Adding Text to Video: Titles, Captions, and Corner Marks
A title, a source credit, a timestamp, a disclaimer — text on video takes seconds in the browser. The hard part is not adding it; it is adding it readably without fighting the picture.
Decide the text’s job first
Three kinds of text, three treatments: titles (fixed, always visible), annotations (appear for a moment, tied to a moment), corner marks (small, permanent, in a corner).
Classify before you type. It prevents the classic mistakes: a permanent title sitting dead-center over the subject, or a three-second notice that never leaves.
Finalize wording before processing. Every wording change means reprocessing the video — thinking costs less than re-rendering.
Position: the 3x3 grid
Positions map to a nine-cell grid: tl, tc, tr across the top; cl, cc, cr through the middle; bl, bc, br along the bottom.
Conventions: titles at tc or cc; disclaimers and credits at bl or br; emphatic one-liners at cc. Avoid cl and cr on vertical video — thumbs live there.
Leave bc for spoken subtitles. If subtitles come later, corner marks must not take bc or the two layers will collide.
Size and color: readability first
Type size scales with frame height, so one setting reads the same at any resolution. Titles generous, annotations medium, corner marks small. When unsure, go bigger — small text is unreadable on phones.
White or black first: white on dark footage, black on light. Yellow and red are for emphasis only. If the background is unpredictable, add a translucent backing or pick a high-contrast pair.
Opacity lets text step back: corner marks at 70-80%; titles fully opaque. Below 50% opacity, text vanishes on bright footage.
Time-limited display
Text supports a display range — show it only when it matters. Annotations like “frame skip at 3s” or “recorded 2024” feel clean when they come and go.
Timing convention matches trimming: appear 0.3s early, leave 0.3s late, giving viewers time to read.
Multiple different texts (opening title plus closing credit): process in two passes, each with its own range, rather than stacking layers in one go.
Relation to burned-in subtitles
This tool is for sparse key text. Full dialogue belongs to the subtitle workflow — edit an SRT and burn it in one pass; pasting sentences one by one does not scale.
Both approaches burn pixels into the picture: text cannot be separated or toggled afterwards. For switchable subtitles, keep them external in the container instead.
After adding text, check once at the start, once at the end, once on a phone in portrait. The small screen is the final exam for readability.