Caption prompts: word-by-word, phrases, and style copies

Say how the words group (one at a time or 3 to 5 word phrases), how they look, where they sit, and which words get bigger.

Minimal:

Add captions to the whole video.

Standard:

Add captions for all speech. Group them into 3-5 word phrases, bold white text with a thin black outline, centered in the lower third. No pop or bounce animation.

Full:

Caption the whole video word by word, TikTok style. Each word appears as it is spoken, bold sans-serif, white, slightly above center so it clears the bottom UI. When I say "eighty percent" and "never again", make those words bigger and yellow. Turn captions off during the 5 seconds of b-roll at the end.

Copying a style from another video:

Copy the caption style from the reference video exactly. Blocky all-caps font, soft glow, hard drop shadow, two lines max. Apply that to my whole video.

One user asked for "nice and blocky with glow and drop shadow" and got something generic the first time. Naming the three visible things fixed it.

Fixes

• Captions over your face. "Move captions to the lower third."
• Wrong word. "At 0:12 the caption should say gesso."
• Too many words on screen. "Max 3 words per caption."

FAQ

Word-by-word or phrases? Word-by-word for fast talking and hooks. Phrases for anything calmer. If you're not sure, phrases.

Can it fix spelling of names? Yes. List them in the prompt. "Spell these as Vyra, sulansart, gesso."

Related

• /prompts/rough-cut
• /prompts/podcast-clip-captions
• /how-to/choose-captions
• /features/ai-captions

Rules that apply to every prompt

1. One target per prompt. Length, platform, purpose.
2. Quote the exact line when you mean a specific moment.
3. Give numbers, not adjectives. "Pauses over half a second," not "snappy."
4. Say what not to add. Zooms, transitions, music, effects.
5. One fix per follow-up message. Ask for a list of changes with timestamps if you want to check.