seedance-video-creator

wwwzhouhui/skills_collection · updated Apr 8, 2026

MDX-style export adds YAML metadata + attribution linking explainx.ai and this canonical listing URL.

$npx skills add https://github.com/wwwzhouhui/skills_collection --skill seedance-video-creator
0 commentsdiscussion
summary

将创意想法转化为专业分镜提示词,通过即梦 API 先生成参考图,再基于图片+提示词生成视频、自动下载。

skill.md

Seedance 2.0 分镜视频创作工具

将创意想法转化为专业分镜提示词,通过即梦 API 先生成参考图,再基于图片+提示词生成视频、自动下载。


前置条件

使用本技能前,确保以下服务可用:

  1. jimeng-free-api-all 服务运行中(默认端口 8000)

    # Docker 部署
    docker run -it -d --init --name jimeng-free-api-all \
      -p 8000:8000 -e TZ=Asia/Shanghai \
      wwwzhouhui569/jimeng-free-api-all:latest
    
  2. 获取 SessionID

  3. 环境变量(可选,也可在交互中指定):

    • JIMENG_API_URL:API 基础地址(默认 http://127.0.0.1:8000),注意只填基础地址,不要包含路径部分。如使用 duckcloud 服务则为 https://jimeng.duckcloud.fun
    • JIMENG_SESSION_ID:即梦 SessionID

核心能力

Seedance 2.0 特性

重要约束:Seedance 2.0 必须至少提供一张参考图片,不支持纯文本生成视频。当用户没有提供图片时,需要先通过文生图 API 生成首帧参考图。

多模态输入(最多12个文件):

  • 图片:≤9张(首帧/尾帧/角色/场景参考)
  • 视频:≤3个,总时长≤15s(运镜/动作/特效参考)
  • 音频:≤3个,总时长≤15s(配乐/音效/对白参考)
  • 文本:自然语言描述

视频参数

  • 模型:seedance-2.0-fast(推荐,快速版)/ jimeng-video-seedance-2.0-fast(等价别名)/ jimeng-video-seedance-2.0(Pro版)/ seedance-2.0 / seedance-2.0-pro(Pro等价别名)
  • 比例:1:1 / 4:3 / 3:4 / 16:9 / 9:16 / 3:2 / 2:3 / 21:9
  • 分辨率:480p / 720p(默认)/ 1080p
  • 时长:4 - 15

交互方式:使用 @1@2(或 @图1@image1)指定素材用途


完整工作流(三阶段)

核心流程:分镜提示词 → 文生图生成首帧 → 图片+提示词生成视频

用户输入创意 → 第一阶段:生成分镜提示词
             → 第二阶段:调用文生图 API 生成首帧参考图 → 下载图片到本地
             → 第三阶段:用参考图 + 分镜提示词调用 Seedance 2.0 → 下载视频

第一阶段:分镜提示词生成

步骤一:理解用户想法

用友好的方式询问用户:

  1. 你想讲什么故事?(一句话概括核心内容)
  2. 视频时长?(4-15秒,默认4秒)
  3. 有参考素材吗?(图片/视频路径)
  4. 是否需要直接生成视频?(是→进入第二、三阶段 / 否→只输出提示词)

步骤二:深入挖掘细节

根据用户的初步回答,逐步引导完善以下维度:

1. 内容叙事

  • 故事的起承转合
  • 关键情节和角色
  • 对白/旁白内容

2. 视觉风格

  • 整体风格:写实/动画/水墨/科幻/复古/电影感
  • 色调氛围:明亮/昏暗/暖色/冷色/黑白
  • 画面比例:竖屏(9:16)/横屏(16:9)/标准(4:3)

