Install any skill in seconds. Free to start, no credit card required.
Get Started Free →漫剧生成器 - 基于 Seedance 的漫画风格短剧生成工具。支持以主角图片为基础,自动生成漫剧分镜脚本并生成视频。适用于创作漫画风格的短视频、角色故事、动画短片等。当用户想要生成漫画风格的视频短剧、角色故事或漫剧时使用此技能。
.claude/skills/freestylefly-manga-drama/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 158% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 86% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 82% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 60% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 35% | 0% |
基于 Seedance 视频生成能力,专门用于创作漫画风格的短剧(漫剧)。
需要设置 ARK_API_KEY 环境变量。
bashcp .canghe-skills/.env.example .canghe-skills/.env
.canghe-skills/.env 文件,填写你的 API Key:ARK_API_KEY=your-actual-api-key-herebashexport ARK_API_KEY="your-api-key"
process.env).canghe-skills/.env~/.canghe-skills/.env需要依赖技能:seedance-video-generation
提供主角图片和主题,自动生成完整漫剧:
bashcd ~/.openclaw/workspace/skills/manga-drama python3 scripts/manga_drama.py generate \ --image /path/to/character.png \ --theme "校园日常" \ --scenes 3 \ --send-feishu
先创建脚本,再生成视频:
bash# 创建脚本模板 python3 scripts/manga_drama.py create-script \ --output my_drama.json \ --title "我的漫剧" \ --character "双马尾女孩" \ --num-scenes 4 # 编辑脚本文件后生成 python3 scripts/manga_drama.py from-script \ --script my_drama.json \ --image /path/to/character.png \ --send-feishu
内置5种漫剧分镜类型:
| 分镜类型 | 名称 | 说明 | |---------|------|------| | introduction | 主角登场 | 介绍主角,展示角色特征 | | action | 动作场景 | 主角进行某个动作 | | emotion | 情感表达 | 表达某种情感 | | interaction | 互动场景 | 与环境或其他元素互动 | | ending | 结尾定格 | 漫剧结尾,定格画面 |
json{ "title": "漫剧标题", "character": "主角描述", "style": "漫画风格", "total_scenes": 3, "scenes": [ { "scene_number": 1, "type": "introduction", "name": "主角登场", "prompt": "双马尾女孩站在画面中央,微笑看向镜头,漫画风格...", "duration": 5, "ratio": "9:16", "resolution": "1080p" } ] }
| 参数 | 必需 | 说明 | |------|------|------| | --image | ✅ | 主角图片路径 | | --theme | ✅ | 漫剧主题/剧情描述 | | --scenes | ❌ | 分镜数量(默认3) | | --output | ❌ | 输出目录(默认~/Desktop) | | --send-feishu | ❌ | 发送到飞书 |
| 参数 | 必需 | 说明 | |------|------|------| | --script | ✅ | 脚本文件路径 | | --image | ✅ | 主角图片路径 | | --send-feishu | ❌ | 发送到飞书 |
| 参数 | 必需 | 说明 | |------|------|------| | --output | ✅ | 输出脚本文件路径 | | --title | ❌ | 漫剧标题 | | --character | ❌ | 主角描述 | | --num-scenes | ❌ | 分镜数量 |
bashpython3 scripts/manga_drama.py generate \ --image ~/Desktop/girl_character.png \ --theme "校园日常" \ --scenes 3 \ --send-feishu
生成3个分镜:
bash# 创建脚本 python3 scripts/manga_drama.py create-script \ --output spring_festival.json \ --title "春节团圆" \ --character "白发奶奶" \ --num-scenes 5 # 编辑 spring_festival.json 文件 # 然后生成 python3 scripts/manga_drama.py from-script \ --script spring_festival.json \ --image ~/Desktop/grandma.png \ --send-feishu
