Extract video frames with FFmpeg and generate visual transcript descriptions.
npx skills add https://github.com/barefootford/buttercut --skill analyze-video
Please help me install this Agent Skill. Skill: analyze-video Source: https://github.com/barefootford/buttercut/tree/main/.claude/skills/analyze-video Command: npx skills add https://github.com/barefootford/buttercut --skill analyze-video
Audio transcripts lack visual context, making it hard to understand the full story of a video. This Skill enriches transcripts with detailed visual descriptions, providing a comprehensive overview for intelligent editing.
Analyze the video at '/path/to/my/product_review.mov' and add visual descriptions to its transcript for the 'product-launch' library.
High-intent search queries and answers about installing and using this skill.
Extract key frames from your video using FFmpeg, analyze them with AI vision to generate visual descriptions, then integrate those descriptions into your audio transcript to create a comprehensive visual transcript for editing.
Yes. FFmpeg intelligently extracts key frames at defined intervals—start, middle, and end for longer videos; a single frame at 2 seconds for short videos—sampling no more frequently than every 30 seconds to balance coverage and efficiency.
You need an existing audio transcript for your video file, FFmpeg installed for frame extraction, and Ruby to run the transcript-cleaning tool. The Skill then generates visual descriptions and updates your transcript JSON incrementally per clip.
Visual transcripts pair audio dialogue with timestamped visual descriptions—like 'close-up of product packaging' or 'user demonstrating feature'—giving editors a complete narrative map for faster rough-cut generation and better scene selection.
Yes. The Skill adapts frame sampling to video duration: longer videos get frames at start, middle, and end; shorter videos get a single frame at 2 seconds, ensuring efficient analysis regardless of length.
Requires FFmpeg for frame extraction from video files, Ruby for transcript processing, and JSON-formatted audio transcripts as input. It outputs updated visual_video.json files with incremental visual entries per clip.
mhagreliusTranscribe spoken content from YouTube videos with optional timestamps.
barefootfordCreate rough cut YAML files and export Final Cut Pro XML from transcripts.
dkyazzentwatwaExtract video metadata and generate JSON or CSV reports.
TheArchitectitExtract and analyze video frames using ffmpeg, ffprobe, and Pillow.
