让 AI 围绕人的欲望主题，按"起承转合"结构、8 秒一个小爆点，生成一组 JSON 格式、每 30 帧一条、最多 900 帧的互相关联的视频帧描述，适合 Stable Diffusion Deforum 视频创作。

中文版提示词

请提供一系列互相关联的视频帧描述，每条描述只聚焦一个主体，格式为 JSON、每 30 帧一条、最多 900 帧，每个帧索引为键、对应值为描述性字幕，描述需契合主题且视觉吸引人。示例格式：{"0":"tiny cute bunny, vibrant diffraction, highly detailed, ultra hd...","30":"anthropomorphic clean cat, surrounded by fractals..."}。视频脚本主题需围绕并解决人的欲望（衣、食、住、行、金钱、尊重、认可、快乐、自信、幸福、自由等），脚本结构为"起（建置）+承（推进）+转+合（升华）"、8 秒一个小爆点，把最有悬念的精彩内容前置、可结合热点，开头要有冲突、打破常规、有爽感有期待，人物需有成长。主题：[主题]

英文版提示词

Provide a series of interconnected video-frame descriptions, each focusing on one main subject, in JSON format with a description every 30 frames up to 900 frames, each frame index as a key with a descriptive caption that fits the theme and is visually engaging. Example: {"0":"tiny cute bunny, vibrant diffraction, highly detailed, ultra hd...","30":"anthropomorphic clean cat, surrounded by fractals..."}. The video theme should revolve around resolving human desires (clothing, food, shelter, mobility, money, respect, recognition, happiness, confidence, freedom, etc.), with a structure of setup + progression + turn + resolution, a small climax every 8 seconds, the most suspenseful highlights front-loaded (optionally tied to hot topics), an opening with conflict that breaks convention and creates anticipation, and characters that grow. Theme: [theme]