Caption type is read in about 400 milliseconds, at arm’s length, over moving footage, often by someone not really trying. That single viewing condition eliminates most of the typefaces people want to use, and it does so for reasons you can name — stroke weight, x-height, aperture, and how the letterforms behave once you add the outline they need to survive a bright background.
The four properties that decide it
- Weight. Caption text is competing with a photographic background. Thin strokes lose, and they lose worse after compression, which eats fine detail first. Bold to black is the working range; regular weights are for interfaces, not video.
- X-height. Lowercase letters that are tall relative to the capitals read bigger at the same point size. This is the single most under-appreciated property in caption typography — two fonts at identical sizes can differ enormously in apparent size.
- Aperture. How open the gaps in c, e, s and a are. Closed apertures turn into blobs at speed, and they close further when you add an outline. This is why some very bold faces are less legible than slightly lighter ones.
- Width. Condensed faces fit more characters per line at a given cap height, which is why they own thumbnails and hero words. On a running caption the trade is real — condensed type is fractionally slower to read — but at three or four words a line it doesn’t bite.
What that looks like
Picks, by job
| Job | Reach for | Why |
|---|---|---|
| Running transcript | Inter Black, Montserrat ExtraBold, Satoshi Black | Open apertures, big x-height, neutral enough to disappear into the content |
| Hero word / title card | Anton, Bebas Neue, Archivo Black | Condensed and heavy: maximum letterform per pixel of height |
| Editorial or authority tone | Playfair Display Black, Instrument Serif, DM Serif Display | High thick/thin contrast reads as considered — one word at a time only |
| Playful, personality-led | Bangers, Luckiest Guy, Fredoka | Comic-lettering ancestry; matches punchy delivery |
| Handmade / diaristic | Permanent Marker, Caveat Brush | Reads as written rather than typeset — accent use only |
Four ways caption type goes wrong
- All caps for the whole transcript. Caps remove the word-shape cues fluent readers rely on. Caps are for the hero word; the transcript should be sentence case.
- An outline heavy enough to close the letters. Outlines exist to separate type from background, and past a certain thickness they start eating the counters. A drop shadow or a soft glow often does the same job with less damage.
- Three typefaces. Two is a system; three is a collage.
- A delicate brand face pressed into service. A typeface chosen for a website will usually fail at a glance over footage. Carry the brand in colour instead.
In Moonshot
Every face in the library is there because it survives the conditions above, and each caption template already pairs a transcript face with a hero face so you aren’t choosing two typefaces under pressure. Switching a template restyles the whole clip instantly — which also means you can audition three type systems on your own footage in about a minute rather than reasoning about specimens.
Which pairing suits which content is the subject of the caption styles field guide.
Quick answers
What's the best font for Reels captions?
A heavy grotesque or a condensed display face, in that order — Inter Black, Montserrat ExtraBold, Anton, Bebas Neue. They win for structural reasons: large x-height, closed apertures, and enough weight to survive an outline. Condensed faces additionally fit more characters per line at a given height, which is why they dominate thumbnails.
Why do thin and script fonts fail on video?
Two reasons. Thin strokes disappear under video compression and against busy footage, and any outline or shadow heavy enough to save them closes up the letterforms. Scripts add a third problem: joined, slanted letters are read as shapes rather than characters, which is fine for a two-word flourish and unusable for a running transcript.
Does the font need to be licensed for commercial use?
Yes, if the video is commercial — and posting for a business or a sponsored clip counts. Most Google Fonts are permissively licensed, but 'free to download' and 'free to use commercially' are different claims, and free-for-personal-use faces are common on font sites. Check before it's baked into a hundred exports.
Should the caption font match my brand font?
Only if your brand font is a heavy sans that survives at speed. Brand typefaces are usually chosen for print and screen reading, not for legibility at a glance on a moving background. Carrying the brand through colour and one accent word is safer than forcing a delicate face into subtitles.
How many fonts should one video use?
Two at most: one for the running caption, one for a hero word or title card. A third is almost always a mistake — the reader stops reading and starts noticing the typography.
Audition type on your own footage
Upload a clip and flip between caption templates live — each one a pre-paired transcript face and hero face, licensed for commercial use.
Open Moonshot free