3. 镜头语言

  • 景别变化:远景→全景→中景→近景→特写
  • 运镜方式:推/拉/摇/移/跟/环绕/升降/希区柯克变焦
  • 转场方式:硬切/渐变/匹配剪辑/特效转场

4. 动作与节奏

  • 主体动作:行走/奔跑/打斗/舞蹈/表情变化
  • 动作节奏:舒缓/急促/有张有弛

5. 声音设计

  • 配乐风格:激昂/温馨/悬疑/欢快/史诗
  • 音效需求:环境音/动作音效/特殊音效

步骤三:构建分镜结构

将故事拆解为时间轴上的镜头序列。

示例结构(4秒视频)

0-1秒:开场镜头,建立场景
1-3秒:发展,核心动作/情感
3-4秒:结尾,画面定格

步骤四:生成两种提示词

根据收集的信息,分别生成:

A. 首帧图片提示词(用于文生图)

描述视频第一帧的静态画面,作为 Seedance 2.0 的首帧参考图。

首帧提示词要求

  • 描述一个静止的画面瞬间(不含动作、运镜等动态描述)
  • 包含:场景、人物/主体外观、光影、色调、构图
  • 画面比例应与视频比例一致
  • 简洁精炼,100-200 字

示例

日落时分的海边沙滩,一位穿着白色飘逸长裙的年轻女孩站在海边,面朝大海,夕阳逆光照射,金色光晕笼罩全身,长发和裙摆被海风轻轻吹动,远处海平线上的落日将天空染成橘红色,脚边浅浅的海浪泡沫

B. 视频分镜提示词(用于 Seedance 视频生成)

完整的分镜提示词,用 @1 引用首帧参考图。

提示词结构模板

【整体描述】风格 + 时长 + 画面比例 + 整体氛围

@1 作为画面首帧参考

【分镜描述】
0-X秒:[镜头运动],[画面内容],[主体动作],[光影/特效]
X-Y秒:[镜头运动],[画面内容],[主体动作],[光影/特效]
...

【声音说明】配乐风格/音效/对白

提示词书写原则

  1. 时间轴清晰,标注每段的秒数范围
  2. 镜头语言明确(推/拉/摇/移/跟/环绕等)
  3. 动作描述具体,避免模糊词汇
  4. 必须使用 @1 引用首帧图片(首帧图片由第二阶段自动生成)
  5. 如果用户提供了多张图片,使用 @1@2 分别引用
  6. 总长度控制在 500 字以内

步骤五:优化与确认

  • 检查是否违反平台规则(写实人脸限制)
  • 确认用户满意后进入第二阶段

第二阶段:文生图生成首帧参考图

目的:Seedance 2.0 必须有参考图片,当用户没有提供图片时,先通过文生图 API 生成首帧画面。

如果用户已提供图片:跳过此阶段,直接进入第三阶段。

步骤一:调用文生图 API

curl -s --max-time 120 -X POST "${API_URL}/v1/images/generations" \
  -H "Authorization: Bearer ${SESSION_ID}" \
  -H "Content-Type: application/json" \
  -d "{
    \"model\": \"jimeng-4.5\",
    \"prompt\": \"${首帧图片提示词}\",
    \"ratio\": \"${RATIO}\",
    \"resolution\": \"2k\"
  }"

参数说明

参数 说明 推荐值
model 图像生成模型 jimeng-4.5(稳定高质量)
prompt 首帧画面描述 第一阶段生成的首帧提示词
ratio 画面比例 与视频比例一致(如 9:16
resolution 图片分辨率 2k

API 返回格式

{
  "created": 1700000000,
  "data": [
    { "url": "https://p6-heycan-hgt.byteimg.com/..." }
  ]
}

步骤二:下载首帧图片到本地

IMAGE_URL=$(echo "${RESPONSE}" | jq -r '.data[0].url')
IMAGE_FILE="/tmp/seedance_frame_$(date +%Y%m%d_%H%M%S).png"
curl -sL -o "${IMAGE_FILE}" "${IMAGE_URL}"

