heygen-com--hyperframes/skills/embedded-captions/modes/standard/_anatomy.md

---
name: caption-template-anatomy
description: The shared, reproducible scene engine every caption template is built on — a matted talking-head with a flowing verbatim foreground caption and a single climax word, driven by one paused GSAP timeline. Read it ONCE per session (it is identical for all 54 templates); each per-template file in templates/ only overrides the style tokens + the named climax entrance/exit.
metadata:
  tags: caption, talking-head, matte, occlusion, verbatim, climax, gsap, hyperframes
---

# Caption Template — Anatomy (the shared engine)

A **caption template** = one complete, reproducible HyperFrames scene:

> **the person (always in frame)** + a **flowing foreground caption** (verbatim, word-by-word, with appear **and** disappear) + a **climax word** (big, behind the speaker, with a designed entrance **and** exit) + a coherent **font / colour / motion** design.

Every file in `templates/` is the SAME engine described here with three things swapped: (1) the **style tokens** (font, fills, accent, optional gradient/stroke), (2) the named **climax entrance** + **exit** (see `_motion.md`), (3) the **copy** (flow lines + climax word) and **scene/person**. So read this once; each template file is short.

HyperFrames-native, so anyone can reproduce it:

- **One paused GSAP timeline per composition**, registered to `window.__timelines[data-composition-id]`.
- All timing in **seconds**; `data-start` / `data-duration` carry the scene window.
- **Deterministic + seek-safe** only — no `Math.random()`, no `Date.now()`, no infinite repeats, no un-seekable CSS animations. Every state is reachable by seeking the timeline to a time `t`.
- GSAP transform aliases (`x`, `y`, `scale`, `rotation`); animate `opacity`, `filter`, `clipPath`, `textShadow`, `backgroundPosition`, `letterSpacing` — never layout props (width/top/left/margin).

For the composition contract see `hyperframes-core`; eases + the animated-property allowlist see `hyperframes-gsap`; caption grouping/positioning/exit guarantees see `hyperframes-captions`.

## 1 · Asset prep (two CLI calls)

```bash
# 1) The person — transparent talking-head cutout over the scene (VP9 + alpha)
bash scripts/prepare.sh   <project>      # matte ∥ transcribe ∥ safe-zones (THIS skill — not remove-background)
#    (a still works too:  remove-background portrait.jpg -o person.png)

# 2) The verbatim word timings that drive the flowing caption
npx hyperframes transcribe subject.mp4 --model small            # → transcript.json
#    shape: [{ "id":"w0","text":"Hello","start":0.0,"end":0.5 }, …]
```

The flow caption consumes that `transcript.json` directly (word `start`/`end` → reveal + active-word emphasis). The climax word is authored by hand (it is the headline beat, not part of the spoken transcript). In production the avatar/matte pipeline yields a pixel-perfect alpha for free — `remove-background` is the fallback for arbitrary footage.

## 2 · The matte sandwich (HTML)

Six layers, back-to-front. The person is layered ON TOP of the climax so they physically **occlude** it — this is what sells "behind the speaker" (blur/opacity alone reads as a flat overlay).

```html
<div
  class="stage {STYLE} {PERSON}"
  id="cap-{id}"
  data-composition-id="cap-{id}"
  data-start="0"
  data-duration="{SCENE_DUR}"
  data-track-index="0"
>
  <!-- z0  background plate (the original frame, full) -->
  <!--     supplied by .{PERSON} as background-image, or a <video> at z-index:0 -->

  <!-- z1  CLIMAX — big, BEHIND the person, occluded -->
  <div class="climax"><span>{CLIMAX_WORD}</span></div>

  <!-- z4  the person cutout (transparent webm/png), aligned to the plate -->
  <video class="cut" src="person.webm" muted playsinline></video>
  <!-- or:  <img class="cut" src="person.png"> -->

  <!-- z5  grade / vignette for depth + legibility -->
  <div class="grade"></div>

  <!-- z6  FLOW — the verbatim caption, IN FRONT, lower third -->
  <div class="flow"></div>
  <!-- words injected from transcript.json -->
</div>
```

