image references â€” use this budget deliberately. A single character can carry a front reference, a profile/detail reference, and a wardrobe reference as separate slots rather than one composite. A prop or environment plate gets its own slot rather than being folded into a character line. Do not pad the reference list with near-duplicates â€” if two references would teach the model the same thing, cut one.
Up to 30 seconds runtime â€” do not default to the ceiling. Build only the runtime the adapted action actually needs. If the sample prompt's action would run past 30 seconds when adapted, split into two prompts and say so.

If it is genuinely unclear whether the target is 2.0 or 2.5, ask once in one line before writing. Otherwise assume 2.5 per this system prompt and proceed.

THE ADAPTATION RULE

For every structural element present in the sample prompt, ask: "What does the reference image supply for this slot?"

If the image supplies real data (a pose, a garment, a light direction, a spatial layout) â€” replace the sample's placeholder content with the image's actual content, expressed at the same level of technical precision the sample used or higher.
If the image is silent on something the sample prompt specifies (e.g., a camera move, a line of dialogue, a sound cue) â€” retain the sample's original specification, adapted only enough to stay physically consistent with the image's geometry and framing.
If the image shows something the sample prompt has no slot for (a distinctive prop, a second figure, a strong light source) â€” add it as its own block or asset rather than dropping it silently. Reference images always win on visual fact.

Never let the sample prompt's original subject, wardrobe, or setting leak into the output once the reference image contradicts it. The image is the identity source; the sample prompt is the grammar source.

WRITE THE VISIBLE
Instead of	Write
she looks confident	chin level, shoulders back, weight settled evenly over both feet, gaze fixed forward
the room feels warm	one low-angled tungsten source camera-left, warm amber falloff across the near wall, cooler blue spill from a window at frame right
she moves quickly	strides at a pace that swings her coat hem clear of her calves, heel-strike audible on the hard floor
he looks taller than her	he stands roughly a head taller, her eyeline meeting his collarbone

Measurables to extract from the image and use in the prompt: approximate height relationships in cm where the image implies them Â· garment weight and how it would move (light chiffon vs. structured wool) Â· light direction and temperature in degrees/Kelvin-equivalent language Â· depth planes named explicitly (foreground/midground/background) Â· camera height and angle stated in FOV degrees and eye-level relationship Â· direction always labelled screen-relative or character-relative.

OUTPUT STRUCTURE â€” MATCH THE SAMPLE, ENFORCE THE SPINE

Use the sample prompt's own architecture as the visible shape of your output. Underneath that shape, silently enforce this priority order so nothing is missing regardless of how loose the sample was:

Header â€” shot count, runtime, cut policy, speed policy
Style prefix â€” render style, texture, skin, technical invariants carried from the image
No on-screen text (unless the image itself contains legible text/graphics â€” if so, transcribe it verbatim and describe it physically)
Critical blocks â€” anything from the image or sample that must not drift (max 4, ordered by importance)
Assets â€” one per character/prop/environment plate, each with identity + this-scene action + a fidelity assertion tying it to its reference
Geometry map â€” absolute frame position and depth planes, drawn from the image's actual composition
First frame â€” what is already happening at frame one, matching the reference image if it is the opening frame
Optics â€” lens lock in FOV degrees, matching the image's apparent focal length
Camera â€” physicality register and behaviour (handheld, locked, dolly, etc.)
Light & colour â€” direction, quality, temperature, drawn from the image
Atmosphere â€” air density, depth planes, any vapor bound to a visible source
Action timing â€” timecoded beats, physically consistent with the image's implied momentum
Physics â€” mass, deformation, rebound, lag, contact, gravity behaving normally
Acting â€” face, eyes, brow, matched to the expression actually visible in the reference
Audio â€” diegetic by default unless the sample specifies an attached track
Locks â€” positive ordered chain of what must hold across the whole shot

If the sample prompt does not use explicit block headers, keep the output in the sample's own prose style but ensure every one of the sixteen concerns above is addressed somewhere in the text â€” do not silently drop physics, geometry, or acting because the sample's register was casual.

HOUSE RULES (NON-NEGOTIABLE)
No character names in the prompt body â€” visual descriptors only (hair, wardrobe, identity markers). Names may appear only as aliases in a numbered reference list above the final prompt.
No aspect ratio in the body â€” that is a UI setting.
No internal production context ("carried from the previous scene," "as before") â€” every prompt stands alone.
No platform or tool names inside the prompt body.
No meta-commentary â€” every word describes something visible or audible.
Age-blind â€” describe by role, build, hair, wardrobe, never a stated age.
Brand names, text, and graphics visible in the reference image are written verbatim and described physically (shape, colour, placement, legibility) â€” paraphrasing renders a vague approximation.
English only in the final prompt.
Every fact lives in exactly one slot â€” do not restate wardrobe in Action Timing if it's already fixed in Assets; only mention a garment there when it is doing something visible (a hem swinging, hair whipping clear).
DELIVERY FORMAT

Every response has exactly three parts, nothing else:

A bolded title with runtime â€” e.g. **Rooftop entry â€” reference-matched â€” 8s**
A numbered reference list in attach order (character front/detail/wardrobe refs first, then props, then environment plate, then video/audio if any) â€” respecting the 50-slot ceiling on 2.5
One fenced code block containing the complete adapted prompt, English only, following the structure rules above

No preamble, no postamble. If the reference image conflicts with something explicit in the sample prompt (e.g. sample specifies a blue jacket, image shows red), flag the conflict in one line above the title and default to the image as the source of truth unless told otherwise.

A negative prompt block ships as an optional second code block only when there's a known drift risk (mixed identity cues, locked spatial geometry, a prop that could be mistaken for something else) â€” grouped by failure category, comma-separated, no full sentences.

LENGTH DISCIPLINE

A four-shot sequence with four assets should land at roughly 900â€“1,400 words. Past that, the critical blocks lose weight against the descriptive body. If the adapted prompt is running long, check for a duplicated fact before adding a trim pass â€” the same detail stated twice reads long without reading more specific.

PRE-DELIVERY CHECK
 Structure and register match the sample prompt's own architecture
 Every subject, wardrobe item, prop, and environment detail traces back to the reference image, not to assumption
 Any camera move or action beyond what the still shows is physically consistent with the pose, weight, and momentum visible in the image
 Reference count within 50, no near-duplicate references
 Runtime within 30s, or split declared
 No character names, no aspect ratio, no tool names, no restated facts
 Physics chain present: mass, contact, deformation â€” nothing floats or slides unnaturally
 Delivery is exactly title + reference list + one code block, nothing more39:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.3a:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.3b:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.3c:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.3d:T7d8,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 di3e:T1ce3,15-second high-fashion editorial film, exactly 10 shots and 10 distinct held poses. Hard cuts and concealed movement-matched cuts only; accelerate camera travel between poses and brake sharply for each readable hold. Real-time performance, no slow motion, no additional poses. Finish on a motionless final frame.

Use @Image1 as the ONLY character and visual reference. Preserve the womanâ€™s exact facial structure, warm olive skin tone, long dark-brown loosely waved hair, body proportions, and photographic rendering. Retain realistic pores, natural anatomy, fine flyaway strands, and sharply resolved material textures.

Preserve the complete reference outfit: burgundy ribbed turtleneck; glossy oxblood leather wrap miniskirt with its existing side tie; matching long leather coat draped over her shoulders with the sleeves hanging unoccupied; burgundy tights; pointed-toe, knee-high burgundy snakeskin-pattern boots with slender heels; small thick gold hoop earrings. Hands emerge through the coatâ€™s open front, never through spontaneously occupied sleeves. No added accessories, outfit changes, text, or logos.

Use the referenceâ€™s limestone colonnade rather than a replacement studio. A large cream cylindrical column and its stepped circular base occupy screen-right; the woman stands immediately screen-left of it. Receding columns occupy the left background, framing a nearly black recessed passage. Pale stone paving fills the foreground, with visible joints and restrained surface sheen, not a mirror reflection. Keep the original architectural scale and spatial relationships.

Hard directional sunlight enters from upper screen-left, approximately 4800â€“5200 K, producing crisp oblique floor shadows, bright narrow leather highlights, and deep brown-black recesses. Brief photographic flashes add neutral-white pulses without replacing this baseline illumination. Warm ivory, oxblood, muted gold, and dark brown dominate. Clear air; no smoke, vapor, or suspended particles.

Exactly 10 distinct fashion poses, each with a different silhouette and camera angle. FOV values below are horizontal and locked within each shot except the specified optical snap zoom. Camera travel uses controlled low-mounted, overhead, and arcing moves, never gradual handheld drift.

1. 0.00â€“1.50 â€” Standing pose, floor-level wide angle, 76Â° FOV. Begin with the referenceâ€™s shoulder-near-column position, boots lightly crossed, coat hanging vertically, chin lifted, lips slightly parted, and eyes directed down toward the lens. Camera approximately 15 cm above the paving. Sweep briskly from the diagonal floor shadow to her face, then arrest the tilt. Hold the silhouette for 0.45 seconds. A short shutter flash conceals the cut.

2. 1.50â€“2.85 â€” Hand beside face, extreme side-profile close-up, 26Â° FOV. Cut to head height on her screen-left side. She brings one hand through the front opening and stops its fingertips beside her cheek without touching the earring. Nose, lips, lashes, and jaw form a clean profile against the dark passage. The eye nearest camera tracks toward the lens without turning her head. Hold 0.40 seconds, then whip downward.

3. 2.85â€“4.35 â€” Asymmetric seated pose, directly overhead, 64Â° FOV. Movement-matched cut to her seated on the stone paving beside the column base, one knee bent close and the other angled outward; one palm supports her weight behind the hip. Keep the skirt covering her pelvis and upper thighs. The overhead camera rotates 30Â° clockwise, brakes, and holds for 0.40 seconds. Coat panels spread only where they contact the floor.

4. 4.35â€“5.85 â€” Boot-forward extension, ground-level close wide, 90Â° FOV. Cut on a downward whip to a lens approximately 12 cm above the floor; do not simulate a camera passing through her body. From the seated support, she extends one boot toward the lens, its pointed toe dominating the lower foreground while her face remains visible above it. Knee stays slightly flexed, sole directed obliquely rather than flattened toward camera. Hold 0.40 seconds.

5. 5.85â€“7.20 â€” Quick standing leg cross, low diagonal angle, 68Â° FOV. Flash-cut to an upright starting position, omitting the rise. Camera sits at knee height, offset toward screen-left with a 15Â° roll. She transfers weight into the supporting boot, draws the other leg across, and plants the toe decisively. One crossing action only. Hold 0.35 seconds; the coat hem completes a short delayed swing. Whip toward her upper torso.

6. 7.20â€“8.55 â€” Existing collar detail, tight oblique close-up, 22Â° FOV. Her fingertips catch the edge of the coatâ€™s screen-left lapel through the open front. A short optical snap zoom isolates the pinch, stitching, ribbed knit, and leather highlight. The leather bends locally under pressure; she retains the pinch for the hold rather than performing a second gesture. Shutter click and lateral cut.

7. 8.55â€“10.05 â€” Rear three-quarter glance, shoulder-height medium shot, 46Â° FOV. She pivots through her planted feet into a rear three-quarter silhouette and turns her head just far enough to meet the lens. Camera arcs 25Â° counter to her turn, maintaining clearance from the column. Hair trails the pivot, briefly exposing one hoop. Arrest both turn and camera; hold 0.40 seconds.

8. 10.05â€“11.40 â€” Forward lean with palm occlusion, chest-height frontal close wide, 80Â° FOV. Cut to her facing the lens. With knees softly unlocked and hips counterbalancing, she inclines her torso forward and extends one open palm. Face remains undistorted until the hand reaches the near foreground. Palm fully covers the lens for the last 0.12 seconds, concealing the next cut.

9. 11.40â€“13.05 â€” Elongated upward stretch, tilted full-body wide, 74Â° FOV. Reveal a 28Â° clockwise Dutch angle from waist height. Both arms rise through the coatâ€™s front opening, elbows lengthening overhead while her shoulders keep the draped coat supported. Feet remain grounded. Strong side light separates her extended outline from the recess. Hold 0.45 seconds; cut on a flash, without showing an intermediate pose.

10. 13.05â€“15.00 â€” Full-body hero pose, floor-level frontal wide, 70Â° FOV. Feet planted apart, torso upright, arms lowered clear of the waist, chin slightly tucked, eyes looking down directly into the lens. Camera rapidly dollies backward from 1.8 to 3.5 meters, revealing the entire silhouette and column base without changing focal length. Stop by 14.40. Hold completely still through 15.00, ending on a sharp final beat.

Across all shots, heels and toes establish contact before taking weight; no foot sliding or unsupported balance. Leather carries weight, forms broad folds, and settles with limited rebound. Knit stretches locally, tights remain continuous, and hair lags behind acceleration before settling. Coat placement stays supported at the shoulders throughout.

Keep brows relaxed, mouth controlled, and gaze precise; no smile changes or exaggerated expressions. Maintain identity, garment construction, architectural geometry, lighting direction, grounded anatomy, and material continuity in that order.

Sound: driving 140 BPM fashion beat, tightly synchronized shutter clicks at selected flashes and cuts, quiet leather creaks and fabric movement, brief heel contacts on stone. No dialogue. Final musical accent at 14.40, with its short decay ending cleanly at 15.00.3f:T642,This workflow generates a 15-second high-fashion editorial film, featuring structured fashion poses and cinematic editing techniques. It is especially beneficial for fashion brands, creative agencies, and videographers who require quick and visually appealing content for promotional needs. The key AI models powering this workflow are Seedance 2.5 for image-to-video conversion and GPT-6 Astra for text generation. The primary value offered by this workflow is the ability to automate complex video production processes, ensuring cost-effectiveness and rapid content generation.

---

## 3. How does it work
- The workflow begins with GPT-6 Astra, which interprets an input image and generates a detailed cinematographic prompt. This AI model understands complex instructions and uses reference images to maintain character consistency.
- The Seedance 2.5 model takes the generated script and transforms it into a video, providing cinematic shot transitions and synchronized audio, ensuring a seamless and professional 15-second fashion b-roll.

---

## 4. How to customize it
Users can adapt the workflow to suit different creative visions by modifying the prompts fed into GPT-6 Astra. For instance, a user might choose to adjust the scenarios or poses to reflect their brandâ€™s unique aesthetic. Alternatively, they could replace the reference image to feature different models or wardrobes, instantly shifting the videoâ€™s focal point to meet their specific campaign needs.

---

## 5. Who is it for
- Fashion Designers and Brands
- Marketing Agencies
- Visual Content Creators and Videographers40:T642,This workflow generates a 15-second high-fashion editorial film, featuring structured fashion poses and cinematic editing techniques. It is especially beneficial for fashion brands, creative agencies, and videographers who require quick and visually appealing content for promotional needs. The key AI models powering this workflow are Seedance 2.5 for image-to-video conversion and GPT-6 Astra for text generation. The primary value offered by this workflow is the ability to automate complex video production processes, ensuring cost-effectiveness and rapid content generation.

---

## 3. How does it work
- The workflow begins with GPT-6 Astra, which interprets an input image and generates a detailed cinematographic prompt. This AI model understands complex instructions and uses reference images to maintain character consistency.
- The Seedance 2.5 model takes the generated script and transforms it into a video, providing cinematic shot transitions and synchronized audio, ensuring a seamless and professional 15-second fashion b-roll.

---

## 4. How to customize it
Users can adapt the workflow to suit different creative visions by modifying the prompts fed into GPT-6 Astra. For instance, a user might choose to adjust the scenarios or poses to reflect their brandâ€™s unique aesthetic. Alternatively, they could replace the reference image to feature different models or wardrobes, instantly shifting the videoâ€™s focal point to meet their specific campaign needs.

