Legibility first
Outline, weight and shadow are what keep text readable over bright, moving footage at phone size. They matter more than any other styling choice.
Caption styling is not decoration - on a muted autoplaying feed it decides whether your words get read at all. CaptionFX gives you presets to start from and real control over type, colour, outline, placement and pacing, with a preview that matches what the exported MP4 will actually look like.
CaptionFX starts you from style presets, then lets you adjust font, colour, outline, placement and pacing, with a preview that matches the exported MP4. The aim is captions that stay legible on a muted phone screen. The styling is drawn into the video when you export.
Last updated
Outline, weight and shadow are what keep text readable over bright, moving footage at phone size. They matter more than any other styling choice.
Start from a preset built for the platform instead of styling from nothing, then change only what your particular footage demands.
How much text appears at once changes a clip's feel as much as the font does. Short punchy phrases and full readable lines are different products.
Caption styling has exactly one hard constraint: the words must be readable against whatever is behind them, on a phone, in a feed, in under a second. Everything else is taste. Fail that one and the styling did nothing, however good it looks on your monitor.
What makes it hard is that the background is moving and you do not control it. A caption sitting over a bright sky, then a dark jacket, then a white wall has to survive all three. This is why a thin light font with no outline - which reads beautifully on a design mockup - is the most common caption mistake there is. It works for the half-second the footage happens to be dark.
The tools that solve it are unglamorous. A heavy enough weight that strokes do not disappear. An outline or a drop shadow so there is always a contrast edge regardless of what is behind the letter. Enough size that a phone viewer is not squinting. CaptionFX exposes all of these directly, and the presets apply sensible versions of them by default, because the default being safe is more valuable than the default being pretty.
Pacing is how much caption text is on screen at any moment, and it is the styling decision most people never consciously make. The same transcript can appear as one or two words at a time, snapping in sync with speech; as short phrases of a few words; or as full lines that sit long enough to read comfortably. These produce genuinely different videos.
Word-by-word is the high-energy short-form look. It forces the eye to track the audio and makes a clip feel fast, which is why it dominates Reels and Shorts. It is also exhausting over anything long, and it breaks down if the speech is quick, because text replaced faster than it can be read is text nobody read.
Full lines suit explanation. If your clip is a coach making a point or an educator working through something, the viewer needs the sentence to stay put long enough to actually process it. CaptionFX handles pacing as a named setting - short phrases, word-by-word, or long-form lines - that sets the phrase and line limits together, so you choose the behaviour rather than tuning four numbers until it feels right.
Highlighting selected words in a different colour is effective and easy to overdo. It works because it gives a scrolling viewer somewhere to look: the highlighted words carry the gist, so someone who reads only those still gets the point. That is a real benefit on a feed.
It stops working the moment too much is highlighted. If half the words are emphasised, none of them are, and you have just made the captions harder to read. The useful rule is one or two words a line at most, chosen because they carry the meaning rather than because they felt important while you were clicking.
CaptionFX can pick emphasis words for you automatically, and it lets you override that choice by hand for any line. The automatic pass is a reasonable starting point; the manual override exists because only you know which word the sentence actually turns on.
The failure that costs the most time in caption styling is a preview that disagrees with the export. You style a clip until it looks right, render for several minutes, and get back a video with a different font, different spacing or captions in the wrong place. Now you are debugging a renderer instead of making a video.
The usual cause is font resolution. The browser and the server pick fonts independently, so a family the browser has and the render host does not gets silently substituted with a default - and silently is the operative word, since nothing errors. It is worse for Devanagari and other scripts that need real glyph shaping.
CaptionFX works to make the two agree deliberately. The fonts are shipped with the renderer rather than assumed present, they are named explicitly instead of left to the host's font configuration, and where a specific weight is needed the renderer draws a real face for that weight rather than approximating it. When a character is missing from the chosen font, the server names a specific fallback that covers it. The preview is drawing what the export will draw on purpose, not by coincidence.
Yes. CaptionFX gives you type, size, colour, outline, shadow, placement and emphasis controls, with presets as starting points. The controls that matter most for readability are weight and outline, since those are what hold up against moving footage.
A heavy weight with a strong outline, sized generously, with word-by-word or short-phrase pacing and one highlighted word per line. The energy comes from the pacing more than the typeface, and the outline is what keeps it readable over whatever is behind it.
Yes. Devanagari needs correct glyph shaping, which fails silently when a renderer lacks a suitable font - you get boxes or wrong conjuncts rather than an error. CaptionFX ships its caption fonts and names them explicitly to the renderer so Devanagari draws the same in the export as in the preview.
How much caption text is on screen at once. CaptionFX offers word-by-word, short phrases and long-form lines, each setting the phrase and line limits together. Word-by-word suits high-energy short-form; full lines suit anything the viewer needs to think about.
That is the intent, and it is engineered rather than assumed. The usual cause of a mismatch is the browser and the render host resolving different fonts, so CaptionFX ships the fonts with the renderer, names them explicitly, and uses real faces for specific weights instead of approximating them.
Yes. The styling is applied when the captions are rendered into the MP4 frames at export. Until then it is data you can keep changing, and you can export the same source video again with different styling as a separate file.