Text Motion

Drop-in building blocks (assets/textmotion.css + assets/textmotion.js) for varied, dynamic text motion and living backgrounds in plain HTML — distilled from launch-reel references and a full documentary build. The point: stop every heading animating the same way, and give the frame constant subtle motion (gradient + blobs) so nothing feels like a static slide.

All CSS timing scales by --R so effects survive slow-clock rendering. Every JS updater takes a logical time t (your clock, in seconds) — call them from your render/rAF loop, never off wall-clock, or motion speeds up under capture.

<link rel="stylesheet" href="assets/textmotion.css">
<script src="assets/textmotion.js"></script>

Pick a different entrance per heading (variety is the fix)

Reference reels never repeat one text move. Assign each scene its own:

Keyword glow-highlight sweep (ref: "Truly highlight what matters", "promotion?")

The strongest, most-copied move: a light bar sweeps behind the key word, which then stays lit.

<h3>Encryption ends where usefulness begins:
    <span class="tm-kw warm">in use</span></h3>

tmSweep(document.querySelector('.tm-kw'), /*cue*/ 1.2, t);   // per frame

Variants: .tm-kw (mint), .tm-kw.purple, .tm-kw.warm. Add .lit fires automatically 0.35s after the sweep.

Scramble-decode headline (ref: cipher reels)

tmScramble($('#h'),'Could a machine compute on what it cannot read?',
           0.1, 1.5, t, 'Could a machine compute on what it <em>cannot read?</em>');

Ticker-roll (ref: single-word heroes swapping)

Your <span class="tm-ticker"><span class="tm-roll" id="rl">
  <i>intelligence</i><i>code</i><i>secrets</i><i>data</i><i>prompts</i></span></span> flows out.

tmTicker(document.getElementById('rl'), /*scene t0*/ 0, t);  // holds word 1, then rolls; mask width tracks each word

Living gradient background (ref: mesh / radial / ripple reels)

A background that narrates: a color arc across the film + slow drift = constant motion with zero "busy" feeling.

<div class="tm-gradbg" id="bg">
  <div class="tm-blob b1"></div><div class="tm-blob b2"></div><div class="tm-blob b3"></div>
</div>
<!-- your content sits above with z-index > 0 -->

const ARC = {                      // one palette per scene = the story turn
  problem:{ga:'#0e1622',gb:'#070a12',b1:'#26375e',b2:'#1a3a4e',b3:'#2a2a44'},
  cost:   {ga:'#161008',gb:'#0a0806',b1:'#6e5220',b2:'#3a2f5e',b3:'#7a4a1a'}, // warm/amber
  answer: {ga:'#0b1512',gb:'#080a12',b1:'#1e5a4e',b2:'#3a2f6e',b3:'#2a5a4a'}, // teal
  brand:  {ga:'#1a1206',gb:'#0a0708',b1:'#8a6522',b2:'#6e3f6e',b3:'#a05a2a'}, // gold reveal
};
tmGrad(bg, ARC[scene]);            // on scene change (CSS blends it)
tmDrift(bg, t);                    // every frame (logical time)

Extras: .tm-rings + tmRings(host) for concentric ripples (ref: "AI SaaS Animation"); .tm-grid for a blueprint field (ref: "Introducing").

Critical for rendering: softness comes from radial-gradient falloff, never filter:blur() — blur is ~10× slower in headless Chromium/Edge (software-rasterized) and drops capture to ~5fps. This one rule is why the blobs are gradient-only.

Glass UI accents (ref: prompt card, "Let's go" pill, live cursors)

The "design-tool" motif: a glass pill/button, a prompt card, or named collaborator cursors gliding to the text.

<span class="tm-pill">Let's go ✦</span>
<img class="tm-cursor" id="cur" src="cursor.svg">
<div class="tm-tag" style="background:#e8552e">Sarah</div>

tmCursor(cur, [[0.3,50,46],[1.0,36,44,true]], t);   // path: [t, x%, y%, click?]

Transitions between scenes (no more plain cross-fades)

• The shared .tm-gradbg persists, so a scene change is a content cross-dissolve over a moving gradient — already feels intentional.
• Use .tm-sc on every scene layer for the fade plus a visibility flip after it — this lets scene backgrounds be transparent (so the gradient shows) without the previous scene ever bleeding through. Toggle .on on the active scene each frame.

Spark-driven text (typewriter · cursor · pill · echoes)

When something is already travelling through the frame — a logo, a dot, the spark from a reference reel — the text can belong to it: the line types, the travelling element is the cursor, the payoff word lands in a highlighter pill, and ghost echoes ripple off it.

tmTypeBuild(line, [{w:'Block',at:0},{w:'it,',at:.31},{w:'in',at:.52},
                   {w:'real',at:.64},{w:'time.',at:.88}], {pay:[4,4], echo:true});
tmType(line, t);                                   // per frame
const c = tmTypeCursor(line, t, {space: stage});   // {x,y} of the typing edge — ride your mark there

The whole family (A17–A22) is written up in PLATTER.md under "A″ · Spark-driven text". The background half of the move — the rings the mark flew along, the lit arc behind it, the glow where it landed — is the mark-driven-ground skill.

Part of the motion-graphics super-skill

This skill is family A (text) of the motion-graphics umbrella. The full named menu of ~40 moves across all families (backgrounds, elements, glass/UI, transitions, UI/data) lives in motion-graphics/PLATTER.md, and the shared building-block library (24 helpers) is mirrored in motion-graphics/assets/motion.{css,js} (same source as this skill's assets/textmotion.{css,js}). Ask for "the platter" to get the whole menu.

Composes with

• kinetic-typography — word-cascade + hero-push staging (this skill is the effect library).
• smooth-render — render to MP4; scale CSS durations by 1/rate via --R, drive JS updaters by logical time, and keep the filter:blur rule above or capture starves. Pause ambient animations when idle: document.documentElement.classList.toggle('tm-paused', !playing).
• vo-sync — set each effect's cue time from VO word timestamps so the sweep/scramble/punch land on the spoken word.

Notes

• Units in the assets are %/em so blocks scale to any container; set the accent palette via --tm-accent / --tm-accent2 / --tm-warm on :root or a wrapper.
• Give each scene a distinct entrance — the whole reason this skill exists is that identical headings read as a slideshow.