Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →漫画风格视频生成器 - 专门生成日式治愈系、国风水墨、美式卡通等漫画风格的动画视频。内置8种漫画风格模板，支持图生视频，一键生成高质量漫画动画。当用户需要生成漫画风格、动画风格、手绘风格的视频时使用此技能。
.claude/skills/freestylefly-manga-style-video/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-16 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -32% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 5% | 0% |
专门用于生成漫画/动画风格的视频，内置多种风格模板，无需复杂的提示词工程。
| 风格代码 | 名称 | 特点 | |---------|------|------| | japanese | 日式治愈系 | 吉卜力风格、手绘水彩、温馨治愈 | | ghibli | 吉卜力风格 | Studio Ghibli、宫崎骏动画风格 | | chinese | 国风水墨 | 中国传统水墨、淡雅诗意、工笔线条 | | cartoon | 美式卡通 | 迪士尼皮克斯、3D卡通、色彩鲜艳 | | sketch | 铅笔素描 | 手绘线条、黑白灰、艺术感 | | watercolor | 水彩手绘 | 透明质感、色彩晕染、艺术插画 | | manga_comic | 日式漫画 | 黑白网点、速度线、动态构图 | | chibi | Q版萌系 | 大头比例、可爱萌系、圆润线条 |
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生成日式治愈风格视频：
bashcd ~/.openclaw/workspace/skills/manga-style-video python3 scripts/manga_style_video.py "女孩在樱花树下读书"
bash# 国风水墨风格 python3 scripts/manga_style_video.py "山水意境" --style chinese # 美式卡通风格 python3 scripts/manga_style_video.py "可爱小动物" --style cartoon # 吉卜力风格 python3 scripts/manga_style_video.py "乡村风景" --style ghibli
bash# 使用角色图片作为参考 python3 scripts/manga_style_video.py "奶奶在包饺子" \ --style japanese \ --image ~/Desktop/character.png
bashpython3 scripts/manga_style_video.py "春节团圆场景" \ --style japanese \ --image character.png \ --duration 10 \ --ratio 9:16 \ --resolution 1080p \ --output ~/Desktop/my_video.mp4
| 参数 | 必需 | 默认值 | 说明 | |------|------|--------|------| | prompt | ✅ | - | 视频内容描述 | | --style | ❌ | japanese | 漫画风格 | | --image | ❌ | - | 参考图片路径 | | --duration | ❌ | 10 | 时长（秒） | | --ratio | ❌ | 9:16 | 比例（16:9/9:16/1:1/4:3） | | --resolution | ❌ | 1080p | 分辨率 | | --output | ❌ | - | 输出路径 | | --no-wait | ❌ | false | 不等待完成 |
bashpython3 scripts/manga_style_video.py \ "白发奶奶在厨房忙碌，窗外阳光洒进来" \ --style japanese \ --duration 8
bashpython3 scripts/manga_style_video.py \ "古代女子在荷塘边弹琴，荷花盛开" \ --style chinese \ --ratio 16:9
bash# 分镜1：角色登场 python3 scripts/manga_style_video.py \ "双马尾女孩站在校门口微笑" \ --style chibi \ --image girl.png \ --output scene1.mp4 # 分镜2：动作场景 python3 scripts/manga_style_video.py \ "女孩在教室认真读书" \ --style chibi \ --image girl.png \ --output scene2.mp4
bashpython3 scripts/manga_style_video.py \ "小猫咪在草地上打滚，超级可爱" \ --style chibi \ --duration 5
bashpython3 scripts/manga_style_video.py --list-styles
输出：
🎨 可用的漫画风格:
japanese - 日式治愈系
ghibli - 吉卜力风格
chinese - 国风水墨
cartoon - 美式卡通
sketch - 铅笔素描
watercolor - 水彩手绘
manga_comic - 日式漫画
chibi - Q版萌系日式动画风格，吉卜力工作室风格，手绘水彩质感，柔和粉彩色调，
线条简洁流畅，温馨治愈，细腻的背景描绘，宫崎骏风格中国传统水墨画风格，国风动画，淡雅色调，山水意境，
工笔线条，古风手绘，诗意唯美Q版大头比例，萌系可爱风格，圆润线条，明亮色彩，
卡通渲染，治愈系表情POST /api/v3/contents/generations/tasksdoubao-seedance-1-5-pro-251215（默认）1. 选择漫画风格 → 获取对应提示词模板
2. 组合用户描述 + 风格提示词
3. 调用 Seedance API 生成视频
4. 等待生成完成 → 下载视频默认保存到 ~/Desktop/，文件名格式：