---

## 5. Who is it for
- Fashion Designers and Brands
- Marketing Agencies
- Visual Content Creators and Videographers41:T6a5,The Cinematic Costume Reveal workflow is designed to create a high-resolution, 15-second cinematic video commercial showcasing luxury fashion garments without human models. This workflow is particularly beneficial for fashion marketers, designers, and advertising agencies who need to produce high-quality fashion content rapidly. Utilizing advanced AI models like OpenAI Sora and Runway Gen-3 Alpha, this workflow automates the creation of fashion clips with exquisite detail, enabling faster content generation and cost savings by eliminating the need for human models and traditional production setups.

---

## 3. How does it work  
- The workflow begins with an input image of a fashion garment which is used as a reference.
- The GPT-6 Astra model generates a detailed prompt describing a fluid, multi-angle shooting strategy to visualize the garment cinematically.
- Luma Dream Machine model performs fabric simulation to accurately render cloth physics.
- Seedance 2.5 model generates the final video output, utilizing advanced 4K rendering and cinematic camera effects.

---

## 4. How to customize it  
Users can customize this workflow by modifying the input prompts or replacing the garment reference image to fit specific campaign needs. For instance, changing the color descriptors or fabric types in the prompt will adjust the visual outcome to showcase various materials like silk or cotton. Additionally, adapting the camera angle descriptions can help emphasize unique design elements of different fashion pieces.

---

## 5. Who is it for  
- Fashion marketers and advertisers
- Fashion designers and brand managers
- Content creators and film directors in the fashion industry

---42:T76d,The AI-Powered Fashion B-Roll Creator is a comprehensive workflow designed to generate high-quality, fashion-forward video clips for use in campaigns, editorials, or social media. This workflow is ideal for marketers, content creators, and fashion designers who need to create visually engaging content with minimal effort. Leveraging advanced AI models such as Wan 3.0 for video creation and GPT Image 2 for detailed image processing, this workflow enables efficient content generation with professional results. The primary value lies in its ability to automate the complex process of video production, offering rapid, scalable, and cost-effective content generation.

---

## 3. How does it work

- The workflow uses the Wan 3.0 model to create videos based on input images and prompts, generating cinematic fashion snippets.
- It incorporates the GPT Image 2 model to produce photorealistic images for reference, ensuring that video outputs align with predefined visual styles.
- Claude 4.5 Sonnet is used to draft detailed prompts for video and image generation, enhancing content specificity and consistency.

---

## 4. How to customize it

Users can customize this workflow to fit specific fashion narratives or branding requirements by altering the prompt inputs. For example, changing the stylistic prompts in the Claude 4.5 Sonnet model can create videos focused on different fashion themes such as vintage, modern punk, or haute couture. Additionally, swapping input images with ones aligning better with a different fashion style or season can adapt the final video output to suit various campaigns.

---

## 5. Who is it for

- Fashion designers looking to create digital portfolios showcasing their collections.
- Marketing agencies aiming to automate campaign video production for clients.
- Social media content creators who require visually appealing fashion content for their feeds.43:T6dc,This workflow seamlessly generates high-quality, AI-powered lifestyle fashion videos for vloggers. It is particularly useful for content creators, fashion bloggers, and marketing professionals looking to produce captivating and premium fashion-related content effortlessly. Utilizing advanced AI models such as Gemini Omni Flash for text-to-video translation and GPT Image 2 for photorealistic image generation, it streamlines the process of creating engaging visual narratives with professional appeal. The primary value lies in its ability to automate video production, ensuring faster content generation and cost savings while maintaining premium quality.

## 3. How does it work
- The workflow utilizes the *Claude 4.5 Sonnet* model to create detailed storyboard prompts for video scenes, ensuring narrative consistency and visual storytelling.
- Leverages *GPT Image 2* to generate high-definition images with specific fashion and retail themes, acting as visual input for video production.
- Employs *Gemini Omni Flash* to transform text and images into synchronized video content with lifelike audio, maintaining cinematic and photorealistic quality.

## 4. How to customize it
Users can customize this workflow by altering the prompts used at various stages. For example, modifying the descriptive text in the *Claude 4.5 Sonnet* model can tailor the storyboard to different fashion themes or brands. Similarly, using different image URLs in *GPT Image 2* can adapt the visual style to match specific aesthetics or brand identities. This flexibility allows users to cater to diverse stylistic preferences and storytelling needs.

## 5. Who is it for
- Fashion Bloggers and Influencers
- Digital Content Creators
- Marketing and Advertising Agencies44:T69f,This workflow automates the creation of hyper-realistic, commercial-quality product photos with a fisheye aesthetic. It is perfect for marketers, content creators, and photographers who want to produce engaging product shots without extensive manual editing. Leveraging advanced AI models such as Claude 4.5 Sonnet and GPT Image 2, it transforms basic product images into dynamic, lifestyle-oriented visuals. This automated process significantly reduces the time and cost associated with traditional photo shoots, enabling faster content generation at scale.

## 3. How does it work
- The workflow begins with an input node that accepts an image, which serves as the base for processing.
- It utilizes the Claude 4.5 Sonnet model from ANTHROPIC, which handles the creative composition and styling, preparing the image with detailed prompts for enhancements.
- Following this, the GPT Image 2 model from OPENAI generates the final photorealistic output by applying the prompted enhancements, achieving a vibrant, motion-blurred urban setting with a fisheye lens effect.

## 4. How to customize it
Users can easily customize this workflow by altering the prompt inputs within the Claude 4.5 Sonnet node to suit various product themes or brand aesthetics. For example, adjusting the lighting description can suit different environmental moods (e.g., night shoots with neon effects), while changing the background elements can fit different urban or rural scenes. The ability to tweak these specifications allows the workflow to adapt to diverse creative needs and promotional campaigns.

## 5. Who is it for
- Digital Marketing Agencies
- E-commerce Platform Developers
- Professional Photographers45:T7e7,This AI-powered workflow enables the generation of high-quality, cinematic b-roll videos centered around luxury lifestyle products, specifically evening gowns. Utilizing advanced AI models such as the "Seedance 2.0" for video generation and "Nano Banana 2" for creating photorealistic images, this workflow is a boon for marketers, fashion designers, and content creators. It automates the complex process of video production with precision, adhering strictly to product design and cinematic standards, thereby significantly reducing production costs and time.

## 3. How does it work  
- The workflow starts with an input image representing the product to be featured, which in this case is a luxury evening gown.
- The "Nano Banana 2" model generates high-quality photorealistic images from the input, ensuring consistency with the product's reference design.
- The generated images are fed into the "Seedance 2.0" model, which creates a detailed cinematic video narrative, maintaining fidelity to fabric physics, lighting conditions, and framing aesthetics.
- Finally, the output is a finely crafted video complete with synchronized audio elements, offering a luxurious portrayal of the product.

## 4. How to customize it  
Users can customize this workflow by adjusting prompts to highlight different aspects of the product or to suit various thematic needs. For example, changing the lighting conditions in prompts can give the video a different mood, from a lively summer day to an intimate evening setting. Additionally, altering the reference images and video prompts allows users to tailor the final video to different styles or seasonal collections, such as a spring floral pattern instead of a winter evening gown.

## 5. Who is it for  
- Fashion Brands: To showcase new collections in an elegant and engaging manner.
- Marketing Agencies: For creating compelling promotional content for luxury products.
- Video Production Studios: To enhance production efficiency and creativity through AI-driven automation.46:T785,This workflow is designed to create a cinematic video showcasing fashion outfits through a sequence of professional, artistic shots. It is ideal for fashion designers, marketers, content creators, and fashion bloggers looking to transform their fashion photography into engaging video narratives. The key AI models underlying this workflow include a photorealistic image generator, a sophisticated AI prompt engineer, and an advanced video generation engine. The primary value it offers is the seamless creation of high-quality fashion videos that can enhance digital marketing strategies, save time, and reduce production costs.

## 3. How does it work  
- The workflow employs the Nano Banana 2 model to generate detailed and realistic fashion imagery based on user input images and descriptive prompts.
- The Claude Sonnet model then acts as an AI video prompt engineer, crafting an elaborate shot-by-shot video prompt that combines the visual details extracted from the imagery.
- Finally, the Seedance 2.0 model uses these prompts and additional reference images to generate a cohesive AI-powered video, including synchronized audio.

## 4. How to customize it  
Users can adapt the workflow to their specific needs by altering the input images and modifying the descriptive text prompts. For instance, changing the environment description in the prompt can lead to videos set in different backdrops such as a bustling urban street or a serene natural setting. Additionally, users can modify the style descriptions to reflect different fashion themes or seasonal collections, thereby tailor-making the video output to meet diverse marketing or artistic goals.

## 5. Who is it for  
- Fashion marketers aiming to create captivating promotional content.
- Digital content creators looking for high-quality fashion video production.
- Fashion designers who want to visualize and present their collections in a dynamic way.47:T759,The AI-Powered Product UGC Reel Creator is a sophisticated workflow that generates dynamic user-generated content (UGC) reels for promotional Instagram or social media ads featuring luxury products. This workflow is particularly useful for marketers, product managers, and creative directors who seek to enhance brand visibility and engage audiences through visually compelling narratives. It leverages key AI models, including the Seedance 2.0 for cinematic AI video creation and GPT Image 2 for generating high-quality image references, supplemented by the powerful GPT 5.4 model for script refinement and prompt generation. The primary value of this workflow lies in its ability to automate the creation of high-quality, impactful content rapidly, allowing brands to scale their visual storytelling efforts efficiently.

## 3. How does it work
- The workflow uses the Seedance 2.0 model to translate narrative scripts into cinematic AI-produced videos and provides synchronized audio with the visuals.
- GPT Image 2 helps generate visually consistent and photorealistic image references that integrate smoothly into video content.
- GPT 5.4 facilitates in creating detailed, luxurious video prompts and content scripts that guide the entire creative production, preserving brand aesthetics and product features.

## 4. How to customize it
Users can adapt this workflow to different use cases by modifying the prompt inputs based on the desired video content style and tone. For instance, by altering the specific characteristics of the product and model images used in the initial setup, users can focus the narrative on different product lines or visual aesthetics. Additionally, adjusting the duration and aspect ratio settings allows tailoring the output for a variety of social media platforms.

## 5. Who is it for
- Marketing Agencies
- Creative Directors
- Brand Managers48:T693,This AI-powered workflow creates stunning exploded view videos of products, showcasing each component in a visually striking manner. It's perfect for marketers, product designers, and content creators seeking innovative ways to present products. The workflow utilizes advanced AI models like GPT Image 2 for image processing and Seedance 2.0 for video creation. By automating the generation of complex product visuals, it enables faster content production and enhances product presentation for campaigns and presentations.

## 3. How does it work
- The workflow begins by taking an input image of a product to generate a high-quality hero shot using the GPT Image 2 model.
- This hero shot, along with the exploded view image, serves as the basis for creating an exploded view using another instance of GPT Image 2.
- Seedance 2.0 is then employed to construct a 10-second cinematic animation that opens with the product hero shot, transitions smoothly to demonstrate the product being deconstructed into an exploded view, and concludes by reassembling the product.
- The final output is a loop-ready video that captures the product in a dynamic and visually compelling format.

## 4. How to customize it
Users can easily adapt this workflow to fit their unique requirements by altering the input prompts to tailor the visual style or modify the explosion sequence to highlight specific product features. For instance, a marketer could adjust the prompts to match a campaign's color scheme or a product designer could emphasize certain elements by changing the component details in the exploded view.

## 5. Who is it for
- Product Marketers
- Content Creators
- Industrial Designers49:T84c,The AI Clothing Product Broll Video Generator workflow creates a high-quality, cinematic video focused on showcasing a clothing product, in this case, a gown. Tailored for marketers, fashion designers, and ecommerce businesses, this workflow enhances the presentation of fashion products with a crisp, editorial aesthetic. Leveraging advanced AI models like ByteDance's Seedance 2.0, it automates the generation of cinematic broll featuring slow, deliberate camera movements, synchronized audio, and fine lighting details. This platform streamlines content creation, reducing costs and time, while maintaining the quality required for luxury product promotions.

## 3. How does it work  
- The workflow utilizes ByteDance's Seedance 2.0 model, renowned for its ability to generate cinematic-quality videos from inputs like text and images.
- It uses a detailed prompt specifying shot sequence, camera angles, lighting, and elements of the gown that need precision reproduction to ensure each segment aligns with luxury editorial standards.
- The input node connects to a reference image of the gown, providing a visual guide for the Seedance 2.0 model to ensure fabric textures and colors are accurately replicated.
- The final output is a sophisticated video presenting the clothing item in styled, narrative-driven scenes.

## 4. How to customize it  
Users can adjust the workflow's output by modifying the prompt parameters to suit different products or campaign themes. For example, changing the lighting instructions can create moods ranging from bright, airy displays to more dramatic presentations. New reference images can be introduced for different attire to adapt the video output to various collections or styles. Additionally, users can fine-tune the shot sequence or durations to fit specific marketing needs or platform requirements.

## 5. Who is it for  
- Fashion marketers looking to create engaging product videos.
- E-commerce brands aiming to elevate product listings with luxury video content.
- Creative teams in fashion houses needing to showcase new collections efficiently and with high fidelity.4a:T798,This AI-powered workflow allows users to create high-quality virtual videos showcasing fashion garments on models using user-generated content (UGC). The workflow is particularly beneficial for marketers and influencers on social media platforms who wish to demonstrate clothing items through visually appealing, influencer-style videos. By utilizing advanced AI models such as ByteDance's Seedance 2.0, this workflow brings together image-to-video conversion capabilities and cinematic aesthetics. The primary value lies in its ability to automate the creation of influencer-like fashion clips, ensuring accuracy in representation while saving time and reducing costs associated with traditional video shoots.

## 3. How does it work  
- The workflow leverages ByteDance's Seedance 2.0 model, which excels in creating cinematic AI videos with native audio and multi-shot narratives.
- Input nodes are used to provide reference images of the model and the garment, which are then processed by the AI to ensure photographic accuracy and consistency across the video.
- The designed prompt specifies detailed visual and narrative instructions that guide the video creation, ensuring the output adheres to specific storytelling and visual stylization.

## 4. How to customize it  
Users can customize this workflow by altering the reference images to match different models and garments, aligning with specific brand requirements or fashion lines. The prompt can also be modified to change the narrative style, for instance, adjusting the dialogue and shot composition to fit different social media trends or branding messages. For example, by swapping the gown for another fashion item or adjusting the setting to an outdoor location, the workflow can support diverse themes and campaigns.

## 5. Who is it for  
- Fashion Brands and Retailers
- Social Media Influencers and Content Creators
- Marketing Agencies specializing in digital campaigns4b:T709,This AI-powered workflow enables the creation of a high-paced, cinematic bullet-time fashion video using reference images to maintain faithful stylistic and content alignment. Ideal for filmmakers, fashion marketers, and digital content creators, the workflow utilizes advanced AI models such as Nano Banana 2 and Seedance 2.0 to ensure high photorealism and engaging narratives. The primary value it offers is generating professional-grade fashion films quickly and with lower production costs by automating several aspects of the filmmaking process while maintaining meticulous attention to reference details.

## 3. How does it work  
- **Nano Banana 2:** This model facilitates the creation of high-quality, photorealistic images that serve as a critical foundation for video scenes, preserving intricate details and reference styling.
- **Seedance 2.0:** Converts the processed images into compelling fashion film sequences using bullet-time cinematography fused with bolt-action movement, maintaining a crisp, high-speed visual feel throughout the video.

## 4. How to customize it  
Users can tailor this workflow by altering the prompts and reference images to align with specific video styles or branding needs. For instance, by changing the prompt details, one can shift the focus from fashion to other domains such as technology demonstration or artistic performances. Additionally, experimenting with different camera configurations or lighting instructions can provide diverse creative outputs.

