Kinetic typography videos with AI: script, sizing and timing

How to make kinetic typography videos with AI: beat-by-beat scripting, type sizes for phones, timing per word, color rules and a full 30-second example.

October 14, 2026, 5 min read

What kinetic typography is and when it works

A kinetic typography video is one where the words are the visuals: they slide, scale, snap and stack in time with a voiceover or music. To make one with AI, write a short script, split it into beats of a few words, pick one type system, and have a tool draw the text as graphics instead of asking a video model to paint letters.

It works best when the message is the product: a manifesto, a launch announcement, one strong argument, or a line you want people to remember. It works badly when the viewer needs to see a thing, like a recipe or a product in use. If your audience would say “show me,” generate footage and use type as a supporting layer.

Kinetic type is cheap to iterate. There are no faces to keep consistent and no hands to fix. Your effort goes into the words and the rhythm, which is where it belongs.

Write the script for the eye, not the ear

Start from a voiceover of 60 to 90 words for a 30-second piece, then cut it into on-screen beats. A beat is the group of words that appears together. The voice can say more than the screen shows. The screen carries the spine of each sentence.

Three to six words per beat. One idea per beat.
Put the key word last in the beat so it lands on the stress of the voice.
Cut filler such as “that,” “really” and “in order to.” The screen has no room for it.
Write numbers as digits. “3 steps” reads faster than “three steps.”
Mark one word per beat for emphasis with size, weight or color. Never more than one.
Read the beats aloud with a timer. If a beat takes longer than about two seconds to say, split it.

Type sizes and safe zones on a phone

Most kinetic type is watched on a phone, often muted, in a 9:16 frame. Design on a 1080 by 1920 canvas and use these as starting points, then test on a real phone at arm’s length.

Hero words: 160 to 240 px font size, one to three words per line.
Supporting lines: 72 to 110 px.
Smallest text, such as a source or a handle: 40 px, and only if it stays up for two seconds or more.
Line length: about 18 characters per line for hero text and 28 for supporting text.
Keep text inside a centered area roughly 900 px wide. Leave the top 220 px and the bottom 380 px clear for the app’s buttons, username and caption. Overlays change, so check the platform’s current spec page.

Use one type family with a heavy weight and a regular weight. Geometric and grotesque sans serifs survive compression best. Hairline weights and thin serifs shimmer and break up after upload.

Pacing each beat

Rhythm is what separates kinetic type from a slideshow. Lock each beat to the voice or the music, not to an even timer.

Entry: 6 to 10 frames at 30 fps, so the word is readable almost at once.
Hold: long enough to read the beat twice at a calm pace. For four words, that is usually 1.2 to 2 seconds.
Exit: 4 to 8 frames, or no exit at all if the next beat pushes the old one off.
Every three or four beats, change the move. A scale-up after three slides, a hard cut after three wipes.
Give the payoff line the longest hold in the video, then half a second of stillness before the call to action.

Easing matters more than the move. Ease-out on entries makes words decelerate into place. Save linear motion for slow background drift.

Color, contrast and motion vocabulary

Use two colors plus one accent. Off-white text on a near-black background, with one bright accent for the emphasized word, is the safest system. Aim for a contrast ratio of at least 4.5:1 for supporting text and 3:1 for very large hero text, which are the WCAG thresholds for normal and large text. Any free contrast checker gives you the ratio in seconds.

Avoid red on blue and red on green. Those pairs vibrate on screen and fail for many colorblind viewers. If text sits over footage, put a solid or blurred panel behind it.

Pick three or four moves and stick to them, for example slide, scale, mask reveal and hard cut. A grid that text snaps to makes simple moves look designed.

A full 30-second example beat sheet

Here is a complete beat sheet for a made-up refillable bottle brand called Tidewell. Times are targets. The voiceover runs underneath, and the accent word is in capitals.

0.0–1.8 s · “Stop buying” (200 px) then “PLASTIC bottles” slides up beneath (96 px). VO: “Stop buying plastic bottles.”
1.8–3.6 s · “You refill your PHONE” snaps in word by word. VO: “You refill your phone every day.”
3.6–5.4 s · “Why not your WATER?” Scale-up on the last word. VO: “Why not your water?”
5.4–8.0 s · “TIDEWELL” fills the width, hard cut in. VO: “Meet Tidewell.”
8.0–11.0 s · “Steel walls” / “Cold all DAY.” Mask reveal left to right. VO: “Double steel walls keep it cold all day.”
11.0–14.0 s · “One bottle.” then “Every TRIP.” Two hard cuts. VO: “One bottle for every trip.”
14.0–17.5 s · “Fits a CUP HOLDER.” Slides in from the left. VO: “It even fits a cup holder.”
17.5–21.0 s · “Wide mouth.” / “Easy to CLEAN.” VO: “A wide mouth makes it easy to clean.”
21.0–25.5 s · “Refill.” then “Repeat.” Longest hold, then half a second of stillness. VO: “Refill. Repeat.”
25.5–30.0 s · “TIDEWELL” small at top, “Link in bio” at 96 px. Music resolves.

That is ten beats, 33 on-screen words and about 60 spoken words. If you need more words, make the video longer rather than packing the beats.

How to make it in ClickPop

Video models still struggle to paint long, correctly spelled text inside a generated scene, so draw the letters as graphics. In ClickPop, Faceless Studio and the motion graphics styles in Create Video, such as Swiss, draw kinetic type in code, so every word is exact.

Paste the voiceover and the beat list, with the accent word marked in each beat.
Choose a style and a palette, and set the format to 9:16.
Pick an AI voice and a music bed, or upload your own voice track.
Preview and fix any beat that feels rushed before export.
Once the vertical cut works, export a 4:5 version for feeds.

Mistakes to avoid

Too many words per beat. If a viewer has to pause to read, the rhythm is gone.
A different move for every word. It reads as noise, not energy.
Text in the bottom caption zone, where the app covers it.
Platform auto captions on top of kinetic type. You get two sets of words fighting.
No hook in the first beat. Open on the strongest line, not the logo.

Questions

Can AI make kinetic typography?

Yes, but the text should be drawn as graphics rather than painted by a video model, which often misspells words. Tools that render type in code give you exact letters with AI handling voice, music and layout.

How many words should be on screen at once?

Three to six words per beat works for most vertical videos. Hero lines can drop to one or two words.

What font works best for kinetic type?

A sans serif with a heavy and a regular weight. Thin weights break up after platforms compress the video.

How long should a kinetic typography video be?

Fifteen to 45 seconds suits most social posts. Let the script decide, and keep beat length steady rather than cramming words.