Why your name plate looks like a template
Because everything in it moves. The bar wipes in, the name flies up, the role fades from the side, a shine sweeps across, and a particle drifts past. Each choice was fine on its own. Together they say: this was assembled from parts.
Motion is a way of pointing. If every element moves, nothing has been pointed at. The professional-looking version usually animates one thing and holds the rest.
Reading time is the constraint nobody budgets for
A viewer needs roughly two and a half to three seconds of *still, settled* text to take in a name and a job title, longer if the name is unfamiliar to them or if they are reading in a second language — which, on the internet, is most of your audience.
Here is where it goes wrong. A three-second lower-third with a 1.2-second animated entrance and a 0.8-second exit leaves one second of readable time. The name was on screen for three seconds and nobody read it. The animation is not reading time.
Budget backwards: decide the hold first (three seconds), then add the in and out on top. Six seconds of lower-third feels long in the timeline and correct to a viewer.
Making the words readable
Type reveals, and how each one feels
- Opacity fade. Neutral, safe, invisible. Good when the graphic is not the point.
- Mask or wipe reveal. A rectangle scaling across the text, revealing it. Reads as printing or typing. Crisp and expensive-feeling. Animate the mask, never the letters.
- Position plus fade. The most common. Keep the distance small: 8–16 pixels of travel, not 200. Long travel on text is the single clearest amateur signature.
- Per-character stagger. Energetic, and illegible past about eight characters. Use it on one short word.
- Scale from 1.04 down to 1.0 with a fade. Subtle, and it reads as authority.
Kinetic typography — text as the main event — follows one rule: the word that changes is the word that moves. And when you cut text to a voiceover, cut on the stressed syllable rather than the word boundary. The stress is where the ear expects the beat, and matching it is the difference between text that dances with the audio and text that merely keeps up.
Where the frame is actually cut
Whatever you design will be reframed. A 16:9 video becomes a 9:16 vertical clip. The platform lays its own interface over the bottom of the frame — captions, a username, a description, a row of buttons down the right edge. The lower third of a 16:9 frame is precisely where a caption bar lands.
Practical rules:
- Keep essential text inside the middle 80 per cent horizontally.
- Keep it above the bottom 20 per cent, even though the name of the graphic tells you to do the opposite.
- If the piece will be cut vertical, check it in a 9:16 crop before you build the graphics, not after.
For legibility, a drop shadow at default settings does very little. What works is a scrim: a soft dark gradient behind the text, tall and low in contrast, or a wide, very soft shadow at low opacity combined with a 1–2 pixel dark outline. Test it against the brightest frame the text will sit over, not an average one.
Fonts carry licences, and a licence for print does not always cover embedding in video. Design foundations covers choosing type; the licence question is worth a minute before you deliver a client's video.
Tools, and the caption style everyone uses
Tools, including the free ones
- DaVinci Resolve (free): the Text+ tool is genuinely powerful — real controls, and the Fusion page underneath it for anything custom. This is the best free option by some distance.
- Premiere Pro: Essential Graphics, with saved templates. Paid, and what most job ads name.
- Kdenlive and Shotcut (free): basic title animation. Fine for a fade and a slide.
- Canva (free tier): animated text presets. Limited easing, no custom curves, and completely adequate for a simple title card.
- CapCut on a phone: auto-captions and text presets, free, and the fastest path if you have no laptop.
- CSS for anything on the web, where a lower-third is just a positioned element with a transition.
The caption style, honestly assessed
One word at a time, large, centred, popping and colour-changing in time with the speech. It works — on a muted feed it holds attention, and the platforms' own tools generate it in one tap.
It is also hard work for anyone reading English as a second or third language, because it removes the ability to read ahead. Chunks of two to four words, held for a beat, keep most of the retention benefit and are considerably kinder. Pick deliberately rather than by default.
And read your auto-captions. They get names, place names, technical terms and any non-English word wrong, reliably, every time.
Today
Mute one of your videos and watch it. If you cannot tell who is talking, what the piece is about, or what you are meant to do at the end, the graphics are decorative rather than working.
Before you move on