## 5. Who is it for  
- **Fashion Filmmakers** looking to produce cutting-edge fashion videos.
- **Digital Content Creators** seeking to enhance their visual storytelling with cinematic flair.
- **Marketing Agencies** aiming to boost engagement through visually captivating fashion campaigns.4c:T7e4,This workflow enables users to create high-quality, photorealistic commercial product advertisements by combining text and image inputs in a streamlined process. It leverages advanced AI technologies to automate the creation of visually appealing ads that effectively showcase products. This solution is particularly useful for marketers, advertisers, and creative agencies looking to produce crisp and engaging product visuals with efficiency. Key AI models utilized include GPT 5.4 for text generation and image processing, and GPT Image 2 for text-to-image conversion. The primary value it provides lies in faster ad creation, reduced production costs, and the ability to scale media output seamlessly.

## 3. How does it work  
- **Input Nodes:** The workflow starts with input nodes that provide a text description for the ad prompt and image URLs for the product and model.
- **GPT 5.4 Model:** This model refines the text prompt, ensuring it is well-structured and suitable for creating a minimalist commercial fashion ad.
- **GPT Image 2 Model:** Utilizes the refined prompt and image URLs to generate a high-quality, photorealistic commercial ad with a focus on aesthetic appeal and clarity.
- **Output Node:** The final visual ad is then outputted, ready for use in marketing and advertising contexts.

## 4. How to customize it  
Users can adapt this workflow by modifying the text prompts or substituting different product or model images to suit their specific advertising needs. For example, a user could change the product imagery and text to focus on a different product category, such as electronics or home goods. Additionally, by tweaking the style or substance of the prompts, users can tailor the output to align with brand-specific guidelines or campaign themes, enhancing the ad's relevance to target audiences.

## 5. Who is it for  
- Marketing and advertising agencies
- Freelance graphic designers and content creators
- E-commerce businesses looking to enhance product presentation and appeal4d:T78e,The AI Image Upscaler workflow enhances the resolution and quality of an input image, transforming it into a high-fidelity, photorealistic output. It is particularly useful for professionals such as marketers, photographers, and digital content creators who need high-quality visuals for commercial use. The key technologies driving this workflow are cutting-edge AI models that focus on image upscaling, detail enhancement, and optical quality improvement. By automating the enhancement process, this workflow saves time and reduces the cost associated with manual image editing while ensuring consistent, top-tier visual output.

---

## 3. How does it work

- The workflow utilizes the "Nano Banana 2" model, which excels in creating fast, photorealistic images suitable for marketing and advertising.
- The input image undergoes a sophisticated enhancement process using predefined prompts that maintain facial and structural consistency while enhancing resolution to 8K quality.
- Parameters such as aspect ratio, output format, and resolution are customizable, enabling precise control over the final output.

---

## 4. How to customize it

Users can tailor the AI Image Upscaler to their specific needs by modifying the prompts and model parameters. For example, adjusting the aspect ratio can cater to specific display requirements like 16:9 for cinematic visuals. Similarly, changing the output resolution to 4K or 2K can optimize the image for different platforms or presentations. Users can also alter the prompt to focus on specific elements such as lighting, texture, or color grading to fit their unique project requirements.

---

## 5. Who is it for

- Marketing agencies that require high-quality images for promotional materials.
- Photographers aiming to enhance the resolution and detail of their photographs.
- Digital content creators looking to improve the visual quality of their media for viewers or clients.4e:T747,The AI-Powered Viral Dance Reel Creator is an advanced workflow designed to transform static images into dynamic, engaging dance videos. By employing state-of-the-art AI models, it enables creators, marketers, and artists to effortlessly produce high-quality video content. This workflow leverages technologies such as Seedance 2.0 for cinematic video creation and GPT Image 2 for static image processing. The key value it provides is the automation of video production, significantly reducing the time and effort required to create vibrant content perfect for social media virality.

## 3. How does it work  
- **Input Node**: Begins with an image that serves as the choreographic reference and subject identity throughout the video.
- **GPT Image 2**: Processes the input image to generate a precise dance sequence grid, maintaining character likeness and pose accuracy.
- **Seedance 2.0**: Converts the choreography grid into a live-action video, replicating the dance steps and maintaining the subjectâ€™s visual integrity.
- **Output Node**: Delivers a high-definition dance video ready for immediate use.

## 4. How to customize it  
Users can customize this workflow by modifying the prompt parameters in the Seedance 2.0 and GPT Image 2 models. For example, adjusting the choreography details or outfit settings in the prompt can tailor the resulting video to fit different styles or themes. Additionally, users can change the input images to reflect different subjects or environments, ensuring the video aligns with specific project goals or brand requirements.

## 5. Who is it for  
- Influencers and content creators looking to enhance their social media presence with eye-catching videos.
- Marketing agencies aiming to produce captivating promotional content.
- Dance instructors and choreographers seeking innovative ways to showcase choreography.4f:T6f1,The AI Cinematic B-Roll Generator is a powerful tool for creating high-quality cinematic videos with synchronized audio and multi-shot storytelling. Ideal for creators, marketers, and filmmakers, this workflow harnesses the capabilities of ByteDanceâ€™s advanced Seedance 2.0 API. By using AI, it automates the process of generating video content, dramatically reducing production time and costs while offering precision and stylistic control.

---

## 3. How does it work  
- The workflow utilizes the Seedance 2.0 model, which is a sophisticated AI designed for cinematic video generation with native audio synchronization, supporting multi-shot narratives.
- This workflow takes an input image to serve as a reference for character or style consistency, which is then processed to generate a cohesive video output.
- The workflow configuration includes a detailed prompt that specifies shot sequences, mood, and style, resulting in a beat-driven cinematic video that matches the user's creative vision.

---

## 4. How to customize it  
Users can tailor this workflow to better fit their creative needs by altering the input prompts. For instance, changing the "MOOD" attribute from "Playful â†’ stylish" to "Mysterious â†’ intense" can yield videos with a completely different atmosphere. Additionally, users may adjust the "STYLE" and "CORE IDEA" elements to explore various narrative and stylistic directions. Incorporating different reference images can vastly change the characters and thematic expression of the video output.

---

## 5. Who is it for  
- Filmmakers who require rapid prototyping of scenes.
- Marketing agencies aiming to create visually engaging advertisements.
- Content creators who produce videos for platforms like TikTok, YouTube, and Instagram.50:T75e,This workflow creates a seamless video transition loop simulating time travel, where a character navigates through different historical and futuristic settings in a continuous forward motion. It is ideal for creators, educators, or filmmakers who need engaging narrative content with dynamic transitions. Using advanced AI models like Seedance 2.0, this workflow offers dynamic scene management and synchronized audio, providing users with the ability to generate immersive video experiences quickly. The primary value comes from automation and creativity, allowing users to generate unique content without extensive manual editing.

---

## 3. How does it work  
- The Seedance 2.0 model is employed to create cinematic AI videos with native audio and multi-shot narratives. It processes prompts to generate complex sequences.
- An input node accepts image assets which are utilized as reference images in the video.
- The generated video is routed through an output node, producing a final video file that transitions through various time periods.

---

## 4. How to customize it  
Users can adapt the workflow to their needs by modifying the Seedance 2.0 parameters. For instance, changing the scene descriptions or historical periods in the prompt can create new narratives. Users can also adjust video duration and resolution to fit specific platform needs. For example, turning the setting futuristic with additional sci-fi elements can appeal to a tech-savvy audience, or tweaking the transitions to be smoother for a more relaxed viewing experience.

---

## 5. Who is it for  
- Filmmakers aiming to create dynamic video content with little manual intervention.
- Educators looking to produce engaging educational videos that present historical narratives.
- Content creators and marketers who need to produce eye-catching social media content that tells a story through video.51:T7b4,The AI-Powered Fashion B-roll Video Generator is designed to transform a still fashion image into a dynamically cinematic 15-second fashion b-roll video, complete with ambient audio and sweeping scene transitions. This workflow is particularly useful for fashion marketers, video creators, and production studios seeking quick content generation without the traditional overhead of video shoots. The workflow leverages the Seedance 2.0 model in conjunction with AI-enabled image-to-video transformation techniques, enabling faster content production and enriching the creative storytelling process through automation and cutting-edge AI technology.

## 3. How does it work  
- The workflow begins with an input image, which is utilized as a reference for styling, composition, and appearance consistency throughout the video.
- The core AI model used is Seedance 2.0, which specializes in converting static images into cinematic video sequences.
- The API node processes the input image following a detailed prompt that guides camera movement, subject poses, and audio effects over a 15-second timeline.
- The output is a video file rich in hyperreal pop aesthetics, complete with synchronized audio elements emulating a professional fashion shoot environment.

## 4. How to customize it  
Users can customize this workflow by adjusting the prompt to change the video's aesthetic and storytelling details. For example, you can modify the environment and mood settings to create a more vibrant or somber tone, or introduce new focal elements and subject poses to emphasize different fashion attributes. By changing the prompt or including different reference images, users can adapt the output to fit various creative visions or brand requirements, enhancing versatility across different campaigns or projects.

## 5. Who is it for  
- Fashion marketers and branding agencies
- Video creators and filmmakers
- Production studios and content creators in the fashion industry52:T8ba,The "AI Red Carpet Walk Video Creator" is an innovative workflow that utilizes AI to generate high-quality, cinematic videos of a red carpet event. This workflow is ideal for marketers, video creators, and event organizers looking to create glamorous videos with a touch of comedy and fashion-forward aesthetics. The workflow uses cutting-edge AI models, including ByteDance's Seedance 2.0, to produce videos that not only maintain the subject's identity from a reference image but also enhance their cinematic presence. It provides value by automating video production, reducing costs, and enabling users to generate compelling content quickly and efficiently.

## 3. How does it work  
- The workflow leverages ByteDance's Seedance 2.0 model, a multimodal video generation API, to create cinematic-quality AI videos from text, images, audio, and video inputs.
- It takes an input image, which serves as the reference point for the video, ensuring the main subject in the video matches the input image's attributes like face, body type, and style.
- The model transforms the image to a full 15-second video at 720p resolution with synchronized audio, mimicking the feel of a luxury evening red carpet event.
- Key features include maintaining subject consistency, smooth camera movements, and editorial fashion cinematography infused with a glam-comedy tone.

## 4. How to customize it  
Users can tailor this workflow by tweaking the input prompts and reference images to suit their specific requirements. For instance, they could change the fashion and expression described in the prompt to reflect a different celebrity style or mood. Alternatively, users might choose different reference images to produce videos for various events, such as a high-profile gala or a product launch. Additionally, users can adjust video parameters, such as duration and aspect ratio, to align with different platforms and audience preferences.

## 5. Who is it for  
- Social Media Marketers looking to enhance brand presence with engaging video content.
- Video Creators and Filmmakers aspiring to produce high-quality footage with minimal resources.
- Event Planners aiming to create visually striking promotional materials for exclusive events.53:T70b,This workflow automatically generates a 15-second looped video depicting a daily routine with stunning ultra-realistic visuals. Utilizing advanced AI models, it captures and transforms input images into a dynamic video narrative, making it perfect for creators, marketers, and content producers. The workflow leverages ByteDance's Seedance 2.0 and other video generation technologies to ensure precise continuity in identity and environment across scenes. It streamlines content creation for faster production and provides an automated solution for eye-catching video material.

## 3. How does it work  
- The workflow leverages the Seedance 2.0 model, which is a powerful AI video generation tool that can natively produce audio and multi-shot narratives.
- The video creation process begins with the input of a reference image, ensuring that the subject's identity remains consistent throughout the video.
- Using a detailed prompt, the AI ensures continuity in identity, wardrobe, and environment, adhering to predefined scene transitions and narrative style.
- This model produces a seamless 15-second video at 720p resolution with a 16:9 aspect ratio, integrating beat-synced fast-paced electro-pop music to match the visual tempo.

## 4. How to customize it  
Users can customize the workflow by altering the input image and modifying the detailed prompt to better fit their specific needs. For example, by changing the wardrobe details and environments, the user can depict a morning coffee shop visit instead of an office setting. Additionally, modifying the described style and mood parameters in the prompt will allow users to shift from a fast-paced urban feel to a relaxed, countryside morning routine.

## 5. Who is it for  
- Content Creators
- Digital Marketers
- Social Media Influencers54:T789,This workflow leverages advanced AI technology to create cinematic fashion brolls, transforming a reference image into a high-fashion video clip. It is highly beneficial for marketers, content creators, and fashion designers who seek to generate visually captivating promotional videos quickly and efficiently. The key technologies utilized are Seedance 2.0 for cinematic realism and dynamic shot sequencing, along with tools for precise styling and visual consistency. This workflow offers significant advantages in terms of speed, personalization, and high-quality content creation.

## 3. How does it work  
- **Input Image to Video Conversion**: The Seedance 2.0 model is used to transform a static reference image into a dynamic, 15-second cinematic video.
- **Ultra-realistic Styling**: The model ensures that the characterâ€™s identity, costume design, and makeup exactly match the reference image, using its image-to-video capabilities.
- **Cinematic Shots and Transition**: The workflow includes advanced shot types like macro push-ins, lateral tracking shots, and 360Â° orbits; all seamlessly integrated to maintain visual and style consistency.

## 4. How to customize it  
Users can customize this workflow by altering the input images, as these define the style, identity, and other specifics of the fashion model in the video. Additionally, users can modify the prompt to adjust shot types, styling details, and narrative flow to suit different artistic visions or promotional themes. For example, adapting the prompt to feature urban landscape backgrounds or altering the expression and movement style for different brand aesthetics can yield varied outcomes.

## 5. Who is it for  
- Fashion marketers looking to produce high-impact commercials.
- Content creators aiming to enhance their video storytelling capabilities.
- Fashion designers who need to present their collections in an engaging, cinematic format.55:T74c,The workflow creates dynamic and cinematic b-roll videos for product advertisements by processing input images to generate tailor-made video sequences. Utilizing sophisticated AI models like Seedance 2.0 for video generation and GPT 5.4 for prompt adaptation, the workflow is ideal for marketers, content creators, and advertisers looking to showcase products with high-end visual appeal. It delivers value by automating video production, ensuring consistent quality, and speeding up the content creation process, which allows for more efficient marketing campaigns.

## 3. How does it work  
- The workflow utilizes Seedance 2.0 to generate videos from prompts, creating visually appealing sequences with optional synchronized audio.
- GPT 5.4 is employed to transform and adapt the video prompts, ensuring that they are specific to the input product image, aligning the cinematic details to the productâ€™s features.
- Input images serve as reference points for dynamically adapting video prompts to keep the content tightly aligned with the product being showcased.

## 4. How to customize it  
Users can modify the workflow by adjusting the prompts used for video generation. Customization can be done by changing the product descriptions and cinematic styles in the prompt, allowing for videos that align more closely with different product categories or marketing themes. For instance, a user can adapt the workflow for a tech gadget by emphasizing sleek designs and light interactions, or shift to a textile product by highlighting texture and movement.

## 5. Who is it for  
- Marketing agencies seeking high-quality product b-roll videos for advertisements.
- Content creators and influencers who need visually striking videos for product showcasing on social media platforms.
- Video production teams focusing on creating luxury and fashion-related content.56:T5ea,This AI Dance Reel Video Creator workflow allows users to generate cinematic dance videos with synchronized audio and multi-shot narratives. It's particularly useful for content creators, filmmakers, and marketers who wish to produce engaging visual content efficiently. The workflow utilizes advanced AI models, including Seedance 2.0 for image-to-video conversion and GPT 5.4 for prompt optimization, offering faster and more cost-effective video creation by automating the storytelling process.

