How to Make Captions Look Good in Short-Form Videos

Good captions are a design problem, not a decoration problem. Here is how to choose type, contrast, position, and timing so the words are easy to read and the video still looks like yours.

Readability first, personality second

Caption design has one non-negotiable job: the words must be readable on a phone, in motion, on a background you do not control. If a viewer has to squint, the style failed no matter how good it looks in the editor. Personality is what you add after the text is legible.

The numbers below are starting points, not universal rules. Type, footage, and platform interfaces vary, so the final test is always the exported file on a phone.

Choose a typeface that survives a phone screen

• Pick a heavy sans with a large x-height. Thin strokes disappear at phone size and over video.
• Use at most two weights in a project. Bold for the caption; a lighter weight only for a secondary line.
• Avoid script and ultra-condensed faces for full captions. They look distinctive in a still and unreadable in motion.
• Use uppercase for short emphasis, not for long lines.
• Keep one family across the project; consistency reads as intent.

Capsly lets you set type, size, color, outline, box, and spacing per caption. The temptation is to use all of it. Resist that.

Build contrast deliberately

White text on a bright sky is not high contrast. White text on a busy street is not high contrast. Contrast is about the local background behind every letter, not the average frame.

• Outline: keeps the footage visible and works on most backgrounds; a heavy outline can look busy on fine detail.
• Box: the most reliable contrast; adds visual weight and hides a little footage.
• Shadow: subtle and soft; the least reliable on bright backgrounds.
• Two-layer approach: a dark outline plus a soft shadow covers most footage without a solid box.

A practical starting point is an outline roughly 4 to 8 percent of the font size. Too thin and it vanishes; too thick and the letters close up. Treat that range as something to test, not a rule: different typefaces and footage need different weights. Check the finished frame in grayscale and at thumbnail size.

Position captions where the eyes already are

For short-form video, center or lower-center is the usual choice: close to the subject’s face, but not on it. Keep a consistent position so the viewer’s eye does not hunt for the next line.

Respect the interface. Reels and TikTok draw controls over the top and bottom of the frame. Keep text out of those bands. Avoid faces, hands, products, and any on-screen text that carries meaning.

If the shot has a natural empty area, such as a wall, sky, or blurred background, use it. Captions work best when they feel placed, not dropped.

Line breaks and reading speed

Captions are read in glances, not paragraphs. Keep one or two lines visible at a time, with roughly three to six words per line. Break at natural phrase boundaries so the viewer’s eye can return to the video.

Aim for roughly 15 to 20 characters per second. If a line needs longer, split it into two cues rather than shrinking the type. Never break a name, a number, or a two-word phrase across lines if you can avoid it.

Word-level timing helps here: adjust the in and out points instead of leaving a line on screen after the speaker has moved on.

Three caption looks that work

The examples below are illustrations rendered in the browser, not screenshots from an export. Each one solves a different problem, and the final check is still the exported file on a phone.

The failure mode to avoid

Use emphasis sparingly

Pick one accent color and use it on the word that carries the meaning, not every other word. Motion should support timing: a quick pop on a new line, a karaoke highlight that follows the voice, a spotlight that moves with the speaker. If the motion makes it harder to read, remove it.

A useful test: pause on any frame. Can you read the caption without the motion? If not, the animation is doing too much.

Style once, reuse everywhere

Consistency is a design decision, not laziness. Save the look as a preset once you have tuned type, outline, position, and motion. Then apply it across a series so episodes feel related. Override a caption only when the footage demands it.

In Capsly, presets live with your account, and per-caption overrides stay available for the exceptions.

To apply these choices to a real transcript, open the subtitle generator. The transcript and the style controls sit in the same editor, so these choices can be tested on the actual frame.

A checklist before export

• Readable at phone size, in motion.
• Contrast holds on the busiest frame.
• Two lines maximum; no orphans.
• Text inside the safe area.
• One accent, used with intent.
• Style consistent across the project.

Then watch the export once at full speed. The goal is not a style that gets noticed; it is a style that lets the words disappear into understanding. For the platform-specific version of this advice, see the Reels caption guide.