图像转换

Name: 图像转换
Author: yikart

AI 智能社区

作者仓库星标 17,824
作者更新于实时读取
作者仓库 AiToEarn

流狐档案事实层信息：领域、Agent、信任分、运行环境、权限与网络

领域: AI 智能
兼容 Agent: Claude Code
Cursor
Cline
Codex
Windsurf
Gemini CLI

+20
信任分: 88 / 100 · 社区维护
作者 / 版本 / 许可: @yikart · 未声明 license
Token 消耗评级: 低消耗
接入复杂程度: 需简单配置
是否需要外部 API Key: 不需要
兼容的系统: 未声明（默认跨平台）
底层运行要求: 无特殊要求
文件与系统权限: 只读
允许写入 / 修改
网络行为: 仅限本地
安装命令数: 26 条

档案由构建时根据 SKILL.md 与安装命令自动衍生，可能与作者实际意图存在差异。

需要注意： 未限定 allowed-tools，默认拥有全部工具权限。

输出预览 editing-videos.preview

---
name: editing-videos
description: Video editing using Volcengine Track structure. Supports cutting, trimming, adding text, sticker…
category: AI 智能
runtime: 无特殊运行时
---

# editing-videos 输出预览

## PART A: 任务判断
- 适用问题：提示词、Agent 工作流、模型评估或自动化推理。
- 输入要求：目标材料、限制条件、期望输出和验收方式。
- 证据边界：围绕“⚠️ CRITICAL: ALL PosX/PosY Are Top-Left Coordinates / When to Use / Workflow”读取原文规则，不把推断写成作者承诺。

## PART B: 执行结果
- **01** 任务判断：确认你的需求是否属于提示词、Agent 工作流、模型评估或自动化推理，并标出输入、限制和预期结果。
- **02** 执行计划：优先按“⚠️ CRITICAL: ALL PosX/PosY Are Top-Left Coordinates / When to Use / Workflow”拆成步骤，说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果：给出可复制的命令、文件改动、检查清单或内容草稿，并说明如何继续迭代。
- **04** 风险边界：结合 读取文件、写入/修改文件、主要在本地完成、通常不需要额外 API Key 给出执行前确认项。

## Running Rules
- 读取文件、写入/修改文件；主要在本地完成；通常不需要额外 API Key。
- 先小样例验证，再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。

SKILL.md SKILL.md

---
name: editing-videos
description: Video editing using Volcengine Track structure. Supports cutting, trimming, adding text, sticker…
category: AI 智能
source: yikart/AiToEarn
---

# editing-videos

## 什么时候使用
- 把AI / Agent方向的常用动作沉淀成 Agent 可调用的技能 适合处理AI Agent、提示词、模型评估与自动化推理，核心价值是把输入、判断、执行、验证和交付边界固定下来，避免 Agent 泛泛回答。 把任务拆成可执行、可检查…
- 面向提示词、Agent 工作流、模型评估或自动化推理，优先处理能明确输入、步骤和验收标准的工作。

## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。

## 执行规则
- 围绕「⚠️ CRITICAL: ALL PosX/PosY Are Top-Left Coordinates / When to Use / Workflow」组织步骤，不把推断写成作者事实。
- 读取文件、写入/修改文件；主要在本地完成；通常不需要额外 API Key。
- 先跑小样例，确认结果可检查后再扩大任务范围。

## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown，不编造命令、平台或依赖。

证据边界与执行链路

作者原文负责流程事实；仓库文件负责来源和命令；流狐只补充适用场景、限制和质量判断。

skill "editing-videos" {
  输入层     -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
  上下文层   -> ⚠️ CRITICAL: ALL PosX/PosY Are Top-Left Coordinates / When to Use / Workflow
  规则层     -> SKILL.md 触发条件 / 执行顺序 / 输出格式
  运行层     -> 无特殊运行时 | 读取文件、写入/修改文件 | 主要在本地完成
  安全层     -> 通常不需要额外 API Key + 小任务验证 + diff / 日志复核
  输出层     -> 可复制结果 + 检查清单 + 下一步迭代
}