manga_{风格}_{时间戳}.mp4--style 参数控制--duration 控制时长| 普通提示词 | 漫画风格提示词（本技能自动生成） | |-----------|------------------------------| | "女孩在樱花树下" | "女孩在樱花树下，日式动画风格，吉卜力工作室风格，手绘水彩质感，柔和粉彩色调，线条简洁流畅，温馨治愈" |
bash# 创建脚本 for i in 1 2 3; do python3 scripts/manga_style_video.py \ "分镜$i的场景描述" \ --style japanese \ --image character.png \ --output scene_$i.mp4 & done wait
如果想要特定的混合风格，可以直接使用 seedance-video-generation 技能，手动编写提示词。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-16 | fail→pass | 9,978 | 2,011 | -80% | 1 | 1 | 0% | 1,605 | 2,337 | +46% | 0 | 0 | — |
case-22 | pass→pass | 10,907 | 6,868 | -37% | 1 | 1 | 0% | 2,029 | 3,309 | +63% | 0 | 0 | — |
case-01 | fail→pass | 23,667 | 2,451 | -90% | 1 | 1 | 0% | 3,692 | 2,502 | -32% | 0 | 0 | — |
case-02 | fail→pass | 9,663 | 2,686 | -72% | 1 | 1 | 0% | 1,588 | 2,469 | +55% | 0 | 0 | — |
case-03 | fail→fail | 10,340 | 4,545 | -56% | 1 | 1 | 0% | 1,474 | 2,693 | +83% | 0 | 0 | — |
case-04 | fail→fail | 16,083 | 3,358 | -79% | 1 | 1 | 0% | 2,438 | 2,659 | +9% | 0 | 0 | — |
case-05 | fail→pass | 12,851 | 3,779 | -71% | 1 | 1 | 0% | 2,031 | 2,704 | +33% | 0 | 0 | — |
case-06 | fail→pass | 15,790 | 3,617 | -77% | 1 | 1 | 0% | 2,492 | 2,619 | +5% | 0 | 0 | — |
case-07 | fail→pass | 14,822 | 9,940 | -33% | 1 | 1 | 0% | 2,305 | 3,518 | +53% | 0 | 0 | — |
case-08 | fail→pass | 11,555 | 5,316 | -54% | 1 | 1 | 0% | 1,828 | 2,942 | +61% | 0 | 0 | — |
case-09 | fail→pass | 13,356 | 4,381 | -67% | 1 | 1 | 0% | 1,733 | 2,735 | +58% | 0 | 0 | — |
case-10 | fail→pass | 14,538 | 4,844 | -67% | 1 | 1 | 0% | 2,451 | 2,947 | +20% | 0 | 0 | — |
case-11 | fail→pass | 7,478 | 10,782 | +44% | 1 | 1 | 0% | 1,100 | 3,118 | +183% | 0 | 0 | — |
case-12 | fail→pass | 11,944 | 2,742 | -77% | 1 | 1 | 0% | 1,699 | 2,541 | +50% | 0 | 0 | — |
case-13 | fail→pass | 10,008 | 2,155 | -78% | 1 | 1 | 0% | 1,566 | 2,400 | +53% | 0 | 0 | — |
case-14 | fail→pass | 9,153 | 2,149 | -77% | 1 | 1 | 0% | 1,311 | 2,393 | +83% | 0 | 0 | — |
case-15 | fail→pass | 13,367 | 1,904 | -86% | 1 | 1 | 0% | 2,141 | 2,335 | +9% | 0 | 0 | — |
case-17 | pass→pass | 5,946 | 5,209 | -12% | 1 | 1 | 0% | 1,023 | 2,920 | +185% | 0 | 0 | — |
case-18 | pass→pass | 14,901 | 7,212 | -52% | 1 | 1 | 0% | 2,293 | 3,338 | +46% | 0 | 0 | — |
case-19 | pass→pass | 15,779 | 3,188 | -80% | 1 | 1 | 0% | 2,191 | 2,531 | +16% | 0 | 0 | — |
case-20 | pass→pass | 8,790 | 8,233 | -6% | 1 | 1 | 0% | 1,496 | 3,598 | +141% | 0 | 0 | — |
case-21 | pass→pass | 13,547 | 15,265 | +13% | 1 | 1 | 0% | 2,536 | 5,043 | +99% | 0 | 0 | — |
case-23 | pass→pass | 21,277 | 20,550 | -3% | 1 | 1 | 0% | 3,513 | 5,561 | +58% | 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. 23 cases were attempted. The headline lift of +61 percentage points is the difference between those two pass rates over the 23 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.