Utility guide
Text should guide the eye, never clutter the frame
On-screen text carries hooks, names, prices, and calls to action that narration alone cannot anchor. This guide covers which text types to use, typography and contrast rules, animation restraint, and placement that survives platform cropping.
From the Mellius product team · Updated September 28, 2026
Conspiracy theorists, you may hate this
AI Director
›
Just type what you need
- ›Remove all silent parts
- ›Add bold kinetic captions
- ›Only have Bill Ackman speaking
- ›Cut to highlights only
- ›Add cinematic transitions
- ›Match cuts to the beat
The short answer
Use titles for hooks, lower thirds for names, and callouts for single facts, setting each in one legible font at TV-safe sizes. Keep text on screen two to five seconds with high contrast and soft backing, animate entrances once with consistent easing, and hold calls to action three seconds.
Instant legibility
Every text element reads in one glance on a phone at arm's length over the brightest scene.
Clear hierarchy
One dominant message per moment, with supporting text visibly secondary in size and weight.
Platform survival
Text sits inside safe zones so feeds, captions, and buttons never cover the message.
Choose the right text for each job
Open with a title hook of five to eight words that restates the spoken promise, since many viewers decide in two seconds with sound off. Introduce people with lower thirds showing name and role in two sizes — name bold, role lighter — held three to five seconds on first appearance. Use callouts for prices, dates, and single statistics, and reserve full captions for accessibility and muted feeds rather than duplicating every title.
Set body text at sizes readable on phones: titles large enough to dominate, lower thirds smaller but never below roughly one-twentieth of frame height. Stick to one or two typefaces per project with sentence case for body copy, and keep line lengths under about forty characters. In Mellius, text lives as editable timeline objects, so you can ask to restyle every lower third or resize hooks for vertical and undo the change if it misses.
- Give hooks, names, and facts each their own text treatment.
- Limit projects to one or two typefaces with clear hierarchy.
- Hold titles two to five seconds; never flash single words.
- Use sentence case for body text, reserve all caps for short hooks.
Style, animate, and place text safely
Guarantee contrast in every scene with cream or white type over a soft dark box, gradient scrim, or subtle stroke — never bare thin text over bright footage. Animate entrances once per element with a quick fade or slide under half a second, then hold still; perpetual motion exhausts viewers and hurts comprehension. Keep animation grammar consistent across the project so titles feel like a system, not a preset sampler.
Place titles in the central safe zone on vertical video, clear of platform captions, like buttons, and progress bars that consume the bottom and right edges. Common mistake: burying the call to action under end-screen elements or feed UI — hold URLs and buttons center-frame for three full seconds. Proof every word on the export, since typos in burned-in text require re-rendering to fix.
A practical sequence
Put it into practice
- 01
Assign each message a format
Decide what is a hook title, a lower third, a callout, or a caption before styling anything.
- 02
Set type for legibility
Choose one or two fonts, size for phones, and guarantee contrast with backing or scrims.
- 03
Place and hold safely
Keep text in safe zones, animate entrances once, and hold calls to action for three seconds.
Frequently asked questions
What font size should on-screen video text be?
Titles should dominate the frame while body text stays above roughly one-twentieth of frame height to survive phone viewing. Test at arm's length on the smallest target screen rather than trusting the editing monitor. When text must shrink, shorten the words instead — fewer bigger words always beat more tiny ones.
How long should text stay on screen?
Two to five seconds covers most titles and callouts, giving viewers time to read without stalling the video. Lower thirds typically hold three to five seconds on first appearance. Calls to action deserve a full three-second hold so viewers can act on the URL or instruction.
Should text be burned in or added as captions?
Burn in titles, hooks, and calls to action that are part of the design, since they must appear everywhere identically. Use sidecar captions for dialogue accessibility on platforms like YouTube that support toggling. For TikTok, Reels, and Shorts, burn in dialogue captions too, because sidecars are not supported.
Continue reading
Describe the edit. Keep the timeline.
Mellius turns conversational direction into cuts, captions, graphics, motion, color, and sound you can still inspect and change.
Start editing free