By @chandlerintelligence | Watch the video this guide is from
<aside> 🎨
The prompts below include the exact styling I used in my video: fonts, colors, sizes and timing. Swap any of it for your own style.
</aside>
Install Claude Code, then paste this:
Set me up to edit videos with you. Install Remotion, FFmpeg, Whisper (with word-level timestamps) and yt-dlp. Check each one runs.
For anything behind your head you need a person cutout. On a Mac it's built in: tell Claude to use Apple Vision person segmentation. On Windows or Linux, run pip install rembg.
Before Claude edits anything, have it write a storyboard, so you both agree on what the video looks like first.
Before you touch the footage, write a storyboard as a table: one row per beat with the time, my exact words, what's on screen, and the tool. Don't render anything until I approve it.
Cut me out with person segmentation. In Remotion, put the title "Text behind / your head" on two lines between the footage and my cutout, so my head covers part of the letters. Font: Geist Pixel Square, 150 px, #F4F4F2, centred at 20% and 27% of the frame height. Each line fades up over 12 frames (ease-out cubic, rising 2% of the frame), the second 4 frames after the first.
Transcribe my clip with Whisper, word-level timestamps. Caption it in Remotion: Helvetica Neue Bold, 64 px, white, soft shadow, centred at 61.5% of the frame height. Up to 3 words at a time. The word being spoken sits on a coral (#D97757) box with 8 px rounded corners; the others have no box. Each chunk cuts on exactly when its first word is spoken.
Make it better: don't put a coloured box behind every caption. Highlight only the word being spoken.
When I name a tool, its logo appears and the name slides out of it.