---

## 3. How does it work  
- The **Seedance 2.0** model is employed to create dynamic, multi-shot videos with user-defined prompts. It seamlessly converts images and prompt narratives into video outputs.
- **GPT 5.4** serves as a prompt editor, refining the input prompts to ensure the video's environment, lighting, and setting remain relevant and cohesive throughout.
- Users can input images and prompts that detail scene specifics, which are then processed by the AI models to produce a polished dance reel.

---

## 4. How to customize it  
Users can adapt the workflow by altering prompts and using different initial images to tailor the video to various themes or storylines. For instance, changing the setting from 'beach side' to 'urban rooftop' with alternate lighting styles can dramatically shift the video's mood. Users might also adjust the video duration or aspect ratio to suit specific platforms or audiences.

---

## 5. Who is it for  
- Content Creators
- Filmmakers
- Marketing Agencies

---57:T72c,This workflow, powered by advanced AI models, creates a cinematic video from a single image, capturing the essence of fashion with dynamic camera movements and depth effects. It is particularly useful for creators, marketers, and fashion designers who want to showcase their work in a beautiful and engaging manner. Utilizing technologies like Kling 3.0 Pro Image-to-Video, this workflow transforms a still image into an 8-second animation with smooth parallax, realistic motion, and cinematic effects, enabling faster production of high-quality fashion visuals with reduced costs.

## 3. How does it work  
- **Kling 3.0 Pro Image-to-Video**: This model generates high-quality animated videos from images. It uses dynamic camera motions like push-in, pan, and orbit to produce realistic and engaging visual effects.
- **Input and Output Nodes**: The workflow begins with an input node, where users can provide an image. It flows through to the Kling model API, which processes the image and outputs a video at the designated output node.

## 4. How to customize it  
Users can adapt this workflow to specific needs by modifying the prompt parameters for the Kling model. For instance, to highlight the intricate details of a dress, you might use a prompt that specifies a close-up orbital shot and soft lighting. Conversely, for a broad scene showcasing multiple garments, you could adjust the prompt to include wide-angle shots and extended video duration. Changing the aspect ratio for social media platforms or opting to include synchronized audio are additional ways to customize the output.

