Rumpelstiltskin AI 提示词：分配舞者与旁观者

用途与限制

这是新画面生成，无法保证原片的精确步伐、剪辑和相貌。配乐取决于生成框所选流程。

先分配人物

单人模式中照片 1 是所选角色；双人模式中照片 1 跳舞，照片 2 观看。分开的照片比合照更容易分配角色。

让剪辑承载笑点

先交替展示舞者脸部与鞋子的特写，再给旁观者反应。描述原地小幅踮脚转动，不要把它写成大幅度舞蹈。

分享前检查

检查远景与特写的人物、角色顺序、鞋子和结尾动作。Prompt 引导生成新画面，不能保证原片的精确时序。

英文提示词示例

以下是可编辑的英文示例，不是已验证的生成结果。按实际照片与角色调整。

Create a new 15-second Rumpelstiltskin barn dance meme. Exactly two adult characters. The tiny tuxedo dancer uses @Image1. The seated spectator uses @Image2. Keep the two portrait identities distinct throughout. Uploaded dancer portraits determine face and hair; the tuxedo, tiny scale and footwork come from the scene direction. Scene references: @Image3 shows the barn; @Image4 shows shoe framing, not another person. STAGING: cramped warm straw-filled barn; tall golden hay bales fill the rear/right, dark vertical timber is visible left. Dancer stands close in front of the seated spectator. In the medium over-the-shoulder view, dancer is left/centre facing toward spectator at camera-right; only the spectator's blurred back-of-head/shoulder occupies the near-right foreground. Dancer is roughly knee-high relative to the seated adult. No cathedral-sized barn, dramatic sun shafts, new windows or wide establishing view. Camera is mostly locked, eye-level medium shots and abrupt close-ups, vintage live-action fantasy film texture. EDIT AND ACTIONS (hard cuts; retain this order and short repeated inserts): 0.00-1.25s medium over-the-shoulder two-person composition. Dancer faces spectator with mischievous grin, knees bent, elbows lifted away from the body. Raise both forearms outward into a broad wing-like stance; make two little buoyant tiptoe bounces. Spectator stays seated, back/shoulder blurred at right, watching. 1.25-2.85s tight dancer chest/head shot against hay, arms stretched sideways at about shoulder height. Lift eyebrows, tilt head slightly, flash a knowing smirk, small rhythmic shoulder bobs. He looks at spectator just off camera, not straight into lens. Do not cut to spectator here. 2.85-4.35s low tight FEET ONLY insert, framed from lower trouser hems to floor. Both black pointed heeled shoes remain close together on straw. Shift weight rapidly from left forefoot to right, lift one heel while the other toes bear weight, swivel toes inward then outward, briefly cross one foot in front and return. Small precise shuffling steps nearly in place; no large travel, running, jumps or generic tap-dance kicks. Long jacket hems sway above ankles. 4.35-5.35s return to the SAME dancer face close-up, both arms still out. Slight chin dip and raised-eyebrow smirk. 5.35-6.85s return to the SAME ankle/shoe framing. Repeat quick alternating toe-supported weight shifts and a tiny crossing step; straw shifts slightly under the soles. 6.85-7.85s SAME dancer head/shoulder close-up, eyes flick toward spectator, grin grows while shoulders gently pulse. 7.85-9.35s SAME feet insert. Rapid heel lifts, narrow inward/outward pivots and alternating crossed/uncrossed feet, staying on the same small patch of straw. 9.35-10.85s SAME dancer face close-up, arms held out, playful intent stare and slight head tilt. No spectator shot yet. 10.85-12.80s SAME feet insert, repeat the tight quick cross-step and forefoot pivots, jacket tails bouncing. Do not show a full-body dance montage. 12.80-13.85s the ONLY spectator reaction close-up. Use @Image2 faithfully, front three-quarter chest-up against hay. Eyes widen and mouth parts in surprised disbelief; bring ONE hand over the mouth and hold the startled look. No broad laughing grin or earlier repeated reactions. 13.85-14.35s return to opening medium over-the-shoulder framing. Dancer straightens slightly and raises ONE index finger beside his face, looking toward seated spectator. 14.35-15.00s tight dancer face with the raised index finger still beside his cheek, a sly confident smile. End on the DANCER, not the spectator. CONTINUITY: exactly two distinct identities, consistent costume, scale, eyelines and hay background. Keep shoe inserts and face inserts visually matched to the same dancer. No extra actors, identity swaps, merged faces, dialogue, subtitles, logos or slow motion. Generate the picture only. Soundtrack is controlled separately by the selected workflow; no lip-sync performance is needed.

生成前与结果检查

• 检查主体、手指、文字和画面边缘。
• 确认素材编号与 Prompt 中的 @Image 一致。
• 提交前核对所选模型、时长、画幅与报价。
• 播放成片并检查首尾帧，不能只看缩略图。

常见问题与调整

资料来源与相关政策

这些是我们编写的素材准备建议与提示词示例。模型能力可能变化，请在生成前检查当前设置。输入要求与使用规则可参阅：

• Kie AI 模型文档
• 内容政策
• 退款政策