步骤三:向用户展示首帧图片

报告生成结果,让用户确认首帧画面满意后进入第三阶段。


第三阶段:Seedance 2.0 视频生成与下载

步骤一:确认生成参数

参数 说明 默认值
API 地址 jimeng-free-api-all 服务地址 http://127.0.0.1:8000
SessionID 即梦平台的 sessionid 环境变量或用户提供
模型 seedance-2.0-fast seedance-2.0-fast
比例 视频画面比例 与首帧图片一致
分辨率 视频分辨率 720p
时长 4-15 秒 4
参考图片 首帧图片路径(第二阶段生成或用户提供) 必需

步骤二:调用 Seedance 视频生成 API

使用本地图片(multipart/form-data)

curl -s --max-time 300 -X POST "${API_URL}/v1/videos/generations" \
  -H "Authorization: Bearer ${SESSION_ID}" \
  -F "model=seedance-2.0-fast" \
  -F "prompt=${VIDEO_PROMPT}" \
  -F "ratio=${RATIO}" \
  -F "resolution=${RESOLUTION}" \
  -F "duration=${DURATION}" \
  -F "files=@${IMAGE_FILE}"

多张图片时

curl -s --max-time 300 -X POST "${API_URL}/v1/videos/generations" \
  -H "Authorization: Bearer ${SESSION_ID}" \
  -F "model=seedance-2.0-fast" \
  -F "prompt=${VIDEO_PROMPT}" \
  -F "ratio=${RATIO}" \
  -F "resolution=${RESOLUTION}" \
  -F "duration=${DURATION}" \
  -F "files=@/path/to/image1.jpg" \
  -F "files=@/path/to/image2.jpg"

重要

  • Seedance 2.0 必须至少上传一张图片,否则返回错误 {"code":-2001,"message":"Seedance 2.0 需要至少一张图片"}
  • API 是同步阻塞的,会自动轮询直到视频生成完成(通常 60-120 秒)
  • curl 超时建议设置为 300 秒:--max-time 300
  • 提示词中的 @1@2 对应 files 参数中图片的上传顺序
  • Authorization 头需要 Bearer 前缀,格式为 Bearer your_sessionid

步骤三:解析结果

API 返回 OpenAI 兼容格式:

{
  "created": 1700000000,
  "data": [
    {
      "url": "https://v3-dreamnia.jimeng.com/...",
      "revised_prompt": "原始提示词"
    }
  ]
}

步骤四:下载视频

VIDEO_URL=$(echo "${RESPONSE}" | jq -r '.data[0].url')
OUTPUT_FILE="seedance_$(date +%Y%m%d_%H%M%S).mp4"
curl -L -o "${OUTPUT_FILE}" "${VIDEO_URL}"
echo "视频已下载到: $(pwd)/${OUTPUT_FILE}"

步骤五:展示结果

向用户报告:

  1. 视频生成成功/失败
  2. 视频文件路径和大小
  3. 使用的首帧图片和分镜提示词(方便用户微调重试)

错误处理

API 错误码

错误 含义 处理方式
code: -2001 Seedance 需要至少一张图片 先调文生图 API 生成首帧,再重试
HTTP 200 成功 提取 data[0].url
HTTP 401 SessionID 过期 提示用户重新获取
HTTP 400 参数错误 检查 model/ratio/duration
HTTP 429 积分不足 提示用户等待或充值
HTTP 500 服务器错误 重试或检查服务状态

常见问题

问题 解决方案
Seedance 报错需要图片 必须先通过文生图生成首帧,再用图片调 Seedance
连接拒绝 检查 jimeng-free-api-all 容器是否运行
生成超时 增加 curl --max-time,或降低时长
内容被过滤 修改提示词,避免敏感内容
视频 URL 失效 URL 有时效性,及时下载
积分不足 每日免费积分有限,次日刷新

完整示例对话

示例一:纯文本创作(无用户图片 → 自动生成首帧)

