Digital human video generation via Qingyun API — avatar-based talking head videos
Scanned 9/5/2026
Install to Claude Code
npx -y skills add aAAaqwq/AGI-Super-Skills --skill digital-human-api --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Digital Human Api?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/aaaaqwq-digital-human-api-agi-super-team)More formats (shields.io, HTML) on the badges page.
---
name: digital-human-api
description: Digital human video generation via Qingyun API — avatar-based talking head videos
---
# digital-human-api v3
基于青云API的通用数字人口播视频生成 Skill。
**v3核心改进:每shot生成专属场景图(Daniel真脸 + 个性化场景),视频自然不抽象。**
## 触发条件
- `数字人视频`、`口播视频`、`digital human`
- 基于剧本生成分镜头数字人视频
## v3 新流程(4步/shot)
```
剧本JSON → [Scene Image] → [TTS] → [Kling Video] → [Lip Sync] → FFmpeg合并
↑ 新增:每个shot独立生成贴合场景的图片
```
**每shot独立流程:**
1. 🖼️ **场景图生成** — Grok依据参考脸生成贴合场景的图片(保持Daniel的脸)
2. 📝 **TTS语音** — Gemini生成口播音频
3. 🎬 **Kling视频** — 场景图 + 动作提示词 → 动态视频
4. 👄 **对口型** — Kling LipSync音画同步
5. 🔗 **FFmpeg合并** — 所有shot + BGM → 最终视频
## v3 剧本格式
```json
{
"title": "视频标题",
"avatar_image": "/path/to/daniel-headshot.jpg",
"shots": [
{
"id": 1,
"text": "口播文案",
"emotion": "sarcastic",
"scene_description": "(可选)场景图详细描述",
"duration": 5
}
]
}
```
### emotion 可选值
| emotion | 动作风格 |
|---------|---------|
| `serious` | 严肃直视镜头 |
| `friendly` | 友好微笑 |
| `excited` | 兴奋手势多 |
| `sarcastic` | 讽刺挑眉 |
| `storytelling` | 讲故事手势 |
| `humorous` | 幽默轻松 |
| `intense` | 紧张/激动 |
| `confident` | 自信权威 |
| `questioning` | 疑惑歪头 |
| `casual` | 日常对话 |
### scene_description 写法
描述越具体,场景图越贴合。建议格式:
- 人物表情+动作(如:raised eyebrow, holding coffee cup)
- 场景(如:modern cafe, restaurant table)
- 光线(如:warm natural lighting)
- 风格(如:realistic photo, shot on iPhone)
## 使用方式
```bash
export QINGYUN_API_KEY=$(pass show api/qingyun)
# 完整流水线
python3 scripts/generate.py --script script.json --concurrent 1
# 分步执行
python3 scripts/generate.py --script script.json --step image # 场景图
python3 scripts/generate.py --script script.json --step tts # TTS语音
python3 scripts/generate.py --script script.json --step video # Kling视频
python3 scripts/generate.py --script script.json --step lipsync # 对口型
python3 scripts/generate.py --script script.json --step merge # 合并
```
## 输出结构
```
output_dir/
├── shot_01_scene.jpg # 场景原图
├── shot_01_scene_768.jpg # 适配Kling的尺寸
├── shot_01_audio.mp3 # TTS语音
├── shot_01_video.mp4 # Kling视频
├── shot_01_lipsync.mp4 # 对口型完成
├── ...
└── final.mp4 # 最终视频
```
## 已知限制
| 问题 | 解决 |
|------|------|
| 视频太抽象 | v3改用场景图,每个shot独立生成 |
| 429限流 | 并发=1,轮询间隔15s |
| 图片像素无效 | 自动resize到768px宽 |
| Grok场景图失败 | 自动降级到无ref生成 |
## 文件清单
```
digital-human-api/
├── SKILL.md # 本文件
├── scripts/
│ ├── generate.py # 主脚本 v3(~800行)
│ └── config.yaml # 配置 v3
```
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!