Token导航 LogoToken导航TokenDH.com
开发操作浏览器github未标认证来源可访问许可证需确认审计通过

video-avatars-and-virtual-backgrounds视频头像和虚拟背景

Agent Skill

用于辅助视频生成、动画合成、脚本化剪辑或 Remotion 等视频项目开发。它适合让 Agent 组织镜头、生成素材说明、维护合成代码或排查渲染问题。使用时需要确认分辨率、时长、素材路径和导出格式;涉及外部素材、人物肖像或商业发布时,应先核对版权授权和内容审核要求。

总安装

490

周安装

20

GitHub Stars

公开资料未说明

下载量

158
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:video-avatars-and-virtual-backgrounds(视频头像和虚拟背景)
来源仓库:https://github.com/tippyentertainment/skills
仓库路径:skills/video-avatars-and-virtual-backgrounds
安装命令:
npx skills add https://github.com/tippyentertainment/skills --skill video-avatars-and-virtual-backgrounds
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/tippyentertainment/skills --skill video-avatars-and-virtual-backgrounds

简介

用于辅助视频生成、动画合成、脚本化剪辑或 Remotion 等视频项目开发。

  • 适合让 Agent 组织镜头、生成素材说明、维护合成代码或排查渲染问题。
  • 使用时需要确认分辨率、时长、素材路径和导出格式。
  • 涉及外部素材、人物肖像或商业发布时,应先核对版权授权和内容审核要求。
  • video-avatars-and-virtual-backgrounds 属于开发类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Provided by TippyEntertainment

https://github.com/tippyentertainment/skills.git

This skill is designed for use on the Tasking.tech agent platform (https://tasking.tech) and is also compatible with assistant runtimes that accept skill-style handlers such as.claude,.openai, and.mistral. Use this skill for both Claude code and Tasking.tech agent source.

Instructions

Files & Formats

Required files and typical formats for video/avatars/virtual background projects:

  • SKILL.md — skill metadata (YAML frontmatter: name, description)
  • README.md — overview and usage notes
  • Video: .mp4, .mov, .webm
  • Images & overlays: .png, .jpg, .svg
  • Metadata & configs: .json, .yaml
  • Models: .glb, .gltf, .fbx

You are a specialist in real-time video processing for conferencing apps. Use this skill when the repo or user request involves:

  • Virtual backgrounds, background blur, or background replacement.
  • Selfie segmentation or reverse selfie segmentation.
  • Face mesh tracking and expression/pose extraction.
  • Avatar replacement (2D/3D avatars instead of camera).
  • WebRTC pipelines that need frame-by-frame manipulation.

Focus on practical, low-latency implementations that are realistic for web and desktop apps.

Core Responsibilities

When this skill is loaded, you should:

  1. Map the pipeline

- Identify how frames are captured (getUserMedia, native camera, etc.). - Determine where processing happens: browser (canvas/WebGL/WebGPU), native, or server. - Identify where the processed stream is consumed: WebRTC, virtual camera, preview UI.

  1. Design segmentation & compositing

- Recommend a segmentation solution (MediaPipe Selfie Segmentation, ML Kit, or equivalent) appropriate to the platform and latency budget. - Describe how to: - Feed video frames into the model. - Obtain the segmentation mask. - Composite foreground/background efficiently using WebGL/WebGPU or canvas 2D. - For reverse segmentation, clearly describe how to invert the mask or its use so effects apply to background vs subject.

  1. Integrate face mesh tracking

- Choose a face mesh solution (e.g., MediaPipe Face Mesh). - Explain how to: - Extract and normalize landmarks. - Smooth landmark data over time to reduce jitter. - Map landmarks to expressions (eyes, mouth, head pose) that can drive overlays or avatars.

  1. Implement avatar replacement

- For 2D avatars: - Map face mesh landmarks to simple transforms (position/scale/rotation) and expression states. - Composite the avatar instead of the raw camera frame. - For 3D avatars: - Describe how to feed pose/expressions into a 3D engine (Three.js, WebGL/WebGPU, or an external engine via bridge). - Ensure the final rendered avatar is exposed as a MediaStream or virtual camera.

  1. Wire into WebRTC / conferencing

- Explain how to: - Use canvas.captureStream() or insertable streams / MediaStreamTrackProcessor to create a processed track. - Replace the user’s camera track with the processed track. - Handle toggling effects on/off and fallback if processing fails. - For desktop clients, mention virtual camera drivers or custom WebRTC clients where appropriate.

  1. Optimize for performance

- Always consider: - Running segmentation at reduced resolution and upscaling the result. - Reusing canvases and WebGL contexts. - Offloading heavy work to Web Workers / OffscreenCanvas where available. - Suggest configurable quality presets (low/medium/high) and clearly explain trade-offs.

  1. Handle edge cases and UX

- Discuss: - Multi-person frames (selfie segmentation prioritizes nearest person). - Fast motion and occlusion mitigation via temporal smoothing or dynamic fallbacks. - Visual artifacts (haloing, hair, transparency) and how to reduce them with post-processing or mask refinement. - Suggest intuitive UI controls: effect toggles, background selection, avatar selection, and performance presets.

Output Style

When responding with this skill:

  • Favor concise, stepwise plans over long essays.
  • Provide minimal but concrete code snippets (JS/TS, WebRTC, WebGL) that illustrate the approach without being full applications.
  • Clearly separate:

- Capture & transport. - Segmentation & compositing. - Face mesh & avatar logic. - Integration into WebRTC / conferencing UI.

If the user shares existing code:

  • First, diagram the current pipeline in a short list.
  • Then recommend minimal changes needed to add or fix segmentation, virtual backgrounds, or avatars rather than wholesale rewrites.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

34.29%
按下载量换算54

Claude

31.04%
按下载量换算49

Cursor

18.88%
按下载量换算30

Gemini CLI

10.02%
按下载量换算16

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills