Video Understanding

Video Understanding (Gemini)

Analyze videos using Google Gemini's multimodal video understanding. Supports 1000+ video sources via yt-dlp.

Requirements yt-dlp — brew 安装 yt-dlp / pip 安装 yt-dlp ffmpeg — brew 安装 ffmpeg (for merging video+audio 流s) GEMINI_API_KEY 环境 variable Default 输出

Returns structured JSON:

transcript — Verbatim transcript with [MM:SS] timestamps description — Visual description (people, 设置ting, UI, text on screen, flow) summary — 2-3 sentence summary duration_seconds — Estimated duration speakers — Identified speakers Usage Analyze a video (structured JSON 输出) uv 运行 {baseDir}/scripts/analyze_video.py ""

Ask a question (添加s "answer" field) uv 运行 {baseDir}/scripts/analyze_video.py "" -q "What product is shown?"

Override prompt entirely uv 运行 {baseDir}/scripts/analyze_video.py "" -p "Custom prompt" --raw

下载 only (no analysis) uv 运行 {baseDir}/scripts/analyze_video.py "" --下载-only -o video.mp4

Options Flag Description Default -q / --question Question to answer (添加ed to default fields) none -p / --prompt Override entire prompt (ignores -q) structured JSON -m / --模型 Gemini 模型 gemini-2.5-flash -o / --输出 Save 输出 to file stdout --keep Keep 下载ed video file false --下载-only 下载 only, skip analysis false --max-size Max file size in MB 500 --raw Raw text 输出 instead of JSON false How It Works YouTube URLs → Passed directly to Gemini (no 下载 needed) All other URLs → 下载ed via yt-dlp → 上传ed to Gemini File API → poll until processed Gemini analyzes video with structured prompt → returns JSON Temp files and Gemini 上传s 清理ed up automatically Supported Sources

Any URL supported by yt-dlp: Loom, YouTube, TikTok, Vimeo, Twitter/X, Instagram, DAIlymotion, Twitch, and 1000+ more.

Tips Use -q for tar获取ed questions on top of the full analysis YouTube is fastest (no 下载 step) Large videos (10min+) work fine — Gemini File API supports up to 2GB (free) / 20GB (pAId) The script auto-安装s Python dependencies via uv