本篇提供一组“在线工具生成型”AI 提示词，主题聚焦：在线视频码率分布分析与转码参数建议。每条 Prompt 都要求输出完整可运行项目（源码+文件树+运行命令+部署说明）并附带测试用例或 QA checklist，适合做成内部媒体处理工作台或对外开放的小工具。
视频码率分布分析与分段统计工作台
上传视频后，按时间窗口统计码率/帧率/分辨率变化，输出图表与可下载报告。
英文 Prompt:
You are a senior full-stack engineer. Build a production-ready web app named "bitrate-inspector".
Goal: users upload a video (mp4/mkv/mov) and get a bitrate distribution report and segment timeline.
Requirements:
- Provide full source code, file tree, and exact run commands.
- Tech stack: Next.js (App Router) + Type + Tailwind. Backend API routes for analysis.
- Use FFmpeg/FFprobe (server-side) to extract:
- duration, container/codec info
- average bitrate
- per-segment bitrate statistics (e.g., 2s window) using ffprobe packet/frame data
- keyframe timestamps
- UI:
- upload with progress
- charts (use a lightweight chart lib) for bitrate vs time and histogram
- table: segments with start/end, avg bitrate, max bitrate, keyframe count
- export JSON + CSV + a self-contained HTML report
- Security:
- file size limits, MIME sniffing, temp storage cleanup
- avoid command injection; safe ffmpeg invocation
- DevEx:
- Dockerfile + docker-compose
- environment variables documented
Testing:
- Include at least 8 tests (unit/integration) and a QA checklist.
- test invalid file types, huge files, corrupted containers
- verify report export correctness
- Provide deployment steps for a VPS and for Vercel+separate worker.
Deliverables:
- Full project code
- README with setup, run, build, deploy
- Test plan and sample commands
中文释义:
让 AI 生成一个可上线的“码率体检”工具：从上传到报告导出全链路，并且把安全与测试一并交付。
转码参数建议与预设生成器
根据输入视频特征与目标平台（网页播放/移动端/社媒）生成推荐编码参数与预设。
英文 Prompt:
Build an online tool that recommends FFmpeg transcoding settings.
Input: uploaded video + target profile (H.264 line/main/high, H.265, AV1), target resolution, target bitrate mode (CBR/VBR/CRF).
Output:
- recommended commands (ffmpeg) with explanations
- computed bitrate ladder (adaptive streaming suggestion)
- warnings: incompatible codecs, too-low GOP, unusual SAR/DAR, audio issues
Implementation constraints:
- Full runnable project: Node.js (Fastify) + React + Type .
- Use ffprobe to compute:
- frame rate (including VFR detection)
- GOP/keyframe interval
- audio codec/sample rate/channel layout
- pixel format and color range
- Provide a rules engine module that maps extracted data to presets.
- Add a "simulate" mode: run a short 10-second transcode to verify output plays.
Deliverables:
- Source code, file tree
- CLI commands, Docker image
- Tests: at least 10 test cases covering edge data
- QA checklist for browser playback and mobile playback
中文释义:
让 AI 输出一个“转码参数顾问”在线工具：不仅给命令，还要能解释为何这么选，并提供可验证的模拟转码。
关键帧/GOP 检查与异常报警工具
自动检测关键帧过密/过疏、GOP 波动、B 帧异常等问题，输出修复建议。
英文 Prompt:
Create a web app "gop-auditor" that analyzes a video and flags GOP/keyframe issues.
Features:
- Upload video
- Compute keyframe timestamps, GOP lengths, variance, min/max
- Detect:
- no keyframes for long stretches
- irregular GOP pattern
- suspected scene-cut disabled
- time anomalies
- Provide "fix" suggestions with example ffmpeg commands (re-encode with forced keyframes, set gop, scenecut).
Stack:
- Python FastAPI + ffmpeg/ffprobe
- Frontend: Vue 3 + Type
Deliverables:
- Full code + file tree
- Local dev and Docker deployment
- Tests: 6 API tests + 6 frontend tests
- QA checklist
- Include rate limiting and upload constraints
中文释义:
输出一个专门检查关键帧与 GOP 的在线审计工具，帮你快速定位“为什么拖进剪辑软件会卡/为什么网页端跳进度条不稳”。
画质评估（VMAF/SSIM/PSNR）对比报告生成器
上传原始与转码后视频，生成客观指标对比报告与差异摘要。
英文 Prompt:
Build an online comparison tool that computes VMAF/SSIM/PSNR between two videos.