解读按原文结构重写，命令、链接、术语均保留；右侧可核对作者原始 SKILL.md

先判断是否适合

核心用途：把AI / Agent方向的常用动作沉淀成 Agent 可调用的技能适合处理AI Agent、提示词、模型评估与自动化推理，核心价值是把输入、判断、执行、验证和交付边界固定下来，避免 Agent 泛泛回答。把任务拆成可执行、可检查、可继续迭代的步骤；通…
适用任务：更适合 AI Agent、提示词、模型评估与自动化推理，尤其是输入、步骤、验收口径可以反复复用的场景。
不宜硬套：如果任务只有一次性问答、来源材料很少，或需要作者没有说明的命令和外部服务，应先降级为普通对话。

作者设计意图

触发依据：使用时机、上下文收集、输出格式和失败处理共同决定它是否适合当前任务。
来源边界：作者原文负责事实，仓库文件负责命令和文件证据；Fluxly 只补充用途、限制和使用判断。

作者的方法与取舍

执行主线：任务围绕「⚠️ CRITICAL: ALL PosX/PosY Are Top-Left Coordinates / When to Use / Workflow」组织，把触发条件、必要上下文、执行规则和交付标准连成一条可复核的链路。
设计重点：重点不在重复教程，而在解释作者为什么规定这些步骤、哪些条件不可省略，以及结果偏离时应回到哪一层检查。
使用判断：先确认任务与原文边界一致，再决定是否启用；来源没有说明的命令、平台和依赖保持未知，不由 Fluxly 补写。

边界和复核

依赖条件：通常不需要额外 API Key，适合先用小任务验证触发是否稳定。
权限提醒：会涉及 read / write 权限，执行前要说明可读写范围、命令边界和回退方式。
质量标准：结果应包含明确产物、必要证据和下一步动作；如果只是泛泛解释，需要补充输入和验收条件后重跑。

Video Editing with Track Structure

Performs video editing using the Volcengine cloud editing API with native Track structure.

⚠️ CRITICAL: ALL PosX/PosY Are Top-Left Coordinates

ALL PosX/PosY values in this API are TOP-LEFT corner coordinates, NOT center point!

This applies to ALL filters: transform, crop, delogo, etc.

For full-screen video display, you MUST use:

{ "Type": "transform", "PosX": 0, "PosY": 0, "Width": <canvas_width>, "Height": <canvas_height> }

NEVER use canvas center coordinates (like 640, 360) for PosX/PosY!

Using center coordinates will cause video to display off-canvas
The coordinate (0, 0) is the top-left corner of the canvas

When to Use

Use this skill when:

Cutting or trimming video segments
Adding text, watermarks, or stickers to videos
Adding or replacing audio tracks
Applying filters or visual effects
Merging multiple video clips
Adding transitions between clips
Changing video playback speed
Removing watermarks (logo blur)

Workflow

Step 1: Get Video Information (Required)

ALWAYS call uploadAndGetVid first to:

Upload video to Volcengine cloud
Get VID for video source reference (format: vid://xxx)
Get Duration for time calculations (in seconds)
Get Dimensions (width, height) for Canvas setup

If you only need metadata without editing (e.g., for display purposes), use probeVideoMetadata instead.

Step 2: Build EditParam

Construct your EditParam following the structure below.

Step 3: Submit Task

Call submitDirectEditTask with Canvas, Output (optional), and Track.

Step 4: Poll for Results

First check: Wait 90 seconds after submission
Subsequent checks: Wait 30 seconds between polls
Call getVideoEditTaskStatus to check progress
Maximum wait time: 20 minutes

Step 5: Return Results

On completion, return the output video URL to user.

EditParam Structure

{
  "Canvas": { "Width": 1920, "Height": 1080 },
  "Output": { "Fps": 30, "Codec": { "VideoCodec": "h264" } },
  "Track": [
    [{ "Type": "video", "Source": "vid://xxx", "TargetTime": [0, 5000] }],
    [{ "Type": "text", "Text": "Hello", "TargetTime": [0, 5000], "Extra": [] }]
  ]
}

Canvas (optional):

If omitted, auto-detected from the primary video source dimensions
If provided manually, MUST match getVideoInfo/uploadAndGetVid output
Width = horizontal pixels, Height = vertical pixels
Example: 1280x720 video → { Width: 1280, Height: 720 } — NOT { Width: 720, Height: 1280 }


**Time unit**: All time values are in **MILLISECONDS** (1 second = 1000 ms)

## Track Structure

Track is a **2D array** `Track[layerIndex][elementIndex]`:

- Outer array: Rendering layers (higher index = higher layer, renders on top)
- Inner array: Elements on the same track

## Element Types

### video

```json
{
  "Type": "video",
  "Source": "vid://your_vid",
  "TargetTime": [0, 5000],
  "Extra": [
    { "Type": "trim", "StartTime": 10000, "EndTime": 30000 }
  ]
}

audio

{
  "Type": "audio",
  "Source": "vid://your_audio_vid",
  "TargetTime": [0, 10000],
  "Extra": [
    { "Type": "a_volume", "Volume": 0.5 }
  ]
}

When to use audio elements:

Adding background music to video
Adding voiceover/narration
Adjusting audio volume separately from video
Replacing original audio with different audio

When NOT to use audio elements:

Simple video concatenation - video elements already contain their audio tracks
When you want to keep the original video audio unchanged

image

{
  "Type": "image",
  "Source": "mid://your_image_mid",
  "TargetTime": [0, 3000],
  "Extra": [
    { "Type": "transform", "PosX": 100, "PosY": 100, "Width": 200, "Height": 200 }
  ]
}

text

{
  "Type": "text",
  "Text": "Your Text Here",
  "TargetTime": [0, 5000],
  "FontType": "SY_Black",
  "FontSize": 120,
  "FontColor": "#FFFFFFFF",
  "AlignType": 1,
  "Extra": [
    { "Type": "transform", "PosX": 60, "PosY": 440, "Width": 1800, "Height": 200 }
  ]
}

subtitle

{
  "Type": "subtitle",
  "Text": "http://example.com/subtitle.srt",
  "TargetTime": [0, 20000],
  "FontType": "ALi_PuHui",
  "FontSize": 60,
  "Extra": [
    { "Type": "transform", "PosX": 40, "PosY": 550, "Width": 1200, "Height": 150 }
  ]
}

Filter Types (Extra)

transform - 2D Transform

Controls element position, size, rotation, and opacity on the canvas.

IMPORTANT:

PosX and PosY are the TOP-LEFT corner coordinates of the element relative to the canvas top-left corner
They are NOT center point coordinates
Rotation is performed around the element's center point

Parameters:

PosX: X coordinate of element's top-left corner (pixels)
PosY: Y coordinate of element's top-left corner (pixels)
Width: Element width on canvas (pixels)
Height: Element height on canvas (pixels)
Rotation: Rotation angle [-360, 360], clockwise is positive (optional)
Alpha: Opacity [0,1], 0 is transparent (optional)

Basic Example (fill canvas):

{ "Type": "transform", "PosX": 0, "PosY": 0, "Width": 1920, "Height": 1080 }

Rotation Example (rotate 1280x720 video 90° to fit 720x1280 canvas):

When rotating, the element rotates around its center. To fill a 720x1280 canvas with a 1280x720 video rotated 90°:

After rotation, the video's display size becomes 720x1280
The center should be at canvas center (360, 640)
Calculate: PosX = 360 - 1280/2 = -280, PosY = 640 - 720/2 = 280

{ "Type": "transform", "PosX": -280, "PosY": 280, "Width": 1280, "Height": 720, "Rotation": 90 }

Centering Formula:

To center an element horizontally: PosX = (CanvasWidth - ElementWidth) / 2
To center an element vertically: PosY = (CanvasHeight - ElementHeight) / 2

Example: Center a 600x60 subtitle on 720x1280 canvas:

PosX = (720 - 600) / 2 = 60
PosY = 1280 - 100 = 1180 (near bottom)

Common Mistakes:

WRONG: Using canvas center (360, 640) or (640, 360) as PosX/PosY - this places the element's top-left at center, causing it to go off-canvas
WRONG: Omitting transform for full-screen video elements - video may not display correctly without explicit positioning
WRONG: Creating separate audio tracks for each video clip in simple concatenation - video elements already contain their audio
CORRECT: Use (0, 0) for top-left alignment when video should fill the canvas