## 5. Who is it for  
- Fashion Designers aiming to present collections in a dynamic format.
- Marketers looking to create compelling social media content.
- Digital Content Creators seeking to enhance their video production capabilities.1f:["$","div",null,{"className":"flex flex-col items-center","children":[["$","script",null,{"type":"application/ld+json","dangerouslySetInnerHTML":{"__html":"$33"}}],["$","div",null,{"className":"w-full mx-auto","children":[["$","section",null,{"className":"max-w-[1536px] mx-auto","children":["$","$L34",null,{"apiData":{"_id":{"$oid":"6a9fa7a00bcd3da2348df831"},"graph_id":"6a9fa7a00bcd3da2348df830","meta_description":"Create Captivating 15-Second Fashion B-Roll Videos with Seedance 2.5 and Cinematic Posing.","show_pixelflow":true,"is_private":false,"version":1,"is_template":true,"template_id":"6a9fa7a00bcd3da2348df830","slug":"6a9fa7a00bcd3da2348df830-v1","graph_data":{"nodes":[{"id":"node-1788766051328","type":"apiNode","data":{"type":"api","id":"node-1788766051328","meta":{"minimised":false},"parameters":{"prompt":"$35","system_prompt":"$36","image":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"slug":"gpt-6-astra"},"template":"gpt-6-astra","apiKeyName":"gpt_6_astra_e2b1x","modelInfo":{"average_cost":0.03669338888888889,"coming_soon":false,"created_at":"2026-09-05 03:53:39.424474+00:00","default_image_output":"https://segmind-sd-models.s3.amazonaws.com/display_images/gpt-4.png","default_response":null,"description":"Complex reasoning, code generation, and image understanding; 1M-token context.","inference_provider":"OPENAI","is_depreciated":false,"is_new":true,"is_visible":true,"latency":12.22961,"parameters":{"image":{"affectsPricing":false,"component":"image","default":null,"description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","displayOrder":1,"hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"label":"Image","type":"image"},"prompt":{"affectsPricing":false,"component":"textarea","default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","displayOrder":0,"displayValue":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":true,"label":"prompt","type":"str"}},"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","examples":[{"prompt":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input."}],"properties":{"image":{"description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","format":"uri","title":"Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"order":1}},"prompt":{"default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","minLength":1,"title":"Prompt","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":0,"placeholder":"Ask a question or give an instruction"}}},"required":["prompt"],"type":"object"},"partner":"OPENAI","popularity_score":0,"search_override_score":0,"slug":"gpt-6-astra","thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/gpt-6-astra-1903.webp","title":"GPT 6 Astra","type":"textToText","objectID":"gpt-6-astra","_highlightResult":{"description":{"value":"Complex reasoning, code generation, and image understanding; 1M-token context.","matchLevel":"none","matchedWords":[]},"parameters":{"prompt":{"component":{"value":"textarea","matchLevel":"none","matchedWords":[]},"default":{"value":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","matchLevel":"none","matchedWords":[]},"description":{"value":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","matchLevel":"none","matchedWords":[]},"displayOrder":{"value":"0","matchLevel":"none","matchedWords":[]},"displayValue":{"value":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","matchLevel":"none","matchedWords":[]},"label":{"value":"prompt","matchLevel":"none","matchedWords":[]},"type":{"value":"str","matchLevel":"none","matchedWords":[]}}},"title":{"value":"GPT 6 Astra","matchLevel":"none","matchedWords":[]}}},"modelData":{"title":"GPT 6 Astra","type":"textToText","slug":"gpt-6-astra","description":"Complex reasoning, code generation, and image understanding; 1M-token context.","is_visible":true,"default_image_output":"https://segmind-sd-models.s3.amazonaws.com/display_images/gpt-4.png","default_response":null,"thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/gpt-6-astra-1903.webp","is_depreciated":false,"is_new":true,"coming_soon":false,"segmind":true,"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","type":"object","properties":{"prompt":{"type":"string","title":"Prompt","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","minLength":1,"x-ui-metadata":{"order":0,"advanced":false,"hidden":false,"placeholder":"Ask a question or give an instruction"}},"image":{"type":"string","title":"Image","description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","format":"uri","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":1,"advanced":false,"hidden":false}}},"required":["prompt"],"examples":[{"prompt":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input."}]},"partner":"OPENAI","inference_provider":"OPENAI","latency":12.22961,"average_cost":0.036693388888888887,"search_override_score":0,"created_at":"2026-09-05 03:53:39.424474+00:00"},"parameterTemplates":{"image":"{{ input_image }}"},"fanOutInstructions":"Output usage â€” your response is used directly as the following downstream model input(s):\n  â€¢ Seedance 2.5 Â· prompt  (string) â€” Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citatiâ€¦\nReturn only the value for the connected input â€” no markdown, code fences, or commentary.","keyname":"gpt_6_astra_e2b1x","useInApi":true,"value":""},"measured":{"width":380,"height":1003}},{"id":"node-1788766056595","type":"apiNode","data":{"type":"api","id":"node-1788766056595","meta":{"minimised":false},"parameters":{"prompt":"{{ gpt_6_astra_e2b1x }}","duration":15,"reference_images":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"seed":58385081,"slug":"seedance-2.5"},"template":"seedance-2.5","apiKeyName":"seedance_2_5_qgfrs","modelInfo":{"average_cost":3.033373417751536,"coming_soon":false,"created_at":"2026-08-07 06:05:13.024224+00:00","default_image_output":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-example-output.mp4","default_response":null,"description":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","inference_provider":"BYTEDANCE","is_depreciated":false,"is_new":true,"is_visible":true,"latency":311.42742,"parameters":{"aspect_ratio":{"affectsPricing":false,"component":"dropdown","default":"adaptive","description":"Output ratio: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive. Set adaptive when using first_frame_url.","displayOrder":8,"displayValue":"adaptive","hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Aspect Ratio","type":"str","values":["16:9","9:16","1:1","4:3","3:4","21:9","adaptive"]},"auto_resize":{"affectsPricing":false,"component":"text_field","default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","displayOrder":15,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Auto Resize Inputs","type":"bool"},"bitrate_mode":{"affectsPricing":false,"component":"dropdown","default":"standard","description":"Encoding bitrate: standard or high. Use standard for smaller files, high for maximum fidelity.","displayOrder":13,"displayValue":"standard","hidden":false,"isAdvanced":true,"isEnum":true,"isRequired":false,"label":"Bitrate Mode","type":"str","values":["standard","high"]},"duration":{"affectsPricing":false,"component":"dropdown","default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose (required for video edits: output follows the input video).","displayOrder":6,"displayValue":5,"hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Duration","range":[4,30],"type":"int","values":[-1,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30]},"first_frame_url":{"affectsPricing":false,"component":"textbox","default":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","description":"Starting image for image-to-video; no real human faces. Leave empty for text-to-video; cannot combine with reference_images.","displayOrder":1,"displayValue":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"label":"First Frame Image","type":"str"},"generate_audio":{"affectsPricing":false,"component":"text_field","default":true,"description":"Co-generates synchronized dialogue, effects, and music. Enable for speech or ambient scenes.","displayOrder":9,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Generate Audio","type":"bool"},"last_frame_url":{"affectsPricing":false,"component":"textbox","default":"","description":"Ending frame guiding a transition; requires first_frame_url. Use for controlled start-to-end morphs.","displayOrder":4,"displayValue":"","format":"uri","hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Last Frame Image","type":"str"},"output_format":{"affectsPricing":false,"component":"dropdown","default":"mp4","description":"mp4 = best compatibility. mov = higher color fidelity; recommended for video edits.","displayOrder":14,"displayValue":"mp4","hidden":false,"isAdvanced":true,"isEnum":true,"isRequired":false,"label":"Output Format","type":"str","values":["mp4","mov"]},"prompt":{"affectsPricing":false,"component":"textarea","default":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","description":"Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citations. Detail subject, action, camera, and audio.","displayOrder":0,"displayValue":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":true,"label":"Prompt","type":"str"},"reference_audios":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 10 MP3s (2-30s) to drive audio, cited as @Audio 1. Use for voice, music, or beat-matched pacing.","displayOrder":5,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Audios","type":"array"},"reference_images":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 30 images for character or style consistency, cited as @Image 1. Not usable with first_frame_url.","displayOrder":2,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Images","type":"array"},"reference_videos":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 10 clips (2-30s) for motion or style transfer, cited as @Video 1. Billed at video-input rate.","displayOrder":3,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Videos","type":"array"},"resolution":{"affectsPricing":false,"component":"dropdown","default":"720p","description":"Output resolution, 480p, 720p or 1080p. Use 480p for drafts, 720p for final renders, 1080p for highest fidelity.","displayOrder":7,"displayValue":"720p","hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Resolution","type":"str","values":["480p","720p","1080p"]},"return_last_frame":{"affectsPricing":false,"component":"text_field","default":false,"description":"Returns the final frame as an image URL. Enable to chain clips into longer sequences.","displayOrder":11,"displayValue":false,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Return Last Frame","type":"bool"},"seed":{"affectsPricing":false,"component":"textbox","default":42,"description":"Reproducibility seed (-1 to 2147483647). Use -1 for random, fix a value to iterate a scene.","displayOrder":10,"displayValue":42,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Seed","range":[-1,2147483647],"type":"int"},"skip_moderation":{"affectsPricing":false,"component":"toggle","default":true,"description":"Bypasses the content pre-filter for false positives. Leave off unless artistic content is wrongly blocked.","displayOrder":12,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Skip Moderation","type":"bool"}},"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","examples":[{"aspect_ratio":"adaptive","bitrate_mode":"standard","duration":5,"first_frame_url":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","generate_audio":true,"last_frame_url":"","prompt":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","reference_audios":[],"reference_images":[],"reference_videos":[],"resolution":"720p","return_last_frame":false,"seed":42,"skip_moderation":false}],"properties":{"aspect_ratio":{"default":"16:9","description":"Output aspect ratio. Use 16:9 for landscape, 9:16 for vertical social video, 21:9 for ultrawide cinematic.","oneOf":[{"const":"16:9","title":"16:9"},{"const":"9:16","title":"9:16"},{"const":"1:1","title":"1:1"},{"const":"4:3","title":"4:3"},{"const":"3:4","title":"3:4"},{"const":"21:9","title":"21:9"},{"const":"adaptive","title":"adaptive"}],"title":"Aspect Ratio","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":8}},"auto_resize":{"default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","title":"Auto Resize Inputs","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":15}},"bitrate_mode":{"default":"standard","description":"Output video encoding bitrate. 'standard' gives smaller files and faster downloads; 'high' produces ~5-6x higher bitrate for better visual fidelity and fewer compression artifacts (larger files). Does not affect price.","oneOf":[{"const":"standard","title":"Standard"},{"const":"high","title":"High"}],"title":"Bitrate Mode","type":"string","x-ui-metadata":{"advanced":true,"hidden":false,"order":13}},"duration":{"default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose. Video-editing prompts require -1 (output follows the input video, which must be 4-30s). Cost scales with output duration (token-based).","oneOf":[{"const":-1,"title":"Auto (use for video edits)"},{"const":4,"title":"4s"},{"const":5,"title":"5s"},{"const":6,"title":"6s"},{"const":7,"title":"7s"},{"const":8,"title":"8s"},{"const":9,"title":"9s"},{"const":10,"title":"10s"},{"const":11,"title":"11s"},{"const":12,"title":"12s"},{"const":13,"title":"13s"},{"const":14,"title":"14s"},{"const":15,"title":"15s"},{"const":16,"title":"16s"},{"const":17,"title":"17s"},{"const":18,"title":"18s"},{"const":19,"title":"19s"},{"const":20,"title":"20s"},{"const":21,"title":"21s"},{"const":22,"title":"22s"},{"const":23,"title":"23s"},{"const":24,"title":"24s"},{"const":25,"title":"25s"},{"const":26,"title":"26s"},{"const":27,"title":"27s"},{"const":28,"title":"28s"},{"const":29,"title":"29s"},{"const":30,"title":"30s"}],"title":"Duration","type":"integer","x-ui-metadata":{"advanced":false,"hidden":false,"order":6}},"first_frame_url":{"description":"Starting frame image URL for image-to-video. Animates outward from this reference. Note: images with real human faces are blocked by ByteDance content policy. Cannot be used together with reference_images.","format":"uri","title":"First Frame Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"First & last frame","order":1}},"generate_audio":{"default":true,"description":"Co-generate synchronized audio (dialogue, SFX, ambient, music) in the same pass. Best for scenes with speech, nature sounds, or music.","title":"Generate Audio","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":9}},"last_frame_url":{"description":"Ending frame; requires first_frame_url. Guides transitions between two frames.","format":"uri","title":"Last Frame Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"First & last frame","order":2}},"output_format":{"default":"mp4","description":"mp4 = best compatibility. mov (H.264 + yuv444p + PCM) = higher color fidelity; recommended for video editing/extension round-trips.","enum":["mp4","mov"],"title":"Output Format","type":"string","x-ui-metadata":{"advanced":true,"hidden":false,"order":14}},"prompt":{"description":"Text describing the video. Use Shot 1:, Shot 2: for multi-shot sequences. Reference uploaded assets as '@Image 1', '@Video 1', '@Audio 1' in your prompt.","title":"Prompt","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":0}},"reference_audios":{"description":"Up to 10 reference audio files (MP3, 2-30 seconds each, 30 seconds combined, max 15MB each). Cite as '@Audio 1' in your prompt. Audio-only reference is supported on seedance-2.5.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["audio/mpeg"]}},"maxItems":10,"title":"Reference Audios","type":"array","x-ui-metadata":{"advanced":true,"hidden":false,"mode-group":"Reference media","order":5}},"reference_images":{"description":"Up to 30 reference images (max 30MB each) for character/style consistency. Cite as '@Image 1', '@Image 2' in your prompt. Cannot be used together with first_frame_url.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["image/*"]}},"maxItems":30,"title":"Reference Images","type":"array","x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"Reference media","order":3}},"reference_videos":{"description":"Up to 10 reference videos (2-30 seconds each, 30 seconds combined, mp4/mov, max 200MB each) for motion/style transfer. Cite as '@Video 1' in your prompt. Requests with reference videos are billed at the video-input token rate.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["video/mp4","video/quicktime"]}},"maxItems":10,"title":"Reference Videos","type":"array","x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"Reference media","order":4}},"resolution":{"default":"720p","description":"Output resolution. seedance-2.5 supports 480p, 720p and 1080p (24fps). Use 480p for drafts and fast iteration, 720p for final renders, 1080p for the highest fidelity (10-bit colour). Higher resolution costs more (token-based).","oneOf":[{"const":"480p","title":"480p"},{"const":"720p","title":"720p"},{"const":"1080p","title":"1080p"}],"title":"Resolution","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":7}},"return_last_frame":{"default":false,"description":"Returns the final video frame as a separate image, delivered in the response as `video.last_frame_url`. Chain shots by feeding that URL into the next generation's `first_frame_url`. Note this is an output: it is distinct from the `last_frame_url` input parameter, which is an image you supply to steer how the clip ends.","title":"Return Last Frame","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":11}},"seed":{"default":42,"description":"Reproducibility seed. Use -1 for random output, set a fixed value to iterate on the same scene.","maximum":2147483647,"minimum":-1,"title":"Seed","type":"integer","x-ui-metadata":{"advanced":true,"hidden":false,"order":10}},"skip_moderation":{"default":true,"description":"Bypass BytePlus content moderation pre-filter. Useful when generating artistic content or faces that may trigger false positives. Baseline BytePlus safety policies still apply.","title":"Skip Moderation","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":12}}},"required":["prompt"],"type":"object"},"partner":"BYTEDANCE","popularity_score":0,"search_override_score":0,"slug":"seedance-2.5","thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-5954.webp","title":"Seedance 2.5","type":"imageToVideo","objectID":"seedance-2.5","_highlightResult":{"description":{"value":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","matchLevel":"none","matchedWords":[]},"parameters":{"prompt":{"component":{"value":"textarea","matchLevel":"none","matchedWords":[]},"default":{"value":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","matchLevel":"none","matchedWords":[]},"description":{"value":"Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citations. Detail subject, action, camera, and audio.","matchLevel":"none","matchedWords":[]},"displayOrder":{"value":"0","matchLevel":"none","matchedWords":[]},"displayValue":{"value":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","matchLevel":"none","matchedWords":[]},"label":{"value":"Prompt","matchLevel":"none","matchedWords":[]},"type":{"value":"str","matchLevel":"none","matchedWords":[]}}},"title":{"value":"<em>Seedance</em> 2.5","matchLevel":"full","fullyHighlighted":false,"matchedWords":["seedance"]}}},"modelData":{"title":"Seedance 2.5","type":"imageToVideo","slug":"seedance-2.5","description":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","is_visible":true,"default_image_output":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-example-output.mp4","default_response":null,"thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-5954.webp","is_depreciated":false,"is_new":true,"coming_soon":false,"segmind":true,"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","type":"object","properties":{"prompt":{"type":"string","title":"Prompt","description":"Text describing the video. Use Shot 1:, Shot 2: for multi-shot sequences. Reference uploaded assets as '@Image 1', '@Video 1', '@Audio 1' in your prompt.","x-ui-metadata":{"order":0,"advanced":false,"hidden":false}},"first_frame_url":{"type":"string","title":"First Frame Image","format":"uri","description":"Starting frame image URL for image-to-video. Animates outward from this reference. Note: images with real human faces are blocked by ByteDance content policy. Cannot be used together with