Skip to content

anbeime/skill / skills/agentkit-multimedia-shopping/agentkit-multimedia-shopping

agentkit-multimedia-shopping

Verified

电商运营与内容创作者在需要制作带货短视频时,用此技能一键生成9:16竖屏数字人成片。自动编排AI绘画、语音合成与视频生成,快速产出“小省导购员”专属形象与专业配音,让多模态视频制作省时省力。

assets/examples/sample-output.md

View on GitHub
# 示例输出

## 概览
本文档提供agentkit-multimedia-shopping Skill的示例输出。

---

## 示例1:生成完整带货视频

### 输入参数
- **角色固定特征**:小省导购员(严格遵循character-profile.md)
- **场景类型**:商务会议室
- **情绪基调**:热情专业
- **话术内容**:"宝子们,今天给你们推荐一款超值商品,性价比超高,闭眼冲不亏!"

### 执行步骤

#### 1. 生成角色形象
```bash
python scripts/generate_character.py \
  --prompt "25岁左右女性数字人,鹅蛋脸、杏眼带笑、浅棕色齐肩卷发,肤色白皙,唇色淡粉,身穿浅灰色修身西装套裙,内搭白色衬衫,脚踩米色细跟鞋,佩戴银色简约项链,气质专业又亲和,手部姿态优雅自然。现代商务场景,背景简约,前景清晰。9:16竖屏,高清晰度,商业化带货风格。" \
  --aspect-ratio "9:16" \
  --resolution "1080x1920" \
  --output "./output/character_reference.jpg"
```

**输出**:`./output/character_reference.jpg`

#### 2. 生成场景背景
```bash
python scripts/generate_scene.py \
  --prompt "现代商务会议室,背景模糊大屏投影,前景简约文件,暖调柔光,侧光照明,明暗对比明显,商业化带货风格,9:16竖屏,高清晰度。" \
  --aspect-ratio "9:16" \
  --resolution "1080x1920" \
  --output "./output/scene_reference.jpg"
```

**输出**:`./output/scene_reference.jpg`

#### 3. 合成导购员语音
```bash
python scripts/generate_voice.py \
  --text "宝子们,今天给你们推荐一款超值商品,性价比超高,闭眼冲不亏!" \
  --voice-type "friendly" \
  --emotion "enthusiastic" \
  --format "16khz_mono_wav" \
  --output "./output/voice.wav"
```

**输出**:`./output/voice.wav`

#### 4. 生成背景音乐
```bash
python scripts/generate_music.py \
  --style "piano" \
  --emotion "enthusiastic" \
  --duration 5 \
  --format "16khz_mono_wav" \
  --output "./output/music.wav"
```

**输出**:`./output/music.wav`

#### 5. 生成视频
```bash
python scripts/generate_video.py \
  --character-image "./output/character_reference.jpg" \
  --scene-image "./output/scene_reference.jpg" \
  --voice-file "./output/voice.wav" \
  --music-file "./output/music.wav" \
  --prompt "主体:25岁女性数字人,鹅蛋脸、杏眼带笑、浅棕色齐肩卷发,浅灰修身西装套裙+白衬衫+米色细跟鞋,银色简约项链;动作时序(5s):0-1s双手轻握平板置于腰侧,1-3s缓慢抬手将平板举至胸前(速度均匀、姿态挺拔),3-5s保持举平板姿势,头部微抬看向镜头方向;场景:现代商务会议室,背景模糊大屏投影,前景简约文件;光影:暖调柔光,侧光勾勒身形,明暗对比明显;氛围:热情专业,面部表情带笑(无夸张),身体无多余晃动。音频类型:导购员语音(语速适中、语气热情)+ 轻快背景音乐(钢琴旋律);音频对齐:抬手动作匹配音乐节奏,语音节奏与唇形同步,无肢体动作与音频断层。技术约束:9:16竖版构图,中景固定镜头,背景虚化程度≥70%,动作强度0.8,面部一致性启用。" \
  --duration 5 \
  --aspect-ratio "9:16" \
  --output "./output/video.mp4"
```