CORRECT: Only add audio tracks when you need background music, voiceover, or volume adjustment

trim - Time Trim

{ "Type": "trim", "StartTime": 5000, "EndTime": 15000 }

crop - Area Crop

{ "Type": "crop", "PosX": 100, "PosY": 100, "Width": 800, "Height": 600 }

speed - Playback Speed

{ "Type": "speed", "Speed": 2.0 }

transition - Transition Effect

{ "Type": "transition", "Source": "1182376", "Duration": 1000 }

lut_filter - Color Filter

{ "Type": "lut_filter", "TargetTime": [0, 5000], "Source": "1184003", "Intensity": 0.8 }

video_animation - Video Animation

{ "Type": "video_animation", "AnimRes": "1180337", "AnimStartTime": 0, "AnimEndTime": 1000 }

a_volume - Audio Volume

{ "Type": "a_volume", "Volume": 0.5 }

a_fade - Audio Fade

{ "Type": "a_fade", "FadeIn": 2000, "FadeOut": 2000 }

delogo - Logo Blur

{ "Type": "delogo", "TargetTime": [0, 10000], "PosX": 1700, "PosY": 50, "Width": 200, "Height": 80, "Sigma": 30, "Radius": 30 }

Resource IDs

Filter IDs (lut_filter)

ID	Name	Description
1184003	Clear	Enhanced clarity
1184004	Afternoon	Warm afternoon tone
1183995	Vintage	Retro film look
1183993	Friends	Friends TV show style

Transition IDs

ID	Name	Description
1182376	CircleOpen	Circle wipe open
1182360	RotateZoom	Rotate and zoom
1182370	DoorOpen	Door opening reveal
1182379	ClockWipe	Clock sweep wipe

Video Animation IDs

ID	Name	Type
1180337	FadeIn	In
1180382	FadeOut	Out
1180335	Shrink	In
1180338	ZoomIn	In

Font IDs

ID	Name
SY_Black	Source Han Sans Black
ALi_PuHui	Alibaba PuHuiTi
PM_ZhengDao	Pangmen Zhengdao Title

Examples

Example 1: Simple Video Concatenation (Most Common)

{
  "Canvas": { "Width": 1920, "Height": 1080 },
  "Track": [[
    {
      "Type": "video",
      "Source": "vid://video1",
      "TargetTime": [0, 5000],
      "Extra": [
        { "Type": "transform", "PosX": 0, "PosY": 0, "Width": 1920, "Height": 1080 }
      ]
    },
    {
      "Type": "video",
      "Source": "vid://video2",
      "TargetTime": [5000, 12000],
      "Extra": [
        { "Type": "transform", "PosX": 0, "PosY": 0, "Width": 1920, "Height": 1080 }
      ]
    },
    {
      "Type": "video",
      "Source": "vid://video3",
      "TargetTime": [12000, 20000],
      "Extra": [
        { "Type": "transform", "PosX": 0, "PosY": 0, "Width": 1920, "Height": 1080 }
      ]
    }
  ]]
}

Key points for video concatenation:

Transform with PosX: 0, PosY: 0 is REQUIRED for full-screen display
NO separate audio tracks - video elements already contain their audio
All video clips on the same track (single inner array)
Sequential TargetTime: video2 starts where video1 ends

Example 2: Concatenate Videos with Transition

{
  "Canvas": { "Width": 1920, "Height": 1080 },
  "Track": [[
    {
      "Type": "video",
      "Source": "vid://video1",
      "TargetTime": [0, 5000],
      "Extra": [
        { "Type": "transform", "PosX": 0, "PosY": 0, "Width": 1920, "Height": 1080 },
        { "Type": "transition", "Source": "1182376", "Duration": 1000 }
      ]
    },
    {
      "Type": "video",
      "Source": "vid://video2",
      "TargetTime": [4000, 9000],
      "Extra": [
        { "Type": "transform", "PosX": 0, "PosY": 0, "Width": 1920, "Height": 1080 }
      ]
    }
  ]]
}

Note: With transitions, TargetTime values overlap (video2 starts at 4000, video1 ends at 5000).

Example 3: Trim Video (10s to 30s)

{
  "Canvas": { "Width": 1920, "Height": 1080 },
  "Track": [[
    {
      "Type": "video",
      "Source": "vid://xxx",
      "TargetTime": [0, 20000],
      "Extra": [
        { "Type": "trim", "StartTime": 10000, "EndTime": 30000 }
      ]
    }
  ]]
}

