# 视频拆解报告：chenxi_ai_daihuo.mp4
**结论**：混合制作：实拍 75% + AI 生成视频 25%（其中 100% 时长实拍还是 AI 证据不足，待确认）；叠加层：花字、文字标注、图形贴片、字幕、标题（代码/剪辑软件合成）；用到真实素材：实拍视频素材（43.1s）；复刻需准备参考：商品参考图（4 镜）；声音：人声 + BGM 114.8BPM（现成曲库配乐） + 音效 31 处；剪辑包装：字幕 300 句、花字 4 处、文字标注 56 处、29 个音效卡在文字/贴纸出现上
- 时长 57.667s · 1080×1920 · 30.0fps · 9:16
- 镜头 21 个，平均 2.75s；转场：cut×20、start×1
- 判断置信度（按时长加权）：70%；判定方式：确定性取证 + VLM 融合
- VLM 已生效（gemini-3.1-pro-preview,gemini-3.1-pro-preview-pc,gcp-gemini-3-pro-preview @ daydream88.fun）；由 claude 看图复核 21/21 个镜头，已并入结论
## 制作方式占比
| 底层画面 | 时长 | 占比 |
|---|---|---|
| 实拍 | 43.13s | 75% |
| AI 生成视频 | 14.53s | 25% |
## 复刻方案
**AI 视频 6 镜 → 6 个生成单元、6 次 Seedance 2.5 调用；实拍 15 镜 → 2 个场景、4 个机位**
- 默认引擎 Seedance 2.0（含 Mini / Fast） 不支持一次多镜头，用它要逐镜生成 6 次；这里按 Seedance 2.5 规划（单次 ≤30s、可一次出多镜头），共 6 次
- 分组规则：同场景、硬切相连的 AI 镜头合并为一次生成（S01）；单次 ≤30s；单元内连续硬切尽量 ≤8s、最多 10s（S06）；在叠化/淡入淡出和换场景处切开（S03）
- 声音：模型可以同时生成动作音效和环境声；口播建议单独 TTS 再混进去，方便改文案和卡时间
### 生成单元
| 单元 | 镜头 | 时间 | 时长 | 方式 | 衔接 |
|---|---|---|---|---|---|
| G1 | #2 | 1.9–4.7s | 2.8s | 单镜 | 首帧：参照原片第 2 镜首帧（shots/shot_002_first.jpg）的构图出图 |
| G2 | #3 | 4.7–6.6s | 1.9s | 单镜 | 首帧：参照原片第 3 镜首帧（shots/shot_003_first.jpg）的构图出图 |
| G3 | #4 | 6.6–7.9s | 1.3s | 单镜 | 首帧：参照原片第 4 镜首帧（shots/shot_004_first.jpg）的构图出图 |
| G4 | #19 | 49.1–51.5s | 2.4s | 单镜 | 首帧：参照原片第 19 镜首帧（shots/shot_019_first.jpg）的构图出图 |
| G5 | #20 | 51.5–54.9s | 3.4s | 单镜 | 首帧：参照原片第 20 镜首帧（shots/shot_020_first.jpg）的构图出图 |
| G6 | #21 | 54.9–57.7s | 2.8s | 单镜 | 首帧：参照原片第 21 镜首帧（shots/shot_021_first.jpg）的构图出图 |
**G1 · AI 生成视频 · 镜头 #2 · 2.8s**
- 参考素材：`@图片1` 人物参考图（每个角色一张正脸 + 半身），保证跨镜头长相一致；`@视频1` 原片 1.9–4.7s 片段：参考动作、运镜和剪辑节奏（只作参考，不复制画面）
- 首帧图像提示词：竖屏 9:16，温馨桌面场景：长发女生穿米白毛衣坐在原木桌前，双手捧着红色苹果造型的陶瓷杯（杯盖是苹果、杯身米色印小树和苹果图案），身后放着印苹果图案的米色礼盒，桌上铺红白格餐布和木勺；柔和自然光，背景浅色墙面虚化，写实带货风格
- 生成 4s 后剪到 2.8s（太短的镜头单独生成不稳定）
```text
竖屏 9:16；正面柔光（平光），柔光（几乎没有明暗分界），偏暖（约 3000–4000K），背景清晰。人物以 @图片1 为准。参考 @视频1 的动作、运镜和剪辑节奏。
女生微笑着拿起桌上的苹果造型陶瓷杯，双手捧到胸前向镜头展示，轻轻转动杯身让苹果图案朝向镜头；镜头从中景缓慢推近到杯子特写，结束时杯子占满画面中心
```
English prompt
```text
A young woman in a cream sweater picks up an apple-shaped ceramic mug from a wooden table, holds it with both hands toward the camera and slowly turns it to show the apple print; the camera pushes in from a medium shot to a close-up of the mug, ending with the mug centered and filling the frame
```
English prompt
```text
A young woman in a cream sweater picks up an apple-shaped ceramic mug from a wooden table, holds it with both hands toward the camera and slowly turns it to show the apple print; the camera pushes in from a medium shot to a close-up of the mug, ending with the mug centered and filling the frame
```
English prompt
```text
A young woman wearing black-framed glasses is talking to the camera. She first raises her right index finger to emphasize a point, then naturally clasps her hands together in front of her chest. Smooth movements, fixed camera.
```
English prompt
```text
A young woman in a cream sweater picks up an apple-shaped ceramic mug from a wooden table, holds it with both hands toward the camera and slowly turns it to show the apple print; the camera pushes in from a medium shot to a close-up of the mug, ending with the mug centered and filling the frame
```
English prompt
```text
A young woman in a cream sweater picks up an apple-shaped ceramic mug from a wooden table, holds it with both hands toward the camera and slowly turns it to show the apple print; the camera pushes in from a medium shot to a close-up of the mug, ending with the mug centered and filling the frame
```
English prompt
```text
A young woman in a cream sweater picks up an apple-shaped ceramic mug from a wooden table, holds it with both hands toward the camera and slowly turns it to show the apple print; the camera pushes in from a medium shot to a close-up of the mug, ending with the mug centered and filling the frame
```