reference_images.","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":1,"advanced":false,"hidden":false,"mode-group":"First & last frame"}},"last_frame_url":{"type":"string","title":"Last Frame Image","format":"uri","description":"Ending frame; requires first_frame_url. Guides transitions between two frames.","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":2,"advanced":false,"hidden":false,"mode-group":"First & last frame"}},"reference_images":{"type":"array","title":"Reference Images","description":"Up to 30 reference images (max 30MB each) for character/style consistency. Cite as '@Image 1', '@Image 2' in your prompt. Cannot be used together with first_frame_url.","maxItems":30,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["image/*"]}},"x-ui-metadata":{"order":3,"advanced":false,"hidden":false,"mode-group":"Reference media"}},"reference_videos":{"type":"array","title":"Reference Videos","description":"Up to 10 reference videos (2-30 seconds each, 30 seconds combined, mp4/mov, max 200MB each) for motion/style transfer. Cite as '@Video 1' in your prompt. Requests with reference videos are billed at the video-input token rate.","maxItems":10,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["video/mp4","video/quicktime"]}},"x-ui-metadata":{"order":4,"advanced":false,"hidden":false,"mode-group":"Reference media"}},"reference_audios":{"type":"array","title":"Reference Audios","description":"Up to 10 reference audio files (MP3, 2-30 seconds each, 30 seconds combined, max 15MB each). Cite as '@Audio 1' in your prompt. Audio-only reference is supported on seedance-2.5.","maxItems":10,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["audio/mpeg"]}},"x-ui-metadata":{"order":5,"advanced":true,"hidden":false,"mode-group":"Reference media"}},"duration":{"type":"integer","title":"Duration","default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose. Video-editing prompts require -1 (output follows the input video, which must be 4-30s). Cost scales with output duration (token-based).","x-ui-metadata":{"order":6,"advanced":false,"hidden":false},"oneOf":[{"const":-1,"title":"Auto (use for video edits)"},{"const":4,"title":"4s"},{"const":5,"title":"5s"},{"const":6,"title":"6s"},{"const":7,"title":"7s"},{"const":8,"title":"8s"},{"const":9,"title":"9s"},{"const":10,"title":"10s"},{"const":11,"title":"11s"},{"const":12,"title":"12s"},{"const":13,"title":"13s"},{"const":14,"title":"14s"},{"const":15,"title":"15s"},{"const":16,"title":"16s"},{"const":17,"title":"17s"},{"const":18,"title":"18s"},{"const":19,"title":"19s"},{"const":20,"title":"20s"},{"const":21,"title":"21s"},{"const":22,"title":"22s"},{"const":23,"title":"23s"},{"const":24,"title":"24s"},{"const":25,"title":"25s"},{"const":26,"title":"26s"},{"const":27,"title":"27s"},{"const":28,"title":"28s"},{"const":29,"title":"29s"},{"const":30,"title":"30s"}]},"resolution":{"type":"string","title":"Resolution","default":"720p","description":"Output resolution. seedance-2.5 supports 480p, 720p and 1080p (24fps). Use 480p for drafts and fast iteration, 720p for final renders, 1080p for the highest fidelity (10-bit colour). Higher resolution costs more (token-based).","oneOf":[{"const":"480p","title":"480p"},{"const":"720p","title":"720p"},{"const":"1080p","title":"1080p"}],"x-ui-metadata":{"order":7,"advanced":false,"hidden":false}},"aspect_ratio":{"type":"string","title":"Aspect Ratio","default":"16:9","description":"Output aspect ratio. Use 16:9 for landscape, 9:16 for vertical social video, 21:9 for ultrawide cinematic.","oneOf":[{"const":"16:9","title":"16:9"},{"const":"9:16","title":"9:16"},{"const":"1:1","title":"1:1"},{"const":"4:3","title":"4:3"},{"const":"3:4","title":"3:4"},{"const":"21:9","title":"21:9"},{"const":"adaptive","title":"adaptive"}],"x-ui-metadata":{"order":8,"advanced":false,"hidden":false}},"generate_audio":{"type":"boolean","title":"Generate Audio","default":true,"description":"Co-generate synchronized audio (dialogue, SFX, ambient, music) in the same pass. Best for scenes with speech, nature sounds, or music.","x-ui-metadata":{"order":9,"advanced":true,"hidden":false}},"seed":{"type":"integer","title":"Seed","default":42,"minimum":-1,"maximum":2147483647,"description":"Reproducibility seed. Use -1 for random output, set a fixed value to iterate on the same scene.","x-ui-metadata":{"order":10,"advanced":true,"hidden":false}},"return_last_frame":{"type":"boolean","title":"Return Last Frame","default":false,"description":"Returns the final video frame as a separate image, delivered in the response as `video.last_frame_url`. Chain shots by feeding that URL into the next generation's `first_frame_url`. Note this is an output: it is distinct from the `last_frame_url` input parameter, which is an image you supply to steer how the clip ends.","x-ui-metadata":{"order":11,"advanced":true,"hidden":false}},"skip_moderation":{"type":"boolean","title":"Skip Moderation","default":true,"description":"Bypass BytePlus content moderation pre-filter. Useful when generating artistic content or faces that may trigger false positives. Baseline BytePlus safety policies still apply.","x-ui-metadata":{"order":12,"advanced":true,"hidden":false}},"bitrate_mode":{"type":"string","title":"Bitrate Mode","default":"standard","description":"Output video encoding bitrate. 'standard' gives smaller files and faster downloads; 'high' produces ~5-6x higher bitrate for better visual fidelity and fewer compression artifacts (larger files). Does not affect price.","oneOf":[{"const":"standard","title":"Standard"},{"const":"high","title":"High"}],"x-ui-metadata":{"order":13,"advanced":true,"hidden":false}},"output_format":{"type":"string","title":"Output Format","default":"mp4","enum":["mp4","mov"],"description":"mp4 = best compatibility. mov (H.264 + yuv444p + PCM) = higher color fidelity; recommended for video editing/extension round-trips.","x-ui-metadata":{"order":14,"advanced":true,"hidden":false}},"auto_resize":{"type":"boolean","title":"Auto Resize Inputs","default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","x-ui-metadata":{"order":15,"advanced":true,"hidden":false}}},"required":["prompt"],"examples":[{"prompt":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","first_frame_url":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","last_frame_url":"","reference_images":[],"reference_videos":[],"reference_audios":[],"duration":5,"resolution":"720p","aspect_ratio":"adaptive","generate_audio":true,"seed":42,"return_last_frame":false,"skip_moderation":false,"bitrate_mode":"standard"}]},"partner":"BYTEDANCE","inference_provider":"BYTEDANCE","latency":311.42742,"average_cost":3.033373417751536,"search_override_score":0,"created_at":"2026-08-07 06:05:13.024224+00:00"},"parameterTemplates":{"prompt":"{{ gpt_6_astra_e2b1x }}","reference_images":"{{ input_image }}"},"keyname":"seedance_2_5_qgfrs","useInApi":true,"value":""},"measured":{"width":380,"height":1420}},{"id":"node-1788766571473","type":"inputNode","data":{"type":"input","id":"node-1788766571473","template":"file","mode":"any","value":"","apiKeyName":"input_image","keyname":"input_image","useInApi":true,"status":"success","outputs":{"value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"data":{"value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"]},"executionTime":0.09999996423721313},"errors":""},"measured":{"width":320,"height":582}},{"id":"node-1788768863220","type":"outputNode","data":{"type":"output","id":"node-1788768863220","value":"","template":"auto","apiKeyName":"Output_Video","keyname":"Output_Video","useInApi":true,"meta":{"sizeMode":"screen","nativeSize":false,"width":1106,"prevWidth":380}},"measured":{"width":1106,"height":700}}],"edges":[{"source":"node-1788766571473","sourceHandle":"string","target":"node-1788766051328","targetHandle":"param-image","id":"xy-edge__node-1788766571473string-node-1788766051328param-image"},{"source":"node-1788766051328","sourceHandle":"text","target":"node-1788766056595","targetHandle":"param-prompt","id":"xy-edge__node-1788766051328text-node-1788766056595param-prompt","selected":false},{"source":"node-1788766571473","sourceHandle":"string","target":"node-1788766056595","targetHandle":"param-reference_images","id":"xy-edge__node-1788766571473string-node-1788766056595param-reference_images"},{"source":"node-1788766056595","sourceHandle":"video","target":"node-1788768863220","targetHandle":"output","id":"xy-edge__node-1788766056595video-node-1788768863220output"}]},"api_graph_data":{"api_input_nodes":[{"id":"node-1788766571473","type":"inputNode","data":{"type":"input","id":"node-1788766571473","template":"file","mode":"any","value":"","apiKeyName":"input_image","keyname":"input_image","useInApi":true,"status":"success","outputs":{"value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"data":{"value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"]},"executionTime":0.09999996423721313},"errors":""},"measured":{"width":320,"height":582}}],"api_output_nodes":[{"id":"node-1788768863220","type":"outputNode","data":{"type":"output","id":"node-1788768863220","value":"","template":"auto","apiKeyName":"Output_Video","keyname":"Output_Video","useInApi":true,"meta":{"sizeMode":"screen","nativeSize":false,"width":1106,"prevWidth":380}},"measured":{"width":1106,"height":700}}]}},"template":{"_id":{"$oid":"6a9fa7a00bcd3da2348df830"},"author":"photonotixmedia@gmail.com","template_name":"Cinematic Fashion B-Roll Videos with Seedance 2.5","template_data":{"nodes":[{"id":"node-1788766051328","type":"apiNode","position":{"x":740,"y":260},"data":{"type":"api","id":"node-1788766051328","meta":{"minimised":false},"parameters":{"prompt":"$37","system_prompt":"$38","image":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"]},"template":"gpt-6-astra","apiKeyName":"gpt_6_astra_e2b1x","modelInfo":{"average_cost":0.03669338888888889,"coming_soon":false,"created_at":"2026-09-05 03:53:39.424474+00:00","default_image_output":"https://segmind-sd-models.s3.amazonaws.com/display_images/gpt-4.png","default_response":null,"description":"Complex reasoning, code generation, and image understanding; 1M-token context.","inference_provider":"OPENAI","is_depreciated":false,"is_new":true,"is_visible":true,"latency":12.22961,"parameters":{"image":{"affectsPricing":false,"component":"image","default":null,"description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","displayOrder":1,"hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"label":"Image","type":"image"},"prompt":{"affectsPricing":false,"component":"textarea","default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","displayOrder":0,"displayValue":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":true,"label":"prompt","type":"str"}},"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","examples":[{"prompt":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input."}],"properties":{"image":{"description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","format":"uri","title":"Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"order":1}},"prompt":{"default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","minLength":1,"title":"Prompt","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":0,"placeholder":"Ask a question or give an instruction"}}},"required":["prompt"],"type":"object"},"partner":"OPENAI","popularity_score":0,"search_override_score":0,"slug":"gpt-6-astra","thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/gpt-6-astra-1903.webp","title":"GPT 6 Astra","type":"textToText","objectID":"gpt-6-astra","_highlightResult":{"description":{"value":"Complex reasoning, code generation, and image understanding; 1M-token context.","matchLevel":"none","matchedWords":[]},"parameters":{"prompt":{"component":{"value":"textarea","matchLevel":"none","matchedWords":[]},"default":{"value":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","matchLevel":"none","matchedWords":[]},"description":{"value":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","matchLevel":"none","matchedWords":[]},"displayOrder":{"value":"0","matchLevel":"none","matchedWords":[]},"displayValue":{"value":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","matchLevel":"none","matchedWords":[]},"label":{"value":"prompt","matchLevel":"none","matchedWords":[]},"type":{"value":"str","matchLevel":"none","matchedWords":[]}}},"title":{"value":"GPT 6 Astra","matchLevel":"none","matchedWords":[]}}},"modelData":{"title":"GPT 6 Astra","type":"textToText","slug":"gpt-6-astra","description":"Complex reasoning, code generation, and image understanding; 1M-token context.","is_visible":true,"default_image_output":"https://segmind-sd-models.s3.amazonaws.com/display_images/gpt-4.png","default_response":null,"thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/gpt-6-astra-1903.webp","is_depreciated":false,"is_new":true,"coming_soon":false,"segmind":true,"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","type":"object","properties":{"prompt":{"type":"string","title":"Prompt","description":"Text instruction or question; be specific and state the output format you want. Use detailed, multi-step prompts for hard reasoning or code.","default":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input.","minLength":1,"x-ui-metadata":{"order":0,"advanced":false,"hidden":false,"placeholder":"Ask a question or give an instruction"}},"image":{"type":"string","title":"Image","description":"Optional image URL (reachable http/https) for vision tasks; pair with a specific question. Omit for text-only prompts.","format":"uri","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":1,"advanced":false,"hidden":false}}},"required":["prompt"],"examples":[{"prompt":"Implement a small arithmetic-expression evaluator in Python that supports +, -, *, /, unary minus, and parentheses with correct operator precedence. Use a tokenizer plus a recursive-descent parser (no eval, no regex-only shortcuts), and raise a clear error on malformed input like '3 + * 2'. Then trace how it parses -3 + 4 * 2 - (10 / 4) step by step, state the final value, and include five unittest cases covering precedence, unary minus, division, parentheses, and a malformed input."}]},"partner":"OPENAI","inference_provider":"OPENAI","latency":12.22961,"average_cost":0.036693388888888887,"search_override_score":0,"created_at":"2026-09-05 03:53:39.424474+00:00"},"parameterTemplates":{"image":"{{ input_image }}"},"fanOutInstructions":"Output usage â€” your response is used directly as the following downstream model input(s):\n  â€¢ Seedance 2.5 Â· prompt  (string) â€” Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citatiâ€¦\nReturn only the value for the connected input â€” no markdown, code fences, or commentary.","outputs":{"text":"$39","data":{"text":"$3a","metrics":{"cost":0.15141,"inference_time":72.77442,"queue_time":0.038667,"remaining_credits":101.6587128706,"total_time":72.813087},"reasoning":null,"timings":{"inference":72.77442},"output":"$3b","choices":[{"message":{"content":"$3c"}}]},"executionTime":74061.29999995232},"errors":"","status":"success"},"measured":{"width":380,"height":1003}},{"id":"node-1788766056595","type":"apiNode","position":{"x":1190,"y":260},"data":{"type":"api","id":"node-1788766056595","meta":{"minimised":false},"parameters":{"prompt":"{{ gpt_6_astra_e2b1x }}","duration":15,"reference_images":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"seed":58385081},"template":"seedance-2.5","apiKeyName":"seedance_2_5_qgfrs","modelInfo":{"average_cost":3.033373417751536,"coming_soon":false,"created_at":"2026-08-07 06:05:13.024224+00:00","default_image_output":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-example-output.mp4","default_response":null,"description":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","inference_provider":"BYTEDANCE","is_depreciated":false,"is_new":true,"is_visible":true,"latency":311.42742,"parameters":{"aspect_ratio":{"affectsPricing":false,"component":"dropdown","default":"adaptive","description":"Output ratio: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive. Set adaptive when using first_frame_url.","displayOrder":8,"displayValue":"adaptive","hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Aspect Ratio","type":"str","values":["16:9","9:16","1:1","4:3","3:4","21:9","adaptive"]},"auto_resize":{"affectsPricing":false,"component":"text_field","default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","displayOrder":15,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Auto Resize Inputs","type":"bool"},"bitrate_mode":{"affectsPricing":false,"component":"dropdown","default":"standard","description":"Encoding bitrate: standard or high. Use standard for smaller files, high for maximum fidelity.","displayOrder":13,"displayValue":"standard","hidden":false,"isAdvanced":true,"isEnum":true,"isRequired":false,"label":"Bitrate Mode","type":"str","values":["standard","high"]},"duration":{"affectsPricing":false,"component":"dropdown","default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose (required for video edits: output follows the input video).","displayOrder":6,"displayValue":5,"hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Duration","range":[4,30],"type":"int","values":[-1,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30]},"first_frame_url":{"affectsPricing":false,"component":"textbox","default":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","description":"Starting image for image-to-video; no real human faces. Leave empty for text-to-video; cannot combine with reference_images.","displayOrder":1,"displayValue":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"label":"First Frame Image","type":"str"},"generate_audio":{"affectsPricing":false,"component":"text_field","default":true,"description":"Co-generates synchronized dialogue, effects, and music. Enable for speech or ambient scenes.","displayOrder":9,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Generate Audio","type":"bool"},"last_frame_url":{"affectsPricing":false,"component":"textbox","default":"","description":"Ending frame guiding a transition; requires first_frame_url. Use for controlled start-to-end morphs.","displayOrder":4,"displayValue":"","format":"uri","hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Last Frame Image","type":"str"},"output_format":{"affectsPricing":false,"component":"dropdown","default":"mp4","description":"mp4 = best compatibility. mov = higher color fidelity; recommended for video edits.","displayOrder":14,"displayValue":"mp4","hidden":false,"isAdvanced":true,"isEnum":true,"isRequired":false,"label":"Output Format","type":"str","values":["mp4","mov"]},"prompt":{"affectsPricing":false,"component":"textarea","default":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","description":"Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citations. Detail subject, action, camera, and audio.","displayOrder":0,"displayValue":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":true,"label":"Prompt","type":"str"},"reference_audios":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 10 MP3s (2-30s) to drive audio, cited as @Audio 1. Use for voice, music, or beat-matched pacing.","displayOrder":5,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Audios","type":"array"},"reference_images":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 30 images for character or style consistency, cited as @Image 1. Not usable with first_frame_url.","displayOrder":2,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Images","type":"array"},"reference_videos":{"affectsPricing":false,"component":"textbox","default":[],"description":"Up to 10 clips (2-30s) for motion or style transfer, cited as @Video 1. Billed at video-input rate.","displayOrder":3,"displayValue":[],"format":"uri","hidden":false,"isAdvanced":false,"isEnum":false,"isRequired":false,"itemType":"str","label":"Reference Videos","type":"array"},"resolution":{"affectsPricing":false,"component":"dropdown","default":"720p","description":"Output resolution, 480p, 720p or 1080p. Use 480p for drafts, 720p for final renders, 1080p for highest fidelity.","displayOrder":7,"displayValue":"720p","hidden":false,"isAdvanced":false,"isEnum":true,"isRequired":false,"label":"Resolution","type":"str","values":["480p","720p","1080p"]},"return_last_frame":{"affectsPricing":false,"component":"text_field","default":false,"description":"Returns the final frame as an image URL. Enable to chain clips into longer sequences.","displayOrder":11,"displayValue":false,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Return Last Frame","type":"bool"},"seed":{"affectsPricing":false,"component":"textbox","default":42,"description":"Reproducibility seed (-1 to 2147483647). Use -1 for random, fix a value to iterate a scene.","displayOrder":10,"displayValue":42,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Seed","range":[-1,2147483647],"type":"int"},"skip_moderation":{"affectsPricing":false,"component":"toggle","default":true,"description":"Bypasses the content pre-filter for false positives. Leave off unless artistic content is wrongly blocked.","displayOrder":12,"displayValue":true,"hidden":false,"isAdvanced":true,"isEnum":false,"isRequired":false,"label":"Skip Moderation","type":"bool"}},"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","examples":[{"aspect_ratio":"adaptive","bitrate_mode":"standard","duration":5,"first_frame_url":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","generate_audio":true,"last_frame_url":"","prompt":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","reference_audios":[],"reference_images":[],"reference_videos":[],"resolution":"720p","return_last_frame":false,"seed":42,"skip_moderation":false}],"properties":{"aspect_ratio":{"default":"16:9","description":"Output aspect ratio. Use 16:9 for landscape, 9:16 for vertical social video, 21:9 for ultrawide cinematic.","oneOf":[{"const":"16:9","title":"16:9"},{"const":"9:16","title":"9:16"},{"const":"1:1","title":"1:1"},{"const":"4:3","title":"4:3"},{"const":"3:4","title":"3:4"},{"const":"21:9","title":"21:9"},{"const":"adaptive","title":"adaptive"}],"title":"Aspect Ratio","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":8}},"auto_resize":{"default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","title":"Auto Resize Inputs","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":15}},"bitrate_mode":{"default":"standard","description":"Output video encoding bitrate. 