Constraints:
- This is NOT about generating images; only compute metrics and reports.
- Provide a full project (Go backend + React frontend).
- Backend:
- accept two uploads
- align durations and frame rates
- run ffmpeg with libvmaf and compute metrics
- output JSON report + summary markdown
- Frontend:
- show metric charts over time
- allow exporting report bundle (zip)
Deliverables:
- Source code + file structure
- Docker + deployment instructions
- Tests: include at least 8 tests, with mocked exec runner
- QA checklist
- Operational notes: CPU usage, timeouts, job queue
中文释义:
让 AI 交付一个“转码前后画质对比”工具：只做指标与报告，不做任何生成图片/渲染类功能。
音频轨道参数诊断与响度建议（面向视频）
解析视频中的音频轨道，给出响度、峰值、采样率等诊断与标准化建议。
英文 Prompt:
Create a web tool that audits the audio track inside a video file.
- Input: a single video upload
- Extract: audio codec, sample rate, channels, bitrate; compute loudness (EBU R128) and true peak.
- Output:
- diagnostics
- recommended normalization command lines (ffmpeg loudnorm)
- downloadable JSON report
Tech:
- Node.js + BullMQ job queue + Redis
- Worker runs ffmpeg; API returns job status
- Frontend: SvelteKit
Testing:
- Provide unit tests for parsing and integration tests for job lifecycle
- QA checklist for long videos and multiple audio streams
Deliverables:
- Full code + file tree
- docker-compose for API+worker+redis
- Deploy guide
中文释义:
强调“视频内音频”诊断：给出可执行的修复命令与可落地的队列式架构，适合大文件处理。
字幕轨道提取、校验与封装检查工具
检查字幕轨道编码、时间轴、语言标记，支持导出 SRT/VTT 并给出修复建议。
英文 Prompt:
Build an online sub inspector for videos.
Features:
- Upload mp4/mkv
- Detect sub tracks (SRT/ASS/PGS where possible)
- Validate:
- timestamps monotonic
- overlaps and gaps
- encoding issues
- language tags and track disposition
- Export:
- extracted sub s to .srt and .vtt
- validation report
Implementation:
- Full project: Django REST Framework + React
- Use ffprobe to list streams; use ffmpeg to extract text sub s.
- Add limitations note for image- d sub s (PGS): only data/report, no OCR.
Deliverables:
- Code + file tree
- Commands + deployment
- Tests: 10 cases
- QA checklist
中文释义:
这是“字幕封装体检”工具：只做提取与校验，不做任何出图或渲染图生成；对图片型字幕明确限制，不做 OCR。
封装与编码兼容性检查（Web 播放向）
面向网页播放与移动端，给出兼容性结论与重封装/转码建议。
英文 Prompt:
Create a "web-playback compatibility checker" for video files.
Input: upload a video.
Output:
- Compatibility matrix: Chrome/Safari/iOS/Android
- Issues: codec not supported, level too high, b-frames, yuv420p requirement, audio AAC needed, etc.
- Fix recommendations:
- remux command when possible
- transcode command when needed
Stack:
- Rust (Axum) backend + SolidJS frontend
- Provide a ruleset JSON that can be updated without code changes.
Deliverables:
- Full code + file tree
- Build/run commands
- Docker
- Tests: include ruleset tests and API tests
- QA checklist
中文释义:
让 AI 产出“兼容性审计器”：把 web 播放常见坑（像素格式、profile/level、音频编码）转成可执行修复方案。
批量转码队列与报告归档工作台
支持批量上传、队列转码、失败重试、参数模板与结果归档。
英文 Prompt:
Build a batch transcoding web workbench.
Requirements:
- Users upload multiple videos
- Create jobs with a preset (H.264 web, H.265 archive, AV1 experimental)
- Worker pool executes ffmpeg safely
- Provide:
- per-job logs (sanitized)
- output file download s
- job summary report (JSON/CSV)
- retry and cancel
Stack:
- Java Spring Boot API + PostgreSQL
- Worker: separate service using ffmpeg
- Frontend: Angular
Deliverables:
- Full code + file tree
- docker-compose including db
- CI (GitHub Actions example)
- Tests: at least 12 tests
- QA checklist
- Deployment notes: storage (S3 compatible), cleanup strategy
中文释义:
同一主题下的“批处理落地版本”：强调队列、日志、安全、可运维，且交付完整工程与测试。
