Raindance AI 潮流提示词

主要提示词：黄金时刻

把你的照片作为首帧或参考图，再粘贴以下提示词。保持 9:16 画幅和 7 秒时长。

Cinematic music-video shot of the same person from the photo, sitting at the very edge of an old wooden pier above a calm open sea at golden hour. Warm low sun behind them, soft rim light on hair and shoulders, gentle sea breeze moving their clothes. The video opens on the still photo, then a hard cut, no fade, to the pier. Slow push-in, 35mm film look, shallow depth of field. They sing slowly and softly straight into the lens. No other people in frame. Vertical 9:16, 7 seconds.

另外两种场景

Cinematic music-video shot of the same person from the photo, sitting at the very edge of an old wooden pier in light summer rain. Overcast silver sky, wet glossy boards, fine raindrops catching the light, calm grey sea. The video opens on the still photo, then a hard cut, no fade, to the pier. Slow push-in, 35mm film look. They sing slowly into the lens while rain falls softly. No other people in frame. Vertical 9:16, 7 seconds.

Cinematic music-video shot of the same person from the photo, at an outdoor evening reception by the water, warm string lights glowing overhead, soft bokeh, deep blue dusk sky with a last band of pink at the horizon. The video opens on the still photo, then a hard cut, no fade. Slow handheld push-in, 35mm film look, warm tones. They sing slowly into the lens. Vertical 9:16, 7 seconds.

双人提示词

情侣或朋友可以把两张照片都作为参考图，并说明每个人的位置。双人视频最适合暖灯宴会或黄金时刻场景。

Cinematic music-video shot of the two people from the reference photos sitting side by side at the edge of an old wooden pier at golden hour. Person one sits on the left, person two on the right. They take turns singing slowly to the camera, then lean in and smile at each other. Slow push-in, 35mm film look, warm low sun. No other people. Vertical 9:16, 7 seconds.

提示词中每个部分为什么重要

• “照片中的同一个人”可以减少模型凭空生成新面孔的情况。
• “直接切换，不要淡入淡出”是这个潮流的标志；没有这句，大多数模型会做交叉淡化。
• “坐在老木码头的最边缘”能确定大家熟悉的构图。
• “缓慢推近”和“35mm 胶片质感”会让画面更像音乐视频，而不是手机随手拍。
• “对着镜头缓慢轻唱”让嘴部动作更轻柔，减少与歌曲节奏不一致的观感。
• “画面中没有其他人”可避免背景人物出现并抢走注意力。

常见错误

• 使用合照或远景照片。面部容易变化，生成的人可能不像你。
• 保留默认 16:9 画幅。TikTok 可能裁掉你的脸。
• 要求嘴型逐字同步歌词。目前模型难以准确对齐，效果反而可能更差。
• 在生成器里添加歌曲。请在 TikTok 内添加，使用平台提供的音乐授权。

常见问题