1. 分析主角图片 → 提取角色特征
2. 根据主题 → 生成分镜脚本
3. 每个分镜 → 调用 Seedance 图生视频
4. 可选 → 发送到飞书~/Desktop/drama_我的漫剧/
├── drama_script_xxx.json # 脚本文件
├── scene_1_introduction.mp4 # 分镜1
├── scene_2_action.mp4 # 分镜2
├── scene_3_emotion.mp4 # 分镜3
└── ...编辑生成的脚本文件,修改 prompt 字段:
json{ "prompt": "{character}在樱花树下读书,花瓣飘落,漫画风格,温馨治愈..." }
创建多个脚本,批量生成:
bashfor script in scripts/*.json; do python3 scripts/manga_drama.py from-script \ --script "$script" \ --image ~/Desktop/character.png done
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | pass→pass | 6,294 | 2,050 | -67% | 1 | 1 | 0% | 898 | 2,057 | +129% | 0 | 0 | — |
case-01 | fail→pass | 5,486 | 7,092 | +29% | 1 | 1 | 0% | 869 | 2,238 | +158% | 0 | 0 | — |
case-02 | fail→fail | 6,823 | 15,944 | +134% | 1 | 1 | 0% | 1,008 | 4,667 | +363% | 0 | 0 | — |
case-07 | fail→pass | 8,252 | 2,547 | -69% | 1 | 1 | 0% | 1,189 | 2,209 | +86% | 0 | 0 | — |
case-03 | fail→pass | 24,668 | 6,959 | -72% | 1 | 1 | 0% | 1,616 | 2,940 | +82% | 0 | 0 | — |
case-04 | pass→pass | 13,507 | 9,615 | -29% | 1 | 1 | 0% | 2,308 | 3,558 | +54% | 0 | 0 | — |
case-05 | pass→pass | 20,199 | 17,315 | -14% | 1 | 1 | 0% | 3,338 | 5,129 | +54% | 0 | 0 | — |
case-06 | pass→pass | 8,154 | 4,576 | -44% | 1 | 1 | 0% | 1,529 | 2,622 | +71% | 0 | 0 | — |
case-08 | fail→pass | 8,362 | 2,011 | -76% | 1 | 1 | 0% | 1,323 | 2,112 | +60% | 0 | 0 | — |
case-10 | fail→pass | 10,516 | 2,568 | -76% | 1 | 1 | 0% | 1,625 | 2,190 | +35% | 0 | 0 | — |
case-11 | pass→pass | 5,238 | 2,423 | -54% | 1 | 1 | 0% | 812 | 2,168 | +167% | 0 | 0 | — |
case-12 | pass→pass | 7,325 | 1,725 | -76% | 1 | 1 | 0% | 1,059 | 1,996 | +88% | 0 | 0 | — |
case-13 | pass→pass | 8,663 | 3,586 | -59% | 1 | 1 | 0% | 1,202 | 2,425 | +102% | 0 | 0 | — |
case-14 | fail→pass | 7,658 | 2,217 | -71% | 1 | 1 | 0% | 1,139 | 2,130 | +87% | 0 | 0 | — |
case-15 | pass→pass | 8,130 | 2,024 | -75% | 1 | 1 | 0% | 1,162 | 2,120 | +82% | 0 | 0 | — |
case-16 | pass→pass | 6,383 | 1,967 | -69% | 1 | 1 | 0% | 787 | 2,075 | +164% | 0 | 0 | — |
case-17 | pass→pass | 8,390 | 2,523 | -70% | 1 | 1 | 0% | 1,261 | 2,169 | +72% | 0 | 0 | — |
case-18 | pass→pass | 12,050 | 1,701 | -86% | 1 | 1 | 0% | 1,655 | 2,063 | +25% | 0 | 0 | — |
case-19 | fail→pass | 6,979 | 4,576 | -34% | 1 | 1 | 0% | 1,025 | 2,549 | +149% | 0 | 0 | — |
case-20 | fail→pass | 7,796 | 2,463 | -68% | 1 | 1 | 0% | 1,088 | 2,120 | +95% | 0 | 0 | — |
case-21 | fail→pass | 10,126 | 1,749 | -83% | 1 | 1 | 0% | 1,376 | 2,037 | +48% | 0 | 0 | — |
case-22 | fail→pass | 8,717 | 1,728 | -80% | 1 | 1 | 0% | 1,401 | 2,029 | +45% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +45 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.