Example 4: Add Text Watermark

{
  "Canvas": { "Width": 1920, "Height": 1080 },
  "Track": [
    [{ "Type": "video", "Source": "vid://xxx", "TargetTime": [0, 10000] }],
    [{
      "Type": "text",
      "Text": "@MyChannel",
      "TargetTime": [0, 10000],
      "FontType": "SY_Black",
      "FontSize": 60,
      "FontColor": "#FFFFFFFF",
      "Extra": [
        { "Type": "transform", "PosX": 1700, "PosY": 50, "Width": 200, "Height": 60 }
      ]
    }]
  ]
}

Example 5: Add Background Music

{
  "Canvas": { "Width": 1920, "Height": 1080 },
  "Track": [
    [{ "Type": "video", "Source": "vid://video", "TargetTime": [0, 30000] }],
    [{
      "Type": "audio",
      "Source": "vid://music",
      "TargetTime": [0, 30000],
      "Extra": [
        { "Type": "a_volume", "Volume": 0.3 }
      ]
    }]
  ]
}

Example 6: Apply Filter

{
  "Canvas": { "Width": 1920, "Height": 1080 },
  "Track": [[
    {
      "Type": "video",
      "Source": "vid://xxx",
      "TargetTime": [0, 10000],
      "Extra": [
        { "Type": "lut_filter", "TargetTime": [0, 10000], "Source": "1183993", "Intensity": 0.8 }
      ]
    }
  ]]
}

Example 7: Logo Blur (Watermark Removal)

{
  "Canvas": { "Width": 1920, "Height": 1080 },
  "Track": [[
    {
      "Type": "video",
      "Source": "vid://xxx",
      "TargetTime": [0, 10000],
      "Extra": [
        { "Type": "delogo", "TargetTime": [0, 10000], "PosX": 1700, "PosY": 50, "Width": 200, "Height": 80, "Sigma": 30, "Radius": 30 }
      ]
    }
  ]]
}

Example 8: Rotate Landscape Video to Portrait

Convert a 1280x720 (landscape) video to 720x1280 (portrait) canvas:

{
  "Canvas": { "Width": 720, "Height": 1280 },
  "Track": [[
    {
      "Type": "video",
      "Source": "vid://xxx",
      "TargetTime": [0, 10000],
      "Extra": [
        {
          "Type": "transform",
          "PosX": -280,
          "PosY": 280,
          "Width": 1280,
          "Height": 720,
          "Rotation": 90
        }
      ]
    }
  ]]
}

Calculation:

Original video: 1280x720, Canvas: 720x1280
After 90° rotation, video displays as 720x1280
Rotation center = (PosX + Width/2, PosY + Height/2) should equal canvas center (360, 640)
PosX = 360 - 1280/2 = -280
PosY = 640 - 720/2 = 280

Important Notes

Always call uploadAndGetVid first - Required for VID and dimensions
Time unit is milliseconds - 1 second = 1000 ms
Track layers render bottom to top - Track[0] is bottom, Track[1] is above it
Transition duration must < clip duration
Subtitle tracks must be separate - Each subtitle occupies one track, max 10 subtitle tracks
Maximum processing time: 20 minutes
Full-screen video requires transform with PosX: 0, PosY: 0 - Do NOT use canvas center coordinates like (640, 360) or (360, 640)
Video elements already contain audio - Do NOT create separate audio tracks for simple video concatenation; only use audio tracks for background music, voiceover, or volume adjustment
Canvas is optional — omit it for auto-detection from video sources; only specify manually for custom canvas sizes

安装命令

一键复制安装选择包管理器,点击右侧复制,粘贴到终端回车即可

npx skills add yikart/AiToEarn/project/aitoearn-backend/apps/aitoearn-ai/src/core/agent/skills/editing-videos -y

查看通用安装教程

查看 Skill 原文

兼容工具

Claude Code
Cursor
Cline
Codex
Windsurf
Gemini
OpenCode
Roo Code
AAide
+Augment
GitHub Copilot
OpenClaw
KKiro
Junie
WWarp
Goose
Trae
Kilo Code
OpenHands
Amp
Zencoder
CContinue
AIAider
CodeBuddy
CodeGeeX
Replit Agent

安全审计

88 / 100

社区

未声明 license/许可。
未声明 version。
未限定 allowed-tools，默认拥有全部工具权限。

元信息

来源: github
yikart/AiToEarn
English

@yikart 的其他技能

分类与标签

分类 AI 智能