用户:帮我生成一个橘猫在海边捉螃蟹的视频,9:16竖屏,4秒

第一阶段 - 生成提示词

首帧图片提示词:

阳光明媚的海边沙滩,一只圆滚滚的橘色猫咪蹲在潮湿的沙滩上,好奇地盯着面前一只小螃蟹,金色阳光照射,海浪轻轻拍打岸边,色彩鲜明活泼,3D动画风格

视频分镜提示词:

可爱3D动画风格,4秒,9:16竖屏,阳光明媚的海边

@1 作为画面首帧参考

0-1秒:中景,画面中的橘猫好奇地盯着沙滩上的小螃蟹,海浪轻拍岸边
1-3秒:近景跟随,橘猫伸出爪子试探性地去抓螃蟹,螃蟹举起钳子反击,橘猫被吓得往后一跳,表情夸张可爱
3-4秒:中景微拉远,橘猫和螃蟹对峙,背景碧蓝大海和白色浪花

背景音效:欢快俏皮卡通配乐 + 海浪声

第二阶段 - 生成首帧参考图

# 1. 调用文生图 API 生成首帧
curl -s --max-time 120 -X POST "${API_URL}/v1/images/generations" \
  -H "Authorization: Bearer ${SESSION_ID}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "jimeng-4.5",
    "prompt": "阳光明媚的海边沙滩,一只圆滚滚的橘色猫咪蹲在潮湿的沙滩上,好奇地盯着面前一只小螃蟹,金色阳光照射,海浪轻轻拍打岸边,色彩鲜明活泼,3D动画风格",
    "ratio": "9:16",
    "resolution": "2k"
  }'

# 2. 下载图片
IMAGE_URL=$(echo "${RESPONSE}" | jq -r '.data[0].url')
curl -sL -o /tmp/seedance_frame.png "${IMAGE_URL}"

第三阶段 - 生成视频

curl -s --max-time 300 -X POST "${API_URL}/v1/videos/generations"
how to use seedance-video-creator

How to use seedance-video-creator on Cursor

AI-first code editor with Composer

1

Prerequisites

Before installing skills in Cursor, ensure your development environment meets these requirements:

  • Cursor installed and configured on your development machine
  • Node.js version 16.0+ with npm package manager (verify with node --version)
  • Active project directory or workspace where you want to add seedance-video-creator
2

Execute installation command

Execute the skills CLI command in your project's root directory to begin installation:

$npx skills add https://github.com/wwwzhouhui/skills_collection --skill seedance-video-creator

The skills CLI fetches seedance-video-creator from GitHub repository wwwzhouhui/skills_collection and configures it for Cursor.

3

Select Cursor when prompted

The CLI will show a list of available agents. Use arrow keys to navigate and space to select Cursor:

◆ Which agents do you want to install to?
│ ── Universal (.agents/skills) ── always included ────
│ • Amp
│ • Antigravity
│ • Cline
│ • Codex
│ ●Cursor(selected)
│ • Cursor
│ • Windsurf
4

Verify installation

Confirm successful installation by checking the skill directory location:

.cursor/skills/seedance-video-creator

Reload or restart Cursor to activate seedance-video-creator. Access the skill through slash commands (e.g., /seedance-video-creator) or your agent's skill management interface.

Security & Verification Notice

We perform automated surface-level scans (Gen AI Scanner, Socket, Snyk) during installation. These checks detect common vulnerabilities but do not guarantee complete security. Always review skill source code and verify the publisher's reputation before production use.

Skills execute code in your development environment. Always verify the publisher's identity, review recent commits, and test in isolated environments before production deployment.

