Two jobs, two kinds of captions
Before choosing a style, decide what the captions are for:
- Subtitles make speech accessible and followable without sound. They should be quiet and consistent: full sentences, one or two lines, at the bottom of the frame. This is the right choice for YouTube, courses, interviews and anything longer than a couple of minutes.
- Short-form captions are part of the visual design. One to four words at a time, big, bold and often highlighted word by word, they pull the eye through fast vertical videos on TikTok, Reels and Shorts.
Mixing the two up is the most common mistake: three-word pop captions on a 20-minute tutorial are exhausting, and small two-line subtitles on a Reel get ignored. With ClipLime's auto subtitles you can switch between Sentences and Short-form at any time without transcribing again.
Size: bigger than you think, smaller than you fear
Captions are read on phones held at arm's length. If you're judging size on a desktop monitor, you're probably making them too small.
- Subtitles: text height around 4 to 5% of the shorter side of the frame works for most landscape videos.
- Short-form captions: 6 to 9% of the frame's width is common, big enough to read at a glance, small enough that three words fit on one line.
- Use a heavy weight (bold or black) for short-form; semibold is plenty for subtitles.
Always check on an actual phone before publishing. If you have to squint, so will everyone else.
Length and timing
Readable captions respect how fast people read. Some rules of thumb used in professional subtitling translate well to creator videos:
- Keep subtitle lines to roughly 40 characters and use at most two lines.
- Give every caption enough time on screen, under about a second is hard to read unless it's only one or two words.
- Avoid captions longer than around seven seconds; split them at a natural pause instead.
- Break at natural points: after punctuation, before words like and, but, because. A line that ends with "the" or "a" makes the reader stumble.
For short-form, timing is easier: show each group of words exactly while it's being said. Word-level timestamps make this automatic.
Placement and safe zones on vertical video
On TikTok, Instagram Reels and YouTube Shorts, the app draws its own interface over your video: a column of buttons down the right side, and the caption, username and audio information along the bottom. Captions placed in those areas get covered.
That's why short-form captions usually sit in the middle of the frame or just below it, comfortably above the bottom interface and away from the right edge. Each app's layout is slightly different and changes over time, so check your actual frame with the safe zone checker rather than relying on memory.
For landscape subtitles, the bottom of the frame is still the right place, just leave a margin so they don't touch the edge and aren't hidden by a player's progress bar.
Styling for contrast, not decoration
Video changes from bright to dark constantly, so captions need contrast that doesn't depend on the background. Three reliable approaches:
- Outline, a dark stroke around light text. Bold and readable over anything; the classic short-form look.
- Shadow, subtler, good for clean subtitles over darker footage.
- Background box, the most readable option, ideal for accessibility and busy footage, at the cost of covering more of the picture.
Highlighting the current word, in a contrasting color or with a box behind it, helps viewers keep their place in fast speech and is one of the most effective short-form techniques. Use one highlight color and stick with it across your videos so it becomes part of your look.
Uppercase text can look punchy, but long uppercase sentences are slower to read. Reserve it for short captions.
Burned in or separate file?
- Burned-in captions become part of the picture. Use them on platforms with limited caption styling: Reels, TikTok, Shorts, ads, LinkedIn, and whenever the look matters.
- Separate subtitle files (SRT or VTT) can be switched on and off, translated and read by search engines. Use them on YouTube, Vimeo and course platforms.
You can do both: burn short-form captions into the vertical cut and upload an SRT for the long version on YouTube. If a subtitle file is out of sync or in the wrong format, the subtitle fixer shifts, stretches and converts it in seconds.
Proofread names and numbers
Automatic transcription has become very accurate, but it still stumbles on names, brands, jargon and numbers, exactly the words viewers notice when they're wrong. Read through every caption once before exporting, and use find-and-replace for anything that's misheard repeatedly.
Edit the video's timing first, for example by removing pauses, and generate captions afterwards, so you only proofread once.