**输出**:`./output/video.mp4`

### 最终输出
- **角色参考图**:`./output/character_reference.jpg`(9:16竖屏,1080×1920)
- **场景参考图**:`./output/scene_reference.jpg`(9:16竖屏,1080×1920)
- **语音文件**:`./output/voice.wav`(16kHz单声道wav)
- **音乐文件**:`./output/music.wav`(16kHz单声道wav)
- **视频文件**:`./output/video.mp4`(9:16竖屏,5秒)

---

## 示例2:仅生成角色参考图

### 输入参数
- **角色固定特征**:小省导购员(严格遵循character-profile.md)
- **情绪基调**:热情专业

### 执行步骤
```bash
python scripts/generate_character.py \
  --prompt "25岁左右女性数字人,鹅蛋脸、杏眼带笑、浅棕色齐肩卷发,肤色白皙,唇色淡粉,身穿浅灰色修身西装套裙,内搭白色衬衫,脚踩米色细跟鞋,佩戴银色简约项链,气质专业又亲和,手部姿态优雅自然。现代商务场景,背景简约,前景清晰。9:16竖屏,高清晰度,商业化带货风格。" \
  --aspect-ratio "9:16" \
  --resolution "1080x1920" \
  --output "./output/character_reference.jpg"
```

### 输出
- **角色参考图**:`./output/character_reference.jpg`(9:16竖屏,1080×1920)

---

## 示例3:仅生成语音

### 输入参数
- **话术内容**:"宝子们,今天给你们推荐一款超值商品,性价比超高,闭眼冲不亏!"
- **语音类型**:亲切
- **情绪基调**:热情

### 执行步骤
```bash
python scripts/generate_voice.py \
  --text "宝子们,今天给你们推荐一款超值商品,性价比超高,闭眼冲不亏!" \
  --voice-type "friendly" \
  --emotion "enthusiastic" \
  --format "16khz_mono_wav" \
  --output "./output/voice.wav"
```

### 输出
- **语音文件**:`./output/voice.wav`(16kHz单声道wav)

---

## 注意事项

1. **提示词质量**:提示词质量直接影响生成效果,建议严格遵循模板
2. **参数格式**:严格按照参数格式要求输入
3. **文件路径**:确保所有文件路径正确且可访问
4. **分辨率匹配**:图像必须为9:16竖屏,分辨率≥1080×1920
5. **音频格式**:音频必须为16kHz单声道wav格式
6. **工作流顺序**:严格按照图像→音频→视频的顺序生成

---

## 与InfiniteTalk的协同

本Skill与`infinitetalk-shopping-avatar` Skill协同工作:

1. **本Skill**:生成角色参考图、场景参考图、语音文件、音乐文件
2. **infinitetalk-shopping-avatar Skill**:生成InfiniteTalk专用提示词
3. **InfiniteTalk**:使用多模态内容和提示词生成视频

协同流程示例:
```
# 步骤1:本Skill生成角色参考图
python scripts/generate_character.py --prompt "..." --output "./output/character_reference.jpg"

# 步骤2:本Skill生成场景参考图
python scripts/generate_scene.py --prompt "..." --output "./output/scene_reference.jpg"

# 步骤3:本Skill合成语音
python scripts/generate_voice.py --text "..." --output "./output/voice.wav"

# 步骤4:本Skill生成音乐
python scripts/generate_music.py --style "piano" --output "./output/music.wav"

# 步骤5:使用infinitetalk-shopping-avatar Skill生成提示词
# (调用infinitetalk-shopping-avatar Skill的提示词生成功能)

# 步骤6:本Skill生成视频(结合InfiniteTalk提示词)
python scripts/generate_video.py \
  --character-image "./output/character_reference.jpg" \
  --scene-image "./output/scene_reference.jpg" \
  --voice-file "./output/voice.wav" \
  --music-file "./output/music.wav" \
  --prompt "InfiniteTalk专用提示词..." \
  --output "./output/video.mp4"
```
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall