Skip to content
cccskills
FreePublished on GitHub

agentkit-multimedia-shopping

Generate character images, backgrounds, synthetic speech, and music for product videos featuring 小省导购员, a digital shopping presenter. Coordinate these assets with InfiniteTalk to produce a 25-second vertical video made from five scenes.

OriginalShow the original description

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。

See how to install

Good for

  • Producing a 25-second vertical product presentation video
  • Generating reference images for the 小省导购员 character
  • Creating business setting backgrounds for a video
  • Generating narration and background music for product videos

Files included (16)

  • SKILL.md7.3 KB
  • assets/examples/sample-output.md6.8 KB
  • assets/templates/avatar-description.md1.6 KB
  • assets/templates/product-recommendation-script.md1.6 KB
  • references/character-profile.md5.1 KB
  • references/scene-templates.md5.5 KB
  • references/script-templates.md4.1 KB
  • references/tool-usage-guide.md5.3 KB
  • references/visual-styles.md5.0 KB
  • scripts/generate_audio.py9.3 KB
  • scripts/generate_character.py3.0 KB
  • scripts/generate_image.py6.3 KB
  • scripts/generate_music.py3.5 KB
  • scripts/generate_scene.py2.9 KB
  • scripts/generate_video.py4.5 KB
  • scripts/generate_voice.py3.5 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

AgentKit多媒体小省导购员数字人带货

任务目标

  • 本技能用于:基于ByteDance agentkit-samples多媒体用例,生成小省导购员数字人带货视频的多模态内容
  • 能力包含:
    • AI绘画生成(小省导购员角色形象、场景背景)
    • 语音合成(导购员语音、背景音乐)
    • 视频生成(多模态组合)
    • 工作流编排(图像→音频→视频→成片)
  • 触发条件:用户需要生成小省导购员数字人带货视频,使用agentkit-samples的多媒体能力

前置准备

依赖说明

本技能依赖以下Python包:

Pillow>=10.0.0
requests>=2.28.0
numpy>=1.24.0

环境准备

  1. 安装agentkit-samples(如需要)
  2. 准备API凭证(如涉及第三方服务调用)
  3. 准备小省导购员角色固定特征描述

前置知识

  • 了解ByteDance agentkit-samples多媒体用例的基本功能
  • 了解AI绘画、语音合成、视频生成的基本原理
  • 了解9:16竖屏视频规格

操作步骤

标准流程

1. 角色形象生成

使用AI绘画生成小省导购员角色形象:

输入信息:

  • 角色固定特征(脸型、发型、服饰、气质)
  • 场景类型(商务场景)
  • 情绪基调(热情、专业、亲切)

生成流程:

  1. 读取 references/character-profile.md,获取小省导购员角色固定特征
  2. 调用 scripts/generate_character.py 生成角色形象
  3. 裁剪为9:16比例,分辨率≥1080×1920
  4. 保存参考图(供InfiniteTalk使用)

输出:角色参考图(9:16竖屏)

2. 场景背景生成

使用AI绘画生成场景背景:

输入信息:

  • 场景类型(商务会议室、书房、洽谈室、大厅、办公室)
  • 情绪基调(冷调、暖调、中性)
  • 光影描述(侧光、顺光、顶光、逆光)

生成流程:

  1. 读取场景模板,选择对应场景类型
  2. 调用 scripts/generate_scene.py 生成场景背景
  3. 裁剪为9:16比例,分辨率≥1080×1920
  4. 保存场景参考图

输出:场景参考图(9:16竖屏)

3. 语音合成

使用TTS生成导购员语音:

输入信息:

  • 话术内容(带货文案)
  • 语音类型(语速、语气)
  • 情绪基调(热情、专业、亲切)

生成流程:

  1. 准备话术内容(符合小省导购员人设)
  2. 调用 scripts/generate_voice.py 合成语音
  3. 导出为16kHz单声道wav格式
  4. 保存语音文件(供InfiniteTalk使用)

输出:语音文件(16kHz单声道wav)

4. 背景音乐生成

使用音乐生成工具生成背景音乐:

输入信息:

  • 情绪基调(热情、专业、紧迫、亲切)
  • 音乐风格(管弦乐、钢琴、弦乐)
  • 时长(5秒/幕)

生成流程:

  1. 根据情绪基调选择音乐风格
  2. 调用 scripts/generate_music.py 生成音乐
  3. 导出为16kHz单声道wav格式
  4. 保存音乐文件(供InfiniteTalk使用)

输出:音乐文件(16kHz单声道wav)

5. 视频生成

使用多模态组合生成视频:

输入信息:

  • 角色参考图(9:16竖屏)
  • 场景参考图(9:16竖屏)
  • 语音文件(16kHz单声道wav)
  • 音乐文件(16kHz单声道wav)
  • 提示词(InfiniteTalk专用)

生成流程:

  1. 读取InfiniteTalk专用提示词(使用infinitetalk-shopping-avatar Skill生成)
  2. 调用 scripts/generate_video.py 生成视频
  3. 生成5幕视频(每幕5秒)
  4. 拼接完整视频(总时长25秒)

输出:完整视频(25秒,9:16竖屏)

可选分支

  • 当仅需生成角色形象:执行步骤1
  • 当仅需生成场景背景:执行步骤2
  • 当仅需生成语音:执行步骤3
  • 当仅需生成背景音乐:执行步骤4
  • 当仅需生成视频:执行步骤1-5

资源索引

注意事项

  • 角色一致性:所有生成的角色形象必须严格遵循角色固定特征
  • 分辨率匹配:所有图像必须为9:16竖屏,分辨率≥1080×1920
  • 音频格式:所有音频必须为16kHz单声道wav格式
  • 情绪适配:语音和音乐必须与场景情绪匹配
  • 工作流顺序:严格按照图像→音频→视频的顺序生成
  • 与InfiniteTalk协同:使用infinitetalk-shopping-avatar Skill生成提示词

使用示例

示例1:生成完整带货视频

功能说明:生成小省导购员数字人带货视频的完整流程

执行方式:

  1. 生成角色形象(调用generate_character.py)
  2. 生成场景背景(调用generate_scene.py)
  3. 合成导购员语音(调用generate_voice.py)
  4. 生成背景音乐(调用generate_music.py)
  5. 生成视频(调用generate_video.py,结合InfiniteTalk提示词)

关键参数:

  • 角色固定特征:严格遵循character-profile.md
  • 场景类型:商务场景
  • 情绪基调:热情专业
  • 分辨率:1080×1920(9:16竖屏)
  • 音频格式:16kHz单声道wav

输出:完整带货视频(25秒,9:16竖屏)

示例2:仅生成角色参考图

功能说明:仅生成小省导购员角色参考图

执行方式:调用generate_character.py

关键参数:

  • 角色固定特征:严格遵循character-profile.md
  • 情绪基调:热情专业
  • 分辨率:1080×1920(9:16竖屏)

输出:角色参考图(9:16竖屏)

工作流总结

角色固定特征描述
    ↓
AI绘画生成(角色形象+场景背景)
    ↓
语音合成(导购员语音+背景音乐)
    ↓
InfiniteTalk提示词生成(使用infinitetalk-shopping-avatar Skill)
    ↓
多模态视频生成(图像+音频+提示词)
    ↓
完整带货视频(25秒,9:16竖屏)

与InfiniteTalk的协同

本技能与infinitetalk-shopping-avatar Skill协同工作:

  1. 本技能:生成多模态内容(图像、音频)
  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:

本技能生成角色参考图 → infinitetalk-shopping-avatar生成提示词 → InfiniteTalk生成视频
本技能生成语音文件 → chinese-wav2vec2-base编码 → InfiniteTalk使用
本技能生成音乐文件 → chinese-wav2vec2-base编码 → InfiniteTalk使用

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

agent-team

FreeSummary

Organize AI agents into execution, management, and review roles for complex projects. Choose preset or custom teams, share project context and decisions, and combine their discussions and work into meeting records, feasibility reports, or implementation outputs.

  • Reviewing a technical architecture from several perspectives
  • Discussing product pricing across market and cost considerations
  • Assessing a new product's feasibility
anbeime/skill8.3KUpdated Oct 17, 2026

Convert PDF, Word, and Excel documents into structured Markdown for search and knowledge bases. The skill tries three parsers in sequence when needed and returns metadata, a confidence score, and the parser used.

  • Prepare documents for retrieval-augmented generation
  • Structure documents for an internal knowledge base
  • Extract structured text from contracts
anbeime/skill8.3KUpdated Oct 17, 2026

Organize parsed documents into four color-coded cards covering facts, explanations, risks, and actions. Preserve source references and generation details so readers can review evidence, identify uncertainty, and consider next steps in research, financial, or legal analysis.

  • Separating facts from interpretations in financial reports
  • Organizing evidence and risks in legal documents
  • Turning research documents into evidence-linked analysis cards
anbeime/skill8.3KUpdated Oct 17, 2026

Record operations across multiple AI agents and connect events into a traceable history. Search records by tracking ID, time window, or responsible component to support audits and investigations, with local storage available when the vector database fails.

  • Trace task dispatch and execution across agents
  • Search audit records by time or responsible component
  • Keep recording events during vector database failures
anbeime/skill8.3KUpdated Oct 17, 2026

Screen documents and external URLs before they enter a processing workflow. Check blocked domains and document permissions, then produce a pass or reject decision, a risk report, and a tracking ID for later review.

  • Screening documents before adding them to a knowledge base
  • Checking external URLs against a domain blocklist
  • Reviewing document permissions before processing
anbeime/skill8.3KUpdated Oct 17, 2026

archify

FreeSummary

Turn plain-language descriptions or Mermaid diagrams into interactive, standalone HTML diagrams of systems, workflows, API exchanges, data flows, and state changes. Validate layouts, explore relationships, switch themes, and export diagrams as images or video.

  • Explain system components and their relationships
  • Visualize workflows and approval steps
  • Map API calls and data pipelines
anbeime/skill8.3KUpdated Oct 17, 2026

See all skills by anbeime

Report a problem with this skill