## 3 · Base CSS (layout · z-order · sizing)

Caption size is in **`cqh`** (% of frame height) via a size container, so it honours broadcast spec at any resolution (~8% cap-height for a 16:9 word). The per-template file only sets the `--ff` / `--cfill` / `--cacc` tokens and any `.climax span` fill (gradient/stroke).

```css
.stage {
  position: relative;
  aspect-ratio: 16/9;
  overflow: hidden;
  container-type: size;
  background-size: cover;
  background-position: center 12%;
  background-color: #0a0a0e;
  font-family: var(--ff);
}
.stage > * {
  position: absolute;
}
.cut {
  z-index: 4;
  inset: 0;
  width: 100%;
  height: 100%;
  object-fit: cover;
  object-position: center 12%;
  pointer-events: none;
}
.grade {
  z-index: 5;
  inset: 0;
  pointer-events: none;
  background: radial-gradient(130% 100% at 50% 26%, transparent 40%, rgba(0, 0, 0, 0.6));
}

/* CLIMAX — big, behind person (z1), occluded. line-height ≥1.15 so clip-reveal
   entrances (inset(0)) never slice glyph tops; pad clips negative for script faces. */
.climax {
  z-index: 1;
  left: 50%;
  top: 37%;
  transform: translate(-50%, -50%);
  white-space: nowrap;
  text-align: center;
  line-height: 1.18;
  font-family: var(--ff);
  color: var(--cfill);
  font-weight: 900;
  font-size: 44cqh;
  text-transform: uppercase;
  text-shadow:
    0 2px 13px rgba(0, 0, 0, 0.6),
    0 0 48px rgba(0, 0, 0, 0.42);
}
.climax span {
  display: inline-block;
  opacity: 0;
} /* GSAP reveals it */

/* FLOW — verbatim caption, in front (z6), lower third */
.flow {
  z-index: 6;
  left: 50%;
  bottom: 9%;
  transform: translateX(-50%);
  width: 90%;
  text-align: center;
  line-height: 1.15;
  font-family: var(--ff);
  font-weight: 700;
  font-size: 7.5cqh;
  color: var(--cfill);
}
.flow .w {
  display: inline-block;
  opacity: 0;
  margin: 0 0.1em;
  color: var(--cfill);
}
.flow .w.act {
  color: var(--cacc);
} /* the currently-spoken word */

/* tokens every template overrides: */
.stage {
  --ff: "Inter";
  --cfill: #fff;
  --cacc: #10a37f;
}
```

**Legibility on busy/bright scenes:** a behind-the-person climax needs separation from the footage, not just a fill colour. For dark or gradient fills on lit scenes give the climax an outline — `-webkit-text-stroke:1px rgba(0,0,0,.5);paint-order:stroke fill` — a dark drop-shadow alone fails against highlights (e.g. a lamp). Gradient/clip fills must live on `.climax span` (the text node), never on the transformed `.climax` container, or the clip detaches and only the shadow shows.

