加载中…
创作者 · itsSaira_1
查看原始来源 ↗Prompt
VIDEO-TO-VIDEO — USE THE PROVIDED VIDEO AS THE EXACT MOTION/SHOT REFERENCE. 0–3.5s: Keep the girl in the exact same position, pose, movement, camera angle and background; she is climbing/holding the structure while golden energy/wings appear behind her. Replace ONLY her identity/appearance with the supplied GIRL REFERENCE IMAGE. 3.5–5.5s: Iron Man appears exactly as in the reference video — same armor, face, proportions, flight pose, camera angle and movement. DO NOT change Iron Man. 5.5–8.5s: Return to the girl; preserve the exact original body position, facial direction, movement, environment and golden energy. Change ONLY the girl's identity to the supplied reference image. 8.5–12s: Girl falls/floats through the sky exactly like the original video, with the same wings, body motion, camera movement, scale and background. 12–15s: Girl moves/falls beside the structure exactly as shown; preserve all motion, perspective, lighting and golden effects. 15–18s: Close shot of the armored character/girl moving onto the structure; preserve the exact original action, framing and camera movement, replacing only the girl's identity. 18–20.5s: Final hero shot — girl stands exactly like the original, same golden wings, armor, pose, environment, lighting and camera framing; ONLY her face/identity/appearance comes from the supplied GIRL REFERENCE IMAGE. IRON MAN: absolute character lock — same identity, armor, face, proportions, colors, movement and appearance throughout. GIRL: replace ONLY with the supplied reference image while preserving original pose, body motion, clothing/armor structure, wings and scene. NO new actions, NO changed camera, NO scene changes, NO reverse motion, NO U-turn, NO 180° flip, NO camera flip, NO text, NO logos, NO extra characters. Cinematic realistic quality, seamless frame-to-frame continuity.
适用模型与稳定度
建议模型
继续探索
提示语
在 Velvet Skies 🌌❤️ 上花了很多时间,融入了每个场景,并尝试通过 shots 在不说话的情况下传达那些情感…… @higgsfield_ai 中的 Seedance 2.0 👇 完整 prompt package;你只需替换角色。 PACKAGE 1 · 15s · 16:9 · 2 refs · 7 shots, 6 cuts SCENE 一段具有 20 世纪 90 年代中期日本电视剧格调的无对白浪漫蒙太奇。第一部分:一名安静的上班族注意到一位女
独居在海边废弃工厂的机器人和机械狗,瞥见飞行英雄后用手工花朵颜料重新装扮自己,但始终没有飞起来。四张参考图分别锁定角色与建筑内外景,情绪克制的写实科幻短片。
提示语
一部温柔的电影感短片讲述 Robot 和 Robot Dog 的故事:他们瞥见一位飞翔的英雄,从中获得灵感,使用手工制作的花卉颜料重新塑造自己。 @[Image 1] 定义 Robot 和 Robot Dog 完整的参考外观与比例;忽略其中的中性背景和重复的 turnaround 图,并让两者各自只保留一个。@[Image 2] 定义 Abandoned Facility 的室内建筑、玻璃窗、湿润混凝土和寒冷的海岸日光;保留其布局,并在地面和凹处密集堆放捡来的家具、生锈的置
中文长 prompt,把动漫版动作戏改写成真人实拍风格:少女借四块持续下坠的石板突围怪物围猎,30 秒一镜到底。每只怪物编号防止混淆,固定 20mm 镜头只靠跟随与环绕变化景别。
提示语
一部真人实拍奇幻动作电影从铃音在风暴虚空中的破碎神社桥上遭到围猎开始,她借四块持续下坠的石板高速突围,摄影机在真实时间内紧贴她的动量完成一镜到底。仅使用 @[character ref] 确定成片中铃音的人物身份、服装、右手护臂与武士刀;忽略参考图的背景、文字、排版、姿势和构图,不让它定义环境、动作、镜头或敌人。 C2、C5、C6、C8、C10是五个彼此独立的剃刀爪兽,每只都有六条肢体、钴蓝色躯壳、浅色腹部与修长前爪。C3、C4、C7、C9是四个彼此独立的剃刀翼兽,每只都身…