List & Monetize Your Skill

Submit your Claude Code skill and start earning

GET_STARTED →

Use Cases

Task Automation & Efficiency

Automate repetitive workflows and reduce manual effort

Example

Generate reports, summarize documents, draft communications

Save 3-5 hours per week on routine tasks

Knowledge Enhancement

Learn new skills, understand complex topics, get expert guidance

Example

Explain concepts, provide examples, suggest learning resources

Accelerate learning and skill development by 2x

Quality Improvement

Enhance output quality through reviews, suggestions, and refinements

Example

Review drafts, suggest improvements, catch errors

Improve work quality by 30-40% with less effort

Implementation Guide

Prerequisites

  • Claude Desktop or compatible AI client with skill support
  • Clear understanding of task or problem to solve
  • Willingness to iterate and refine outputs

Time Estimate

15-45 minutes depending on use case complexity

Installation Steps

  1. 1.Install skill using provided installation command
  2. 2.Test with simple use case relevant to your work
  3. 3.Evaluate output quality and relevance
  4. 4.Iterate on prompts to improve results
  5. 5.Integrate into regular workflow if valuable

Common Pitfalls

  • Expecting perfect results without iteration
  • Not providing enough context in prompts
  • Using skill for tasks outside its intended scope
  • Accepting outputs without review and validation

Best Practices

✓ Do

  • +Start with clear, specific prompts
  • +Provide relevant context and constraints
  • +Review and refine all outputs before using
  • +Iterate to improve output quality
  • +Document successful prompt patterns

✗ Don't

  • Don't use without understanding skill limitations
  • Don't skip validation of outputs
  • Don't share sensitive information in prompts
  • Don't expect skill to replace human judgment

💡 Pro Tips

  • Be specific about desired format and style
  • Ask for multiple options to choose from
  • Request explanations to understand reasoning
  • Combine AI efficiency with human expertise

When to Use This

✓ Use When

Use when skill capabilities match your task, clear ROI on time saved, and you can validate outputs. Best for repetitive tasks, learning, and quality improvement.

✗ Avoid When

Avoid when task requires deep expertise you can't validate, involves sensitive decisions, or when learning process is more valuable than speed of completion.

Learning Path

  1. 1Familiarize yourself with skill capabilities and limitations
  2. 2Start with low-risk, non-critical tasks
  3. 3Progress to more complex and valuable use cases
  4. 4Build expertise through regular use and experimentation

Discussion

Product Hunt–style comments (not star reviews)
  • No comments yet — start the thread.
general reviews

Ratings

4.828 reviews
  • Kofi Chen· Dec 24, 2024

    Registry listing for seedance-video-creator matched our evaluation — installs cleanly and behaves as described in the markdown.

  • Fatima Sharma· Dec 20, 2024

    seedance-video-creator reduced setup friction for our internal harness; good balance of opinion and flexibility.

  • Michael Taylor· Nov 15, 2024

    Useful defaults in seedance-video-creator — fewer surprises than typical one-off scripts, and it plays nicely with `npx skills` flows.

  • Zara Farah· Nov 11, 2024

    I recommend seedance-video-creator for anyone iterating fast on agent tooling; clear intent and a small, reviewable surface area.

  • Nia Brown· Oct 6, 2024

    I recommend seedance-video-creator for anyone iterating fast on agent tooling; clear intent and a small, reviewable surface area.

  • Emma Bansal· Oct 2, 2024

    Useful defaults in seedance-video-creator — fewer surprises than typical one-off scripts, and it plays nicely with `npx skills` flows.

  • Nia Thompson· Sep 25, 2024

    Solid pick for teams standardizing on skills: seedance-video-creator is focused, and the summary matches what you get after install.

  • Hiroshi Park· Sep 21, 2024

    Keeps context tight: seedance-video-creator is the kind of skill you can hand to a new teammate without a long onboarding doc.

  • Benjamin Ramirez· Sep 21, 2024

    seedance-video-creator has been reliable in day-to-day use. Documentation quality is above average for community skills.

  • Sakshi Patil· Sep 1, 2024

    Registry listing for seedance-video-creator matched our evaluation — installs cleanly and behaves as described in the markdown.

showing 1-10 of 28

1 / 3