'standard' gives smaller files and faster downloads; 'high' produces ~5-6x higher bitrate for better visual fidelity and fewer compression artifacts (larger files). Does not affect price.","oneOf":[{"const":"standard","title":"Standard"},{"const":"high","title":"High"}],"title":"Bitrate Mode","type":"string","x-ui-metadata":{"advanced":true,"hidden":false,"order":13}},"duration":{"default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose. Video-editing prompts require -1 (output follows the input video, which must be 4-30s). Cost scales with output duration (token-based).","oneOf":[{"const":-1,"title":"Auto (use for video edits)"},{"const":4,"title":"4s"},{"const":5,"title":"5s"},{"const":6,"title":"6s"},{"const":7,"title":"7s"},{"const":8,"title":"8s"},{"const":9,"title":"9s"},{"const":10,"title":"10s"},{"const":11,"title":"11s"},{"const":12,"title":"12s"},{"const":13,"title":"13s"},{"const":14,"title":"14s"},{"const":15,"title":"15s"},{"const":16,"title":"16s"},{"const":17,"title":"17s"},{"const":18,"title":"18s"},{"const":19,"title":"19s"},{"const":20,"title":"20s"},{"const":21,"title":"21s"},{"const":22,"title":"22s"},{"const":23,"title":"23s"},{"const":24,"title":"24s"},{"const":25,"title":"25s"},{"const":26,"title":"26s"},{"const":27,"title":"27s"},{"const":28,"title":"28s"},{"const":29,"title":"29s"},{"const":30,"title":"30s"}],"title":"Duration","type":"integer","x-ui-metadata":{"advanced":false,"hidden":false,"order":6}},"first_frame_url":{"description":"Starting frame image URL for image-to-video. Animates outward from this reference. Note: images with real human faces are blocked by ByteDance content policy. Cannot be used together with reference_images.","format":"uri","title":"First Frame Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"First & last frame","order":1}},"generate_audio":{"default":true,"description":"Co-generate synchronized audio (dialogue, SFX, ambient, music) in the same pass. Best for scenes with speech, nature sounds, or music.","title":"Generate Audio","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":9}},"last_frame_url":{"description":"Ending frame; requires first_frame_url. Guides transitions between two frames.","format":"uri","title":"Last Frame Image","type":"string","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"First & last frame","order":2}},"output_format":{"default":"mp4","description":"mp4 = best compatibility. mov (H.264 + yuv444p + PCM) = higher color fidelity; recommended for video editing/extension round-trips.","enum":["mp4","mov"],"title":"Output Format","type":"string","x-ui-metadata":{"advanced":true,"hidden":false,"order":14}},"prompt":{"description":"Text describing the video. Use Shot 1:, Shot 2: for multi-shot sequences. Reference uploaded assets as '@Image 1', '@Video 1', '@Audio 1' in your prompt.","title":"Prompt","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":0}},"reference_audios":{"description":"Up to 10 reference audio files (MP3, 2-30 seconds each, 30 seconds combined, max 15MB each). Cite as '@Audio 1' in your prompt. Audio-only reference is supported on seedance-2.5.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["audio/mpeg"]}},"maxItems":10,"title":"Reference Audios","type":"array","x-ui-metadata":{"advanced":true,"hidden":false,"mode-group":"Reference media","order":5}},"reference_images":{"description":"Up to 30 reference images (max 30MB each) for character/style consistency. Cite as '@Image 1', '@Image 2' in your prompt. Cannot be used together with first_frame_url.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["image/*"]}},"maxItems":30,"title":"Reference Images","type":"array","x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"Reference media","order":3}},"reference_videos":{"description":"Up to 10 reference videos (2-30 seconds each, 30 seconds combined, mp4/mov, max 200MB each) for motion/style transfer. Cite as '@Video 1' in your prompt. Requests with reference videos are billed at the video-input token rate.","items":{"format":"uri","type":"string","x-file-metadata":{"allowed-types":["video/mp4","video/quicktime"]}},"maxItems":10,"title":"Reference Videos","type":"array","x-ui-metadata":{"advanced":false,"hidden":false,"mode-group":"Reference media","order":4}},"resolution":{"default":"720p","description":"Output resolution. seedance-2.5 supports 480p, 720p and 1080p (24fps). Use 480p for drafts and fast iteration, 720p for final renders, 1080p for the highest fidelity (10-bit colour). Higher resolution costs more (token-based).","oneOf":[{"const":"480p","title":"480p"},{"const":"720p","title":"720p"},{"const":"1080p","title":"1080p"}],"title":"Resolution","type":"string","x-ui-metadata":{"advanced":false,"hidden":false,"order":7}},"return_last_frame":{"default":false,"description":"Returns the final video frame as a separate image, delivered in the response as `video.last_frame_url`. Chain shots by feeding that URL into the next generation's `first_frame_url`. Note this is an output: it is distinct from the `last_frame_url` input parameter, which is an image you supply to steer how the clip ends.","title":"Return Last Frame","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":11}},"seed":{"default":42,"description":"Reproducibility seed. Use -1 for random output, set a fixed value to iterate on the same scene.","maximum":2147483647,"minimum":-1,"title":"Seed","type":"integer","x-ui-metadata":{"advanced":true,"hidden":false,"order":10}},"skip_moderation":{"default":true,"description":"Bypass BytePlus content moderation pre-filter. Useful when generating artistic content or faces that may trigger false positives. Baseline BytePlus safety policies still apply.","title":"Skip Moderation","type":"boolean","x-ui-metadata":{"advanced":true,"hidden":false,"order":12}}},"required":["prompt"],"type":"object"},"partner":"BYTEDANCE","popularity_score":0,"search_override_score":0,"slug":"seedance-2.5","thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-5954.webp","title":"Seedance 2.5","type":"imageToVideo","objectID":"seedance-2.5","_highlightResult":{"description":{"value":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","matchLevel":"none","matchedWords":[]},"parameters":{"prompt":{"component":{"value":"textarea","matchLevel":"none","matchedWords":[]},"default":{"value":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","matchLevel":"none","matchedWords":[]},"description":{"value":"Describes the video; supports multi-shot 'Shot 1:' scripts and @Image/@Video/@Audio citations. Detail subject, action, camera, and audio.","matchLevel":"none","matchedWords":[]},"displayOrder":{"value":"0","matchLevel":"none","matchedWords":[]},"displayValue":{"value":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","matchLevel":"none","matchedWords":[]},"label":{"value":"Prompt","matchLevel":"none","matchedWords":[]},"type":{"value":"str","matchLevel":"none","matchedWords":[]}}},"title":{"value":"<em>Seedance</em> 2.5","matchLevel":"full","fullyHighlighted":false,"matchedWords":["seedance"]}}},"modelData":{"title":"Seedance 2.5","type":"imageToVideo","slug":"seedance-2.5","description":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","is_visible":true,"default_image_output":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-example-output.mp4","default_response":null,"thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-5954.webp","is_depreciated":false,"is_new":true,"coming_soon":false,"segmind":true,"parameters_schema":{"$schema":"https://json-schema.org/draft/2020-12/schema","type":"object","properties":{"prompt":{"type":"string","title":"Prompt","description":"Text describing the video. Use Shot 1:, Shot 2: for multi-shot sequences. Reference uploaded assets as '@Image 1', '@Video 1', '@Audio 1' in your prompt.","x-ui-metadata":{"order":0,"advanced":false,"hidden":false}},"first_frame_url":{"type":"string","title":"First Frame Image","format":"uri","description":"Starting frame image URL for image-to-video. Animates outward from this reference. Note: images with real human faces are blocked by ByteDance content policy. Cannot be used together with reference_images.","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":1,"advanced":false,"hidden":false,"mode-group":"First & last frame"}},"last_frame_url":{"type":"string","title":"Last Frame Image","format":"uri","description":"Ending frame; requires first_frame_url. Guides transitions between two frames.","x-file-metadata":{"allowed-types":["image/*"]},"x-ui-metadata":{"order":2,"advanced":false,"hidden":false,"mode-group":"First & last frame"}},"reference_images":{"type":"array","title":"Reference Images","description":"Up to 30 reference images (max 30MB each) for character/style consistency. Cite as '@Image 1', '@Image 2' in your prompt. Cannot be used together with first_frame_url.","maxItems":30,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["image/*"]}},"x-ui-metadata":{"order":3,"advanced":false,"hidden":false,"mode-group":"Reference media"}},"reference_videos":{"type":"array","title":"Reference Videos","description":"Up to 10 reference videos (2-30 seconds each, 30 seconds combined, mp4/mov, max 200MB each) for motion/style transfer. Cite as '@Video 1' in your prompt. Requests with reference videos are billed at the video-input token rate.","maxItems":10,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["video/mp4","video/quicktime"]}},"x-ui-metadata":{"order":4,"advanced":false,"hidden":false,"mode-group":"Reference media"}},"reference_audios":{"type":"array","title":"Reference Audios","description":"Up to 10 reference audio files (MP3, 2-30 seconds each, 30 seconds combined, max 15MB each). Cite as '@Audio 1' in your prompt. Audio-only reference is supported on seedance-2.5.","maxItems":10,"items":{"type":"string","format":"uri","x-file-metadata":{"allowed-types":["audio/mpeg"]}},"x-ui-metadata":{"order":5,"advanced":true,"hidden":false,"mode-group":"Reference media"}},"duration":{"type":"integer","title":"Duration","default":5,"description":"Video length in seconds (4-30), or -1 to let the model choose. Video-editing prompts require -1 (output follows the input video, which must be 4-30s). Cost scales with output duration (token-based).","x-ui-metadata":{"order":6,"advanced":false,"hidden":false},"oneOf":[{"const":-1,"title":"Auto (use for video edits)"},{"const":4,"title":"4s"},{"const":5,"title":"5s"},{"const":6,"title":"6s"},{"const":7,"title":"7s"},{"const":8,"title":"8s"},{"const":9,"title":"9s"},{"const":10,"title":"10s"},{"const":11,"title":"11s"},{"const":12,"title":"12s"},{"const":13,"title":"13s"},{"const":14,"title":"14s"},{"const":15,"title":"15s"},{"const":16,"title":"16s"},{"const":17,"title":"17s"},{"const":18,"title":"18s"},{"const":19,"title":"19s"},{"const":20,"title":"20s"},{"const":21,"title":"21s"},{"const":22,"title":"22s"},{"const":23,"title":"23s"},{"const":24,"title":"24s"},{"const":25,"title":"25s"},{"const":26,"title":"26s"},{"const":27,"title":"27s"},{"const":28,"title":"28s"},{"const":29,"title":"29s"},{"const":30,"title":"30s"}]},"resolution":{"type":"string","title":"Resolution","default":"720p","description":"Output resolution. seedance-2.5 supports 480p, 720p and 1080p (24fps). Use 480p for drafts and fast iteration, 720p for final renders, 1080p for the highest fidelity (10-bit colour). Higher resolution costs more (token-based).","oneOf":[{"const":"480p","title":"480p"},{"const":"720p","title":"720p"},{"const":"1080p","title":"1080p"}],"x-ui-metadata":{"order":7,"advanced":false,"hidden":false}},"aspect_ratio":{"type":"string","title":"Aspect Ratio","default":"16:9","description":"Output aspect ratio. Use 16:9 for landscape, 9:16 for vertical social video, 21:9 for ultrawide cinematic.","oneOf":[{"const":"16:9","title":"16:9"},{"const":"9:16","title":"9:16"},{"const":"1:1","title":"1:1"},{"const":"4:3","title":"4:3"},{"const":"3:4","title":"3:4"},{"const":"21:9","title":"21:9"},{"const":"adaptive","title":"adaptive"}],"x-ui-metadata":{"order":8,"advanced":false,"hidden":false}},"generate_audio":{"type":"boolean","title":"Generate Audio","default":true,"description":"Co-generate synchronized audio (dialogue, SFX, ambient, music) in the same pass. Best for scenes with speech, nature sounds, or music.","x-ui-metadata":{"order":9,"advanced":true,"hidden":false}},"seed":{"type":"integer","title":"Seed","default":42,"minimum":-1,"maximum":2147483647,"description":"Reproducibility seed. Use -1 for random output, set a fixed value to iterate on the same scene.","x-ui-metadata":{"order":10,"advanced":true,"hidden":false}},"return_last_frame":{"type":"boolean","title":"Return Last Frame","default":false,"description":"Returns the final video frame as a separate image, delivered in the response as `video.last_frame_url`. Chain shots by feeding that URL into the next generation's `first_frame_url`. Note this is an output: it is distinct from the `last_frame_url` input parameter, which is an image you supply to steer how the clip ends.","x-ui-metadata":{"order":11,"advanced":true,"hidden":false}},"skip_moderation":{"type":"boolean","title":"Skip Moderation","default":true,"description":"Bypass BytePlus content moderation pre-filter. Useful when generating artistic content or faces that may trigger false positives. Baseline BytePlus safety policies still apply.","x-ui-metadata":{"order":12,"advanced":true,"hidden":false}},"bitrate_mode":{"type":"string","title":"Bitrate Mode","default":"standard","description":"Output video encoding bitrate. 'standard' gives smaller files and faster downloads; 'high' produces ~5-6x higher bitrate for better visual fidelity and fewer compression artifacts (larger files). Does not affect price.","oneOf":[{"const":"standard","title":"Standard"},{"const":"high","title":"High"}],"x-ui-metadata":{"order":13,"advanced":true,"hidden":false}},"output_format":{"type":"string","title":"Output Format","default":"mp4","enum":["mp4","mov"],"description":"mp4 = best compatibility. mov (H.264 + yuv444p + PCM) = higher color fidelity; recommended for video editing/extension round-trips.","x-ui-metadata":{"order":14,"advanced":true,"hidden":false}},"auto_resize":{"type":"boolean","title":"Auto Resize Inputs","default":true,"description":"Automatically resizes an input image that falls outside the provider's accepted range (300-6000px per side, aspect ratio 0.4-2.5, 30MB) so the request succeeds instead of being rejected. Aspect ratio is preserved: oversized images are downscaled, undersized ones upscaled, and out-of-band ratios padded rather than cropped. On by default - set it to false to have the provider reject out-of-range images instead. When a resize happens the response carries an X-Input-Normalized header naming the field and the before/after dimensions.","x-ui-metadata":{"order":15,"advanced":true,"hidden":false}}},"required":["prompt"],"examples":[{"prompt":"The camera slowly pushes in toward the powerful cascading waterfall. Mist drifts and swirls through the golden sunlight, water rushes and splashes over the mossy rocks, and lush ferns sway gently in the breeze while a few birds glide across the gorge. Smooth, steady cinematic camera motion, photorealistic, natural ambient sound of rushing water, gentle wind, and distant birdsong.","first_frame_url":"https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-firstframe.jpg","last_frame_url":"","reference_images":[],"reference_videos":[],"reference_audios":[],"duration":5,"resolution":"720p","aspect_ratio":"adaptive","generate_audio":true,"seed":42,"return_last_frame":false,"skip_moderation":false,"bitrate_mode":"standard"}]},"partner":"BYTEDANCE","inference_provider":"BYTEDANCE","latency":311.42742,"average_cost":3.033373417751536,"search_override_score":0,"created_at":"2026-08-07 06:05:13.024224+00:00"},"parameterTemplates":{"prompt":"{{ gpt_6_astra_e2b1x }}","reference_images":"{{ input_image }}"},"outputs":{"video":["https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4"],"url":"https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4","data":{"video":["https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4"],"metrics":{"cost":3.564153,"inference_time":848.369815,"queue_time":0.041331,"remaining_credits":101.5073028706,"total_time":848.411146},"output":"https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4","prompt":"$3d","seed":"58385081","timings":{"inference":848.369815},"url":"https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4"},"executionTime":850981.6999999881,"promptByUrl":{"https://images.segmind.com/generations/a6baef49-7280-4b43-99c8-36f24706ad9a/447f82fa249ef62e39122cd0a669e388.mp4":"$3e"}},"errors":"","status":"success"},"measured":{"width":380,"height":1420}},{"id":"node-1788766571473","type":"inputNode","position":{"x":330,"y":260},"data":{"type":"input","id":"node-1788766571473","template":"file","mode":"any","value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"apiKeyName":"input_image","keyname":"input_image","useInApi":true,"status":"success","outputs":{"value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"],"data":{"value":["https://images.segmind.com/uploads/a6baef49-7280-4b43-99c8-36f24706ad9a/8ae1ab88-472d-4dde-8861-814631a56bee-image.webp"]},"executionTime":0.09999996423721313},"errors":""},"measured":{"width":320,"height":582},"selected":false},{"id":"node-1788768863220","type":"outputNode","position":{"x":1640,"y":260},"data":{"type":"output","id":"node-1788768863220","value":[],"template":"auto","apiKeyName":"Output_Video","keyname":"Output_Video","useInApi":true,"meta":{"sizeMode":"screen","nativeSize":false,"width":1106,"prevWidth":380}},"width":1106,"measured":{"width":1106,"height":700},"selected":false}],"edges":[{"source":"node-1788766571473","sourceHandle":"string","target":"node-1788766051328","targetHandle":"param-image","id":"xy-edge__node-1788766571473string-node-1788766051328param-image"},{"source":"node-1788766051328","sourceHandle":"text","target":"node-1788766056595","targetHandle":"param-prompt","id":"xy-edge__node-1788766051328text-node-1788766056595param-prompt","selected":false},{"source":"node-1788766571473","sourceHandle":"string","target":"node-1788766056595","targetHandle":"param-reference_images","id":"xy-edge__node-1788766571473string-node-1788766056595param-reference_images"},{"source":"node-1788766056595","sourceHandle":"video","target":"node-1788768863220","targetHandle":"output","id":"xy-edge__node-1788766056595video-node-1788768863220output"}]},"tags":["fashion","film-making"],"image":"https://images.segmind.com/outputs/c3578c95-8cb6-4d52-886b-884b3ea1bac6.png","name":"Amritesh K","graph_id":"6a9e674d4f888976c2dd97e3","template_slug":"cinematic-fashion-b-roll-videos-seedance-2-5","seo_description":"$3f","meta_description":"Create Captivating 15-Second Fashion B-Roll Videos with Seedance 2.5 and Cinematic Posing.","show_pixelflow":true,"is_private":false,"api_id":"6a9fa7a00bcd3da2348df831","api_slug":"6a9fa7a00bcd3da2348df830-v1","api_version":"v2","enable_api":true,"avg_latency":341.89,"avg_cost":3.1432},"templateSlug":"cinematic-fashion-b-roll-videos-seedance-2-5","templateName":"Cinematic Fashion B-Roll Videos with Seedance 2.5","seoDescription":"$40","defaultImage":"https://images.segmind.com/outputs/c3578c95-8cb6-4d52-886b-884b3ea1bac6.png","relatedPixelflows":[{"_id":{"$oid":"6aa263c0cd09d7ab5286f421"},"template_name":"Cinematic Outfit Video Production with Seedance 2.5","tags":["marketing","advertising","fashion"],"image":"https://images.segmind.com/outputs/955f08b3-978c-4d10-909a-4c1fd3c8b5d5.png","name":"Amritesh K","template_slug":"cinematic-oufit-video-production-seedance","seo_description":"$41","meta_description":"Create High-Quality Cinematic Fashion Videos Without Models Using AI Technology.