<< All versions

Skill v1.0.0

currentAutomated scan100/100
jagz999/explainer-video/kinetic-typography
──Details
PublishedSeptember 28, 2026 at 04:05 AM
Content Hashsha256:8c33992fe1d36134...
Git SHAf5f4b7d175bf
──Files
Files (1 file, 5.4 KB)
SKILL.md5.4 KBactive
SKILL.md · 77 lines · 5.4 KB

version: "1.0.0" name: kinetic-typography description: >- Add animated "kinetic typography" to an HTML page — headlines whose words cascade in one after another, a centered hero headline that gets pushed aside as content/cards slide in, top/bottom supporting lines that slide out of the headline, and a fractured/"cracked" word effect. Use when building motion-graphics-style titles, an animated headline reveal, a product explainer's text, or any "the words move like that launch video" request. Triggers: "animate the headline", "kinetic typography", "make the title words fly/rise in", "cracked text effect", "headline that slides aside when the cards come in", "word-by-word reveal".


Kinetic Typography

Drop-in building blocks (assets/kinetic.css + assets/kinetic.js) for motion-graphics text in plain HTML/CSS — the look of polished SaaS launch videos. Three composable effects:

1. Word-cascade reveal

Each headline line masks its words; the words rise into place one after another behind the mask. splitWords() splits .head .ln > span into per-word .wd spans (preserving an inline colored/gradient word or a .crackword as one unit) and tags each with a global --wi index; the CSS @keyframes wordrise staggers by --wi so the whole headline "assembles" line by line.

html
<div class="head"><span class="ln"><span>Find the <span class="grad">cracks</span></span></span>
<span class="ln"><span>first.</span></span></div>
<script src="kinetic.js"></script><script>splitWords();</script>

Play it by adding .show to an ancestor (e.g. the active scene).

2. Hero → push → reveal staging

A headline appears centered as one line (hero), then content slides in and pushes it aside while the kicker (top line) and subtitle (bottom line) slide out of the headline. Structure each scene as a .lead (kicker + .head + .sub) next to a .cardstage, mark the scene .choreo dk-left or .choreo dk-right, and toggle .cardsin when the content phase begins:

html
<section class="scene choreo dk-right">
<div class="lead"><div class="kicker">Section · Label</div>
<div class="head"><span class="ln"><span>Govern every agent.</span></span></div>
<div class="sub">One line of supporting copy.</div></div>
<div class="cardstage"><!-- your cards / panel --></div>
</section>
  • No .cardsin → headline centered (--heroX recenters it from its docked side), kicker/sub hidden, stage off-screen (--stageHide).
  • Add .cardsin → lead slides to its dock, stage slides in from the opposite edge, kicker/sub slide out. Alternate dk-left/dk-right across scenes for variety.

3. Fractured / cracked word

Wrap a word <span class="crackword" data-text="cracks">cracks</span>. It renders as two copies clipped along a jagged seam; the CSS variable --ck ramps the fracture (0 = solid, 1 = fully cracked) with a red glint in the gap. Drive it in your frame loop so it cracks on cue (e.g. when a voiceover says the word):

js
setCrack(el, clamp((t - crackStart) / crackDur, 0, 1)); // from kinetic.js

Keep --ck at 0 until the word has finished its cascade-in, so it reads "word appears → then cracks."

4. Typewriter + cursor + pill

A fourth staging, for when a travelling element (a logo, a dot, the spark from a reference reel) is already flying through the frame: instead of the words cascading in, the line types, and the travelling element is the cursor typing it.

  • Type without reflow. Build the line once in its final layout — a span per word, an

inline-block span per letter — and per frame set each letter's opacity plus a translateX accumulator that starts at +.18em of tracking and closes to 0 over .25 s. Never animate letter-spacing: that re-lays the line out every frame, so the words crawl and a long line can re-wrap mid-type. Step ≈ 35 ms per letter, capped so the whole line takes at most 1.1 s.

  • Ride the mark to the typing edge. After it lands, lerp the mark from its resting point to the

right edge of the last-typed letter (arc it over the line with a sine so it never crosses the words it just wrote, and shrink it to cursor size on the way out), then ease it back when the line is done — well before the cut, so every exit still starts from the resting point.

  • Land the payoff word in a highlighter pill. The last word sits in a solid accent pill that

wipes in from its left edge (clip-path:inset(0 var(--wipe) 0 0), 100 %→0 over .3 s) as that word's first letter appears, with the letters flat white over it, and the run settles out of a 6° tilt over .35 s.

Helpers and classes for all three live in the text-motion skill (tmTypeBuild · tmType · tmTypeCursor, .tm-type / .tm-pay), and the background half of the move — the rings the mark travelled along and the glow where it landed — is the mark-driven-ground skill. Recipes: "A″ · Spark-driven text" in ../motion-graphics/PLATTER.md.

Notes

  • Units are container-query units (cqw/cqh); put the scene inside a container-type: size stage, or swap to vw/%.
  • These are logical-time friendly: the cascade/push are CSS, the crack is variable-driven — so they render cleanly with the smooth-render skill (scale the transition/animation durations by 1/RATE at render time).
  • To sync cues (a word cracking, a card raising) to a voiceover, pair with the vo-sync skill.
All versions