Captions that look right: styles, bar rules, fixing words, safe zones
Generate captions from the transcript, pick a look, set how bars are cut, fix a word, and keep them out from under the Reels UI.
Captions in BetterEdits are a track of bars. Each bar is a start, an end and its text, linked to nothing. Claude writes them from the transcript; you or Claude fix them like any clip.
You need a video whose footage has been analysed. If it has not, Claude does that first when you ask for captions.
Generate
Caption the video in the default style.
Or in the app: with the video track active, press ⌥⌘T, or click Generate Subtitles in the track's panel. Claude and the button do the same thing: read the words of each clip, map them to the timeline, cut them into bars by the rules in Settings.
Bars never cross a cut. A speaker change ends a bar.
Pick a look
Eight styles ship with the app: Soft Pill, Airy Caps, Editorial, Ink, Mono, Bold Outline, Yellow Pop and Outlined Serif.
Switch the captions to Bold Outline.
Or click the Subtitles row, open the Track tab, and pick from the style menu. Each name is drawn in its own look.
To make your own, open Settings and the Subtitle Styles tab: font, size, colours, background, outline, shadow, uppercase, word highlight, with a live 9:16 preview. A style you save appears in the menu.
A style is copied onto the track when you apply it. Changing the saved style later does not change tracks that already wear it; apply it again.
Set the bar rules
Settings, Subtitles tab. Three numbers decide how words become bars:
- The most words in a bar.
- The longest a bar may last.
- The pause that starts a new bar.
Short bars read faster and suit vertical video. Change the rules, then generate again to cut an existing track by them.
Caption it again with the new rules.
Generating again replaces every bar, edits included. Do captions last, or accept redoing the fixes.
Fix a word
Click the Subtitles row and open the Bars tab. Every bar is there for inline editing: click a time to jump to it, type to fix the text, right-click to merge, split or delete. Split cuts at the middle word and shares the time by text length.
Or say it:
In the captions, "there" at 0:42 should be "their". Merge the two bars around 1:10 into one.
Position and the safe zone
Captions sit at the top, middle or bottom of the canvas, with an inset, per track.
Put the captions at the bottom, a bit higher than default.
On Reels and Stories the app's own UI covers the bottom of the picture. Turn on the Guides menu in the viewer: it shades the top and bottom zones, the feed crop and the profile grid. Keep captions inside the clear area, or ask:
Move the captions up until they clear the Reels caption box.
Claude renders a frame to check.
Hide or burn in
The eye on the Subtitles track hides it from the preview and the export. Otherwise captions are burned in at export with the same renderer the viewer uses, so what you see ships. In an HDR export captions and text are drawn brighter so they sit with the footage.
Captions without a transcript
A clip whose footage has not been analysed gives no words, and Generate says so. You can still write bars by hand in the Bars tab, or in timeline.json, one bar per line.