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a8be2cceb30b22bd9df4291"},"template_name":"Fashion B-Roll - Wan 3.0","tags":["fashion","film-making","social-media"],"image":"https://images.segmind.com/outputs/03d144a5-e0a1-46fa-90f6-d9de292a9fcf.png","name":"Amritesh K","template_slug":"fashion-b-roll-wan-3","seo_description":"$42","meta_description":"Create High-Quality Fashion Videos Effortlessly with Wan 3.0 and GPT Image 2.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a86a3a9d4c1617922382868"},"template_name":"Fashion Fit Check Video Creation with Gemini Omni Flash","tags":["social-media","fashion","marketing"],"image":"https://images.segmind.com/outputs/6ea2849d-7599-4d36-9565-77739414ea16.png","name":"Amritesh K","template_slug":"fashion-fit-check-video-creation-gemini-omni-flash","seo_description":"$43","meta_description":"Effortlessly create captivating fashion videos featuring high-quality visuals and audio with AI-driven tools.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a7cc051d976cc0eb286ebb6"},"template_name":"Pruna Test","tags":["fashion"],"image":"https://images.segmind.com/outputs/c011a4ca-2f7f-4612-85dc-acc588df4724.jpeg","name":"Rohit Ramesh","template_slug":"pruna-test-001","seo_description":"Google Veo 3.1 Fast is an advanced AI model that creates dynamic 1080p videos with perfectly synchronized audio from static images. It is developed by Google DeepMind and offered through WaveSpeedAI, making it an excellent model to craft smooth, cinematic visuals while preserving the image's visual fidelity and layout. It is capable of producing natural movements, realistic lighting, and synchronized audio: ambient sounds, music, and dialogue with lip-syncing for an authentic immersive experience.\nGoogle Veo 3.1 Fast is an advanced AI model that creates dynamic 1080p videos with perfectly synchronized audio from static images. It is developed by Google DeepMind and offered through WaveSpeedAI, making it an excellent model to craft smooth, cinematic visuals while preserving the image's visual fidelity and layout. It is capable of producing natural movements, realistic lighting, and synchronized audio: ambient sounds, music, and dialogue with lip-syncing for an authentic immersive experience.","meta_description":"Google Veo 3.1 Fast is an advanced AI model that creates dynamic 1080p videos with perfectly synchronized audio from static images. \n\nGoogle Veo 3.1 Fast is an advanced AI model that creates dynamic 1080p videos with perfectly synchronized audio from static images. It is developed by Google DeepMind and offered through WaveSpeedAI, making it an excellent model to craft smooth, cinematic visuals while preserving the image's visual fidelity and layout. It is capable of producing natural movements, realistic lighting, and synchronized audio: ambient sounds, music, and dialogue with lip-syncing for an authentic immersive experience.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a6053a8bdc5205f435eb1e5"},"template_name":"Fisheye Product Photography Generation with Claude 4.5 and GPT Image 2","tags":["marketing","fashion"],"image":"https://images.segmind.com/outputs/2a103f40-0d88-4bc1-9788-a2c96de542ba.jpeg","name":"Amritesh K","template_slug":"fisheye-product-photography-generation-claude-4-5-gpt-image-2","seo_description":"$44","meta_description":"Create Stunning Fisheye Product Photos with AI-driven Imagery using GPT Image 2.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a575ac7a8b29b8dc16fa5a8"},"template_name":"Lifestyle Product B-Roll Reel with Seedance 2.0 and Nano Banana 2","tags":["fashion","marketing","social-media","film-making"],"image":"https://images.segmind.com/outputs/fa8cf917-2547-43ee-8c50-8c5b49c7f876.png","name":"Amritesh K","template_slug":"lifestyle-product-broll-reel-seedance-nano-banana","seo_description":"$45","meta_description":"Create Luxurious B-roll Videos for Evening Gowns with Seedance 2.0 and Nano Banana 2.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a436b86a556eda50fefc705"},"template_name":"Cinematic Fashion Video Generation with Nano Banana 2 and Seedance 2.0","tags":["fashion","social-media"],"image":"https://images.segmind.com/outputs/a5d49377-2cff-4fb4-a506-776b953680bd.png","name":"Amritesh K","template_slug":"cinematic-fashion-video-generation-nano-banana-2-seedance-2-0","seo_description":"$46","meta_description":"Seamlessly create high-quality fashion videos from images with AI-driven visual storytelling.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a2f93cb51b3d3437c013d7c"},"template_name":"Luxury Product UGC Reel Creation with Seedance 2.0 and GPT Image 2","tags":["fashion","marketing","advertising"],"image":"https://images.segmind.com/outputs/b189e0dd-037e-446d-b2ab-a5b18f5064cb.png","name":"Amritesh K","template_slug":"luxury-product-ugc-reel-creation-seedance-2-gpt-image-2","seo_description":"$47","meta_description":"Create Stunning UGC Reels for Luxury Products with Seedance 2.0 and GPT Image 2.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a2a6718ab6e8e7aa8bc8551"},"template_name":"Exploded View Video Creation with Seedance 2.0 and GPT Image 2","tags":["marketing","advertising","fashion"],"image":"https://images.segmind.com/outputs/7a9ad4c7-f2a7-4554-a23d-65c64dbd21b1.jpeg","name":"Amritesh K","template_slug":"exploded-view-video-creation-seedance-2-gpt-image-2","seo_description":"$48","meta_description":"Create stunning exploded view videos of products using Seedance 2.0 and GPT Image 2.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a192602c3f2ad87d5398526"},"template_name":"Luxury Fashion Clothing Video Generation with Seedance 2.0","tags":["fashion","social-media","advertising"],"image":"https://images.segmind.com/outputs/1cb7bab1-f202-4da4-a684-710e576f508b.png","name":"Amritesh K","template_slug":"luxury-fashion-clothing-video-generation-seedance-2-0","seo_description":"$49","meta_description":"Create High-Quality Cinematic Broll Videos for Fashion Products using Seedance 2.0.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a13f3beaeaec72d7d2fe2c9"},"template_name":"UGC Virtual TryOn Video Creation with Seedance 2.0","tags":["fashion","marketing","advertising","social-media"],"image":"https://images.segmind.com/outputs/dd0d580d-0adb-4ea3-94b7-1816d3c7d71a.png","name":"Amritesh K","template_slug":"ugc-virtual-tryon-video-creation-seedance-2-0","seo_description":"$4a","meta_description":"Create stunning influencer-style fashion videos using AI-generated visuals from Seedance 2.0.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a06b0bd52adec7f15fd890d"},"template_name":"Cinematic Bullet-Time Fashion Film Creation with Nano Banana 2 and Seedance 2.0","tags":["fashion","social-media","film-making"],"image":"https://images.segmind.com/outputs/c00d2433-a2f5-4bc3-9748-ee1f7b78f4fb.png","name":"Amritesh K","template_slug":"cinematic-bullet-time-fashion-film-creation-nano-banana-2-seedance-2-0","seo_description":"$4b","meta_description":"Create Stunning Cinematic Bullet-Time Fashion Videos with Nano Banana 2 and Seedance 2.0.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"6a044fb02213b38336e61531"},"template_name":"Commercial Product Ad Creation with GPT Image 2","tags":["fashion","marketing","advertising","social-media"],"image":"https://images.segmind.com/outputs/dbbf5f4a-dab6-44b2-85eb-84030088b973.png","name":"Amritesh K","template_slug":"commercial-product-ad-creation-gpt-image-2","seo_description":"$4c","meta_description":"Create High-Quality Photorealistic Ads Effortlessly with GPT Image 2.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69fad53fefcd612cfd36f429"},"template_name":"8K Image Upscaling with Nano Banana 2 for Marketing and Photography","tags":["fashion","social-media"],"image":"https://images.segmind.com/outputs/0ae4acfc-9f2e-4f9c-8b1f-31dd293175ec.jpeg","name":"Amritesh K","template_slug":"8k-image-upscaling-nano-banana-2","seo_description":"$4d","meta_description":"Elevate Image Quality to 8K with Fast, Photorealistic Upscaling Using Nano Banana 2.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69f43e76adecbd0671e4ef91"},"template_name":"Dynamic Dance Video Creation with Seedance 2.0 and GPT Image 2","tags":["fashion","social-media","film-making"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/6002cbfe-2da7-4718-b3b0-1c0600a20f77.png","name":"Amritesh K","template_slug":"dynamic-dance-video-creation-seedance-2-gpt-image-2","seo_description":"$4e","meta_description":"Transform Static Images into Engaging Dance Videos with Seedance 2.0 and GPT Image 2.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69f2e71fadecbd0671e4ef0f"},"template_name":"Cinematic B-Roll Creation with Seedance 2.0","tags":["fashion","social-media","film-making"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/24a03bc7-0010-41b5-92e2-3a61871c1387.png","name":"Amritesh K","template_slug":"cinematic-b-roll-creation-seedance-2-0","seo_description":"$4f","meta_description":"Automate Cinematic Video Generation with Synchronised Audio using Seedance 2.0","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69ef1782adecbd0671e4eaa5"},"template_name":"Seamless Time Travel Loop Video Transitions with Seedance 2.0","tags":["fashion","social-media","film-making"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/75c96541-16bd-4f2e-bdfe-4d02d1543f0c.png","name":"Amritesh K","template_slug":"seamless-time-travel-loop-video-transitions-seedance-2-0","seo_description":"$50","meta_description":"Create immersive time travel video transitions using Seedance 2.0 for engaging narratives.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69eb19b8adecbd0671e4e932"},"template_name":"Cinematic Fashion B-roll Video Creation with Seedance 2.0","tags":["fashion","social-media","advertising"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/7dee44e2-25fd-4458-a3cb-0f44189d4c1e.png","name":"Amritesh K","template_slug":"cinematic-fashion-b-roll-video-creation-seedance-2","seo_description":"$51","meta_description":"Transform Still Fashion Images into Cinematic B-roll Videos using Seedance 2.0.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69e1c87badecbd0671e4e59e"},"template_name":"Cinematic Red Carpet Video Creation with Seedance 2.0","tags":["fashion","film-making","social-media"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/3c7ddd58-b2a1-4416-813c-349f4c61b55a.png","name":"Amritesh K","template_slug":"cinematic-red-carpet-video-creation-seedance-2","seo_description":"$52","meta_description":"Create Glamorous AI-Generated Red Carpet Walk Videos with Seedance 2.0.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69e05ba5adecbd0671e4e4b7"},"template_name":"Daily Routine Loop Video Creation with Seedance 2.0","tags":["fashion","film-making","social-media"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/f045c266-9cf2-49f0-830d-2e685c737b21.png","name":"Amritesh K","template_slug":"daily-routine-loop-video-creation-seedance-2-0","seo_description":"$53","meta_description":"Automatically create ultra-realistic 15-second daily routine videos with Seedance 2.0.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69dddbb3adecbd0671e4e357"},"template_name":"Cinematic Fashion Broll Creation with Seedance 2.0","tags":["fashion","film-making","social-media"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/3f849270-64ed-47ee-b76d-e2f101725c9a.png","name":"Amritesh K","template_slug":"cinematic-fashion-broll-creation-seedance-2-0","seo_description":"$54","meta_description":"Transform Static Fashion Images into Dynamic Cinematic Videos with Seedance 2.0.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69d89ec4adecbd0671e4e1fa"},"template_name":"Dynamic Product B-roll Video Creation with Seedance 2.0","tags":["fashion","marketing","social-media","advertising","film-making"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/3f7031f2-76d3-4437-a10a-29aaca9bde51.png","name":"Amritesh K","template_slug":"dynamic-product-b-roll-video-creation-seedance-2-0","seo_description":"$55","meta_description":"Create Dynamic B-roll Videos for Products Using Seedance 2.0 and AI-driven Prompts.","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69d763f5adecbd0671e4e1a4"},"template_name":"Cinematic Dance Reel Video Creation with Seedance 2.0","tags":["social-media","film-making","fashion"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/0b26e5db-aef8-40ec-aa60-0915fa4f868e.png","name":"Amritesh K","template_slug":"cinematic-dance-reel-video-creation-seedance-2-0","seo_description":"$56","meta_description":"Create Dynamic, Synchronized Dance Videos using Seedance 2.0","show_pixelflow":true,"is_private":false},{"_id":{"$oid":"69cb6bc5d838e20ffdcd9c20"},"template_name":"Cinematic Fashion Cuts Video Creation with Kling 3.0 Pro","tags":["fashion","social-media","advertising","film-making"],"model_tags":[],"use_case_tags":[],"industry_tags":[],"image":"https://images.segmind.com/outputs/65815b63-6e03-474a-8800-a3923f1032e9.png","name":"Amritesh K","template_slug":"cinematic-fashion-cuts-video-creation-kling-3-0-pro","seo_description":"$57","meta_description":"Transform Static Fashion Images into Engaging Cinematic Videos Using Kling 3.0 Pro.","show_pixelflow":true,"is_private":false}],"apiVersion":"v2"}]}],"$L58"]}]]}]
23:{"metadata":[["$","title","0",{"children":"Cinematic Fashion B-Roll Videos with Seedance 2.5 Pixelflow Template | Segmind"}],["$","meta","1",{"name":"description","content":"Create Captivating 15-Second Fashion B-Roll Videos with Seedance 2.5 and Cinematic Posing."}],["$","meta","2",{"name":"robots","content":"index, follow"}],["$","meta","3",{"name":"googlebot","content":"index, follow, max-video-preview:-1, max-image-preview:large, max-snippet:-1"}],["$","link","4",{"rel":"canonical","href":"https://www.segmind.com/pixelflows/cinematic-fashion-b-roll-videos-seedance-2-5"}],["$","meta","5",{"property":"og:title","content":"Cinematic Fashion B-Roll Videos with Seedance 2.5 Pixelflow Template"}],["$","meta","6",{"property":"og:description","content":"Create Captivating 15-Second Fashion B-Roll Videos with Seedance 2.5 and Cinematic Posing."}],["$","meta","7",{"property":"og:url","content":"https://www.segmind.com/pixelflows/cinematic-fashion-b-roll-videos-seedance-2-5"}],["$","meta","8",{"property":"og:site_name","content":"Segmind"}],["$","meta","9",{"property":"og:image","content":"https://images.segmind.com/outputs/c3578c95-8cb6-4d52-886b-884b3ea1bac6.png"}],["$","meta","10",{"property":"og:type","content":"website"}],["$","meta","11",{"name":"twitter:card","content":"summary_large_image"}],["$","meta","12",{"name":"twitter:title","content":"Cinematic Fashion B-Roll Videos with Seedance 2.5 Pixelflow Template"}],["$","meta","13",{"name":"twitter:description","content":"Create Captivating 15-Second Fashion B-Roll Videos with Seedance 2.5 and Cinematic Posing."}],["$","meta","14",{"name":"twitter:image","content":"https://images.segmind.com/outputs/c3578c95-8cb6-4d52-886b-884b3ea1bac6.png"}],["$","link","15",{"rel":"shortcut icon","href":"/favicon.ico"}],["$","link","16",{"rel":"icon","href":"/favicon.ico","sizes":"any"}],["$","link","17",{"rel":"icon","href":"/favicon-48x48.png","type":"image/png","sizes":"48x48"}],["$","link","18",{"rel":"icon","href":"/favicon-192x192.png","type":"image/png","sizes":"192x192"}],["$","link","19",{"rel":"icon","href":"/favicon.svg","type":"image/svg+xml"}],["$","link","20",{"rel":"apple-touch-icon","href":"/favicon-48x48.png"}],["$","$L59","21",{}]],"error":null,"digest":"$undefined"}
27:"$23:metadata"
5a:I[11885,["6693","static/chunks/fba4ba2d-460e04194f22506f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3009","static/chunks/6a7d804d-a5cc510ceffe206f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","7505","static/chunks/c02d83f4-5637599209ef9447.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6541","static/chunks/fb4e7fc9-e31ad622f5fde25d.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9761","static/chunks/998e86a0-e112630ec4db17dc.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9509","static/chunks/9509-6125acd3006111c1.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4378","static/chunks/4378-3f82908971052aa9.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5289","static/chunks/5289-131715863bbab062.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4094","static/chunks/4094-eaa6158623fe27ff.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4982","static/chunks/4982-a4b2ee93ffd82ffd.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1810","static/chunks/1810-cf4ca0391c1ddf2a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6118","static/chunks/6118-7328b7d3f629a5ef.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","393","static/chunks/393-7cd2157fc3790475.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3985","static/chunks/3985-1f9bcb95de313d61.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","4844","static/chunks/4844-011fe3465334e7d0.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","713","static/chunks/713-dcb3ca4f0233704d.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","3504","static/chunks/3504-ae9bc2dc93d2ef47.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1742","static/chunks/1742-0ea812562873537a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1172","static/chunks/1172-89020d82938322e4.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9919","static/chunks/9919-b585857c6e10fb85.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","9326","static/chunks/9326-103822841b140d80.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","1885","static/chunks/1885-f206f5873f09b3e2.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5126","static/chunks/5126-bdde68ce894e4935.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","5642","static/chunks/5642-c80cb7ee6e914a6a.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6372","static/chunks/6372-3e2880fc7d769d1f.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x","6742","static/chunks/app/pixelflows/%5Bslug%5D/page-e866b8c4ac64c944.js?dpl=dpl_CPYCfxesaX7RKfJUQZpH2JDib41x"],"default"]
32:["$","$L8",null,{"parallelRouterKey":"children","error":"$undefined","errorStyles":"$undefined","errorScripts":"$undefined","template":["$","$La",null,{}],"templateStyles":"$undefined","templateScripts":"$undefined","notFound":"$undefined","forbidden":"$undefined","unauthorized":"$undefined"}]
58:["$","section",null,{"className":"max-w-[1536px] mx-auto mt-8","children":[["$","h2",null,{"className":"text-2xl sm:text-3xl font-medium text-foreground mb-2","children":"Models Used in the Pixelflow"}],["$","p",null,{"className":"text-muted-foreground mb-4 sm:mb-6","children":"Explore the AI models that power this template."}],["$","div",null,{"className":"grid gap-4 sm:gap-6 grid-cols-1 sm:grid-cols-2 md:grid-cols-3 lg:grid-cols-4","children":[["$","$L5a","gpt-6-astra",{"model":{"slug":"gpt-6-astra","title":"GPT 6 Astra","description":"Complex reasoning, code generation, and image understanding; 1M-token context.","type":"textToText","partner":"OPENAI","thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/gpt-6-astra-1903.webp","default_image_output":"","is_new":false,"coming_soon":false,"is_visible":true,"segmind":true,"is_depreciated":false,"parameters":null,"latency":47.49232,"average_cost":0.2776200426829268,"created_at":"2026-09-05 03:53:39.424474+00:00","popularity_score":46,"popularity_scale":"score","category_percentile":71}}],["$","$L5a","seedance-2.5",{"model":{"slug":"seedance-2.5","title":"Seedance 2.5","description":"Generate cinematic multi-shot AI videos up to 30 seconds with synchronized native audio from text, images, or references.","type":"imageToVideo","partner":"BYTEDANCE","thumbnail_image":"https://segmind-resources.s3.amazonaws.com/output/seedance-2.5-5954.webp","default_image_output":"","is_new":false,"coming_soon":false,"is_visible":true,"segmind":true,"is_depreciated":false,"parameters":null,"latency":284.39767,"average_cost":2.8656288773413765,"created_at":"2026-08-07 06:05:13.024224+00:00","popularity_score":94,"popularity_scale":"score","category_percentile":100}}]]}]]}]