## 4 · One paused GSAP timeline (the loop, made seek-safe)

The gallery used a `setInterval` loop; HyperFrames needs the same beats as **absolute-time tweens on one paused timeline**. The cycle is **FLOW line → (FLOW line) → CLIMAX in → hold → out**. Restraint is the rule: flow stays clean; the one big mood move happens only at the climax.

```html
<script src="https://cdn.jsdelivr.net/npm/gsap@3.14.2/dist/gsap.min.js"></script>
<script>
  window.__timelines = window.__timelines || {};
  const stage = document.getElementById("cap-{id}");
  const flow = stage.querySelector(".flow");
  const climax = stage.querySelector(".climax span");
  const tl = gsap.timeline({ paused: true });

  // --- FLOW: render words from transcript.json, reveal each at its start time ---
  // WORDS = grouped transcript lines: [{ words:[{text,start,end}], end }] in scene-local seconds.
  function renderLine(line) {
    flow.innerHTML = line.words
      .map((w, i) => `<span class="w" data-i="${i}">${w.text}</span>`)
      .join(" ");
    return [...flow.querySelectorAll(".w")];
  }
  WORDS.forEach((line) => {
    const spans = renderLine(line); // (one renderLine per active line; see captions skill for multi-line groups)
    spans.forEach((el, i) => {
      const w = line.words[i];
      tl.add(FLOW_IN(el), w.start); // FLOW_IN  = the template's flow entrance (see _motion.md)
      tl.set(spans, { className: "w" }, w.start); // active-word sweep: only the spoken word gets .act
      tl.set(el, { className: "w act" }, w.start);
    });
    tl.add(FLOW_OUT(spans), line.end); // FLOW_OUT = flow exit
    tl.set(flow, { autoAlpha: 0 }, line.end + FOUT); // hard-hide the group so old text can't linger
    tl.set(flow, { autoAlpha: 1 }, line.end + FOUT + 0.001);
  });

  // --- CLIMAX: entrance at the beat → hold ≥1s → exit (named recipes from _motion.md) ---
  const T = CLIMAX_AT; // beat time (after the flow lines)
  tl.add(CLIMAX_IN(climax), T); // e.g. SLAM / DEBLUR / INK-LOOM / CHROME-SWEEP …
  tl.add(CLIMAX_OUT(climax), T + CLIMAX_HOLD); // hold ≥1s, then exit. CLIMAX_OUT ends opacity:0 (hard exit)

  window.__timelines["cap-{id}"] = tl;
</script>
```

`FLOW_IN` / `FLOW_OUT` / `CLIMAX_IN` / `CLIMAX_OUT` are the named recipes in **`_motion.md`** — each returns a GSAP tween/timeline so the per-template file just picks four names. A simpler equivalent for the flow active-word _glow_ (rather than discrete reveal) is the single-driver envelope in `hyperframes-animation/rules/asr-keyword-glow.md`.

## 5 · How to choose values

- **SCENE_DUR** — must equal `data-duration`. Typical 6–10 s for a looping demo card.
- **WORDS grouping** — 2–4 words / line, ~380–520 ms per word (premium pacing is slower than Hormozi). Group from `transcript.json`; keep `end < next.start` (monotonic).
- **CLIMAX_AT** — place the climax _after_ the flow lines clear, on the narration's emphasis beat.
- **CLIMAX_HOLD** — **≥1 s** of settled dwell _after_ the entrance finishes (the climax is the headline beat). Entrances run 0.6–1.6 s, so e.g. hold = entranceDur + 1.0–1.6.
- **FOUT** — flow exit ≈ 0.5 s. **Exit ≈ 75 % of entry** for every element (arrival deliberate, departure swift; see `_motion.md`).
- **Climax size** — base 44 cqh; long words bleed off-frame (intended cinematic); 3-char words behind a centred subject need size + an outline so they peek.

## Critical constraints (HyperFrames)

- Timeline **paused**; registry key = `data-composition-id`.
- **No CSS keyframe animation** on caption elements — all motion is GSAP tweens at absolute times (seek-safe).
- No `Math.random` / `Date.now` / infinite repeats.
- `display:inline-block` on every `.w` and the climax `span`.
- **Hard-hide** each flow group at its end time; **CLIMAX_OUT ends at `opacity:0`** (or fully-clipped) so nothing lingers.
- Gradient / `background-clip:text` / stroke fills go on `.climax span`, not the transformed `.climax`.
- `.climax` `line-height ≥ 1.15`; pad clip-reveal entrances with negative insets for script/decorative fonts.

## Pairs with HF skills

- `media-use` — `remove-background` (the matte) + `transcribe` (word timings).
- `hyperframes-captions` — transcript consumption, grouping, positioning, exit guarantees, `fitTextFontSize`.
- `hyperframes-animation/rules/asr-keyword-glow.md` — the verbatim active-word envelope.
- `hyperframes-gsap` — single paused timeline, transform aliases, ease palette.
- `_motion.md` (this folder) — the named flow/climax entrance + exit recipes.