准备并引用参考素材:以 Image1 分镜网格作为角色/场景/道具/构图的精确视觉参考,Image2 为角色表,Image3 为公寓外部与包。 原文:「Use the storyboard grid Image1 as the exact visual reference for character design, apartment layout, clothing, handbag, props, lighting, and composition. Animate the six storyboard frames as one continuous 15-second cinematic sequence. Maintain the character’s exact appearance throughout. Maintain the apartment as one fixed physical environment. Do not alter furniture, architecture, kitchen layout, décor, handbag design, clothing, hairstyle, or character proportions. Movement should feel like polished feature-film 3D character animation: natural body weight, realistic walking rhythm, subtle secondary movement in hair and clothing, expressive facial animation, believable hand interaction with objects, and smooth cinematic camera movement. Jarvis is an AI voice only. He is never physically shown. Character sheet is Image2 Apartment exterior and bag is Image3」
写提示词主体约束块:保持角色外貌与公寓环境固定,运动为精致3D角色动画,Jarvis 仅以 AI 语音出现、永不实体展示。 原文:「Maintain the character’s exact appearance throughout. Maintain the apartment as one fixed physical environment. Do not alter furniture, architecture, kitchen layout, décor, handbag design, clothing, hairstyle, or character proportions. Movement should feel like polished feature-film 3D character animation: natural body weight, realistic walking rhythm, subtle secondary movement in hair and clothing, expressive facial animation, believable hand interaction with objects, and smooth cinematic camera movement. Jarvis is an AI voice only. He is never physically shown.」
写第一组分镜脚本块:按镜头 1–6 描述角色进门、放钥匙、扔包、倒水、听到收益、面部反应等动作、对白、音频与摄影机要求。 原文:「SHOT 1 — 0:00–0:02.5 — 2.5 SECONDS Begin from inside the apartment facing the entrance. The front door opens and the character walks naturally into the apartment after returning from work.」
写第二组分镜脚本块:按镜头 1–6 描述喝水、Jarvis 安排相亲、震惊到黑屏结尾等动作、对白、音频与摄影机要求。 原文:「SHOT 1 — 0:00–0:02.5 — 2.5 SECONDS Begin with the character behind the kitchen counter, matching Panel 1. She raises the clear glass to her lips and takes a natural sip of water.」
添加全局连续性约束块:保持玻璃杯与 Birkin 包位置固定,Jarvis 仅以语音存在、无视觉化身,动作自然克制,无变形或夸张卡通动作。 原文:「The glass stays on the kitchen counter after Shot 3. The Birkin stays on the sofa throughout the entire sequence. Jarvis is voice only. No Jarvis character, man, robot, hologram, speaker, floating interface, screen, glow, or visual manifestation. Keep movement natural and restrained. No duplicates, unnecessary gestures, random head turns, excessive blinking, floating objects, deformed hands, prop morphing, handbag movement, wardrobe changes, hairstyle changes, apartment changes, or exaggerated cartoon motion.」 未给项:工具平台=@budgetpixel;模型版本=GPT Image 2 + Seedance 2.0;时长=exactly 15 seconds;画幅=16:9 landscape;分辨率=原帖未给;参考素材=Image1/@Image1、Image2/@Image2、Image3/@Image3
在提示词中写入开场场景、创作者台词与耳机使用蒙太奇的叙事主体。 原文:「A young creator walks into a bright living room holding a pair of sleek wireless earbuds still inside their charging case. Looking directly into the camera, they smile and say, 'These have honestly become my everyday essential.' They open the case, put the earbuds in, and instantly transition into a quick montage of working on a laptop, making coffee, and taking a short walk outside while music plays.」
在提示词中补充产品特写、收尾台词、定格画面文案与影像风格参数。 原文:「Close-up shots highlight the earbuds' premium design and secure fit. The creator finishes by holding the charging case toward the camera and says, 'Great sound, all-day comfort, and I barely need to recharge them.' End with a clean product hero shot on a table and on-screen text: 'Upgrade your everyday audio.' Natural facial expressions, smooth handheld camera movement, cinematic depth of field, authentic UGC feel, high-quality audio, realistic lighting, premium commercial quality.」 未给项:工具平台=Higgsfield;模型版本=Seedance 2.0 4K;时长=原帖未给;画幅=原帖未给;分辨率=原帖未给;参考素材=原帖未给
选用 Seedance 2.0 on @SocialSight 作为生成平台与模型。 原文:「Seedance 2.0 on @SocialSight」
采用 “Picture to 15 seconds video” 的图片转视频模式。 原文:「-Picture to 15 seconds video」
在提示词中设定输出为 15 秒、竖屏 9:16 的 UGC 风格时尚广告。 原文:「Create a 15-second vertical (9:16) UGC-style fashion advertisement that feels authentic, premium, and social-media ready.」
将 attached reference image 作为图生视频输入,并在提示词中要求模特身穿参考图中完全一致的礼服、不做任何改动。 原文:「The model is wearing the EXACT dress shown in the attached reference image. Do not redesign, modify, recolor, simplify, or reinterpret the dress in any way. Maintain identical embroidery, fabric, silhouette, neckline, sleeves, dupatta, borders, patterns, proportions, textures, embellishments, and overall craftsmanship throughout the entire video.」
写入提示词中的人物与环境描述,指定明亮现代家居、自然晨光、模特动作及发质等。 原文:「A confident young woman stands in a bright, modern home with soft natural morning light. She gently showcases the dress while smiling warmly as if recommending it to a friend. Her hair is long, healthy, shiny, and naturally bouncy.」
在提示词中加入模特对镜头说话的口播脚本。 原文:「She speaks directly to the camera: "I've finally found a dress that feels as beautiful as it looks. The craftsmanship is stunning, the fabric feels luxurious, and every detail looks incredibly premium. It's elegant, comfortable, and honestly makes me feel confident every time I wear it. If you're looking for a timeless outfit, you have to try this!"」
按顺序给出 5 个场景的分镜脚本。 原文:「Storyboard: Scene 1: Close-up of the dress in her hand as she introduces it by gently showcasing the embroidery, fabric texture, and premium craftsmanship. Scene 2: She gently spreads the dupatta and lightly runs her hand across the embroidery, showing the intricate detailing and luxurious fabric movement. Scene 3: Beauty shots of her wearing the dress while walking naturally, adjusting the dupatta, lightly twirling, and showcasing realistic fabric flow, elegant movement, and premium embroidery from different angles. Scene 4: After styling herself, she confidently flips her silky, glossy hair while smiling at the camera, allowing the dress to flow naturally as she turns. Scene 5: Final hero shot holding the dupatta beside her while showcasing the complete outfit against a clean, premium interior background. She smiles and says, "Highly recommend!"」
在提示词末尾加入画质、运镜、表情、转场、唇同步及服装全程可见性等技术要求。 原文:「The video should have realistic smartphone camera quality, natural handheld movement, genuine facial expressions, smooth transitions, accurate lip sync, premium visibility of the dress throughout, and a polished yet authentic UGC aesthetic. The dress remains perfectly identical to the attached reference image, with every embroidery detail, color, fabric texture, silhouette, and design element consistently visible in every shot.」 未给项:工具平台=Seedance 2.0 on @SocialSight;模型版本=Seedance 2.0;时长=15 seconds;画幅=vertical (9:16);分辨率=原帖未给;参考素材=the attached reference image
未给项:工具平台=Seedance 2.0 on @SocialSight;模型版本=Seedance 2.0;时长=15 seconds;画幅=vertical (9:16);分辨率=原帖未给;参考素材=the attached reference image
在 @openart_ai 的 Director 中使用 Seedance 模型 原文:「Product ad made on @openart_ai using Director with Seedance model」
撰写提示词的人物与产品设定段落,固定主角形象并点明智能手表 原文:「Create an ultra-realistic premium smartwatch commercial featuring the same young Korean woman throughout the entire sequence.」
按时间顺序撰写提示词的叙事场景段落,覆盖街道开场、运动、支付、城市使用、屋顶休憩到结尾镜头 原文:「The video opens with the young Korean woman already in frame, smiling naturally toward the camera while walking through a beautiful tree-lined city street.」
在提示词中加入环境音约束与画面纯净度约束 原文:「Natural ambient audio only: birds chirping, footsteps, gentle breeze, smartwatch tap sounds, subtle notification chimes, café ambience, coffee machine sounds, city ambience, leaves rustling. No background music, no subtitles, no logos, no watermarks, and no on-screen text.」
将视频设为15秒时长并采用一镜到底的拍摄方式。 原文:「FORMAT: 15s / ONE CONTINUOUS SHOT」
设定主体角色、环境与整体情绪:让穿着细节丰富迷你裙的女性静止站在灯光明亮的展会地面上,背景为模糊的霓虹灯与剪影,情绪从亲密展示突变为惊悚恐怖。 原文:「SUBJECTS: An alluring, highly attractive female figure. She stands completely still, wearing a highly detailed mini skirt outfit with visible fabric textures, skin pores, and faint perspiration. ENVIRONMENT: A brightly lit convention floor. The background is a blur of neon booth lights and passing silhouettes, heavily grounded in realistic textures. MOOD: Starts as an observational and intimate showcase, twisting sharply into jarring psychological terror.」
设定色彩逻辑为自然胶片冲印 emulation 风格。 原文:「COLOR LOGIC: Naturalistic Film Print Emulation」
拍摄 0–5 秒段落:从极低视角的脚部极特写开始,以 pedestal 上升镜头缓慢上移,使用 50mm 镜头、浅景深,并加入 muffled crowd ambience 与布料摩擦声效。 原文:「0:00-0:05: ECU. Bare feet planted firmly on a polished convention floor. Camera begins a slow, mesmerizing pedestal movement upwards from a worm's eye view, capturing extreme micro-details of skin and floor texture. 50mm lens, shallow depth of field. SFX: (muffled crowd ambience, close fabric rustling).」
拍摄 5–12 秒段落:保持连续不剪切,镜头沿腿部与迷你裙继续上摇,停在下巴位置、脸部不入境,使用 50mm 镜头持续上跟,并加入低频轰鸣声效。 原文:「0:05-0:12: MS. The continuous camera movement smoothly glides up her shapely legs and highly detailed mini skirt outfit, emphasizing her attractive figure. The camera settles precisely at her chin, keeping her full face just out of frame. 50mm lens, continuous upward tracking. SFX: (room tone fades out, low frequency rumble builds).」
拍摄 12–15 秒段落:不切换镜头,让柔软微笑扭曲为苍白恐怖的鬼脸并伸出长舌直逼镜头,使用 50mm 镜头、macro close focus,并加入突兀死寂与潮湿音频 glitch。 原文:「0:12-0:15: CU. Without cutting, her soft smile shudders and distorts, her flesh smoothly instantly twisting into a pale, ghastly supernatural face with wet dark seams. She opens her mouth impossibly wide and extends a long, glistening tongue directly at the camera. 50mm lens, macro close focus. SFX: (sudden dead silence, followed by a visceral wet sound and a harsh audio glitch).」 未给项:工具平台=原帖未给;模型版本=原帖未给;时长=15s;画幅=原帖未给;分辨率=原帖未给;参考素材=原帖未给