Hubbestaiskill.com

Home / All Skills / Content & Media Creation / captions-overlay

Content & Media Creation · heygen-com/hyperframes

captions-overlay

Overlay doctrine for the embedded-captions workflow — the caption MODEL (drop / rail / embed) and the rule that captions are an OVERLAY composited on top of the film, never a reserved bottom band you shift content up to avoid. Load when adding captions/subtitles to a talking-head or launch video, when deciding whether a phrase should be dropped, ride the verbatim rail, or be promoted to a scarce embedded climax, when laying out a composition that will carry captions (do NOT reserve a keep-out band), or when centering a composition on the true frame center under captions. Quotes the rail+embed model from embedded-captions and constraint #13 (captions overlay, keep-out band retired) from the product-launch-video scene agent. Applies ON TOP of embedded-captions.

L1 Basic · Source registration onlyV1 Commit-PinnedDual-pass Audit

Author heygen-comGitHub heygen-com/hyperframes ↗Stars 44283License Apache-2.0（仓库根 LICENSE 为 Apache License Version 2.0）Commit d2f0bc7f34

Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

The doctrine fuses two rules into one: (1) The caption model (every spoken sentence has only three fates: drop / rail / embed, where embed is a scarce, 'earned' climax); (2) The overlay law (caption lines are an overlay composited 'on top of' the film, not a reserved area, so content must never be shifted up or given a dead margin).

.claude/skills/captions-overlay/SKILL.md

Two ideas combine here. First, the **caption model** — every spoken phrase is `drop`, `rail`, or `embed`, and embed is the scarce earned peak, not the default. Second, the **overlay law** — a caption line is composited ON TOP of the film as an overlay; it is NOT a reserved zone, so you never shift content up or leave a dead band to "make room" for it.

What this does: This is not an executable workflow, but a criteria document—it provides classification decisions (which of the three states each phrase belongs to) and composition prohibitions (never reserve a keep-out band); actual rendering is performed upstream by embedded-captions and product-launch-video finalize.

Scarcity is a hard constraint: rail carries the vast majority of text, embed ≤1/beat, never adjacent or in the same frame, separated by ≥ one beat; short videos typically have one embed, long explainers about one per section. 'Embedding every word' is a common mistake.

.claude/skills/captions-overlay/SKILL.md

**The rail carries most of the text; embed is the scarce, earned peak** — ≤1 per beat, never two adjacent/co-visible, spaced ≥ a beat apart. A short clip → usually one embed; a long explainer → ~one per section. Embedding every word is the common mistake.

What this does: It uses an upper bound to eliminate the most common mistake of 'embedding everything', while presenting both embed contexts: Standard mode (rail = verbatim lower third, embed = climax behind the subject) and Cinematic mode (rail dropped, full embed-style).

The verifiable expression of the overlay law is 'compose around the true vertical center': y = H/2 (540 for landscape, 960 for portrait), and content can be drawn all the way to the canvas bottom; 0.42×H plus a bottom dead zone is a bug, not a fix.

.claude/skills/captions-overlay/SKILL.md

- **Center the composition on the TRUE vertical center — y = H / 2** (landscape 540, portrait 960). Do not shift content up to "make room" for captions; a composition centered at 0.42 × H with a dead lower band is the bug, not the fix. - Content may extend to the canvas bottom.

What this does: It provides testable geometric numbers (0.42×H as a negative example), turning 'did we secretly yield space for captions' into a verifiable proposition rather than aesthetic taste.

The only retained soft avoidance: do not place small, critical readable text (URL line, legal copy, secondary subtitles) right in the bottom ~80px center zone; large graphics, cards, and environmental elements under captions are fully allowed.

.claude/skills/captions-overlay/SKILL.md

**One soft courtesy rule:** avoid parking _critical small readable text_ (a URL line, a legal line, a sub-caption) exactly in the bottom ~80px center span where the caption line sits — the overlay would fight it. Large imagery / cards / ambient content under the captions is fine; the caption skin is designed to read over content.

What this does: It narrows avoidance down from an 'area-level' rule to 'critical small text' alone, clarifying that it is a soft rule (courtesy) rather than a machine gate.

Machine gates are deprecated: the legacy captions.mjs keepout check has been retired; readability of captions over content is now judged visually via finalize snapshot QA.

.claude/skills/captions-overlay/SKILL.md

- There is no machine keep-out gate (the old `captions.mjs keepout` check is retired). Finalize snapshot QA judges caption-over-content legibility visually.

What this does: Clarifies that there is 'no automated acceptance' and explicitly shifts acceptance responsibility to visual snapshot inspection—this is this skill's acceptance standard and what distinguishes it from previous approaches with keepout checks.

The relationship with the upstream skill is 'additive' rather than 'merged': this doctrine applies ON TOP of embedded-captions; do not expect the upstream skill to absorb it.

.claude/skills/captions-overlay/SKILL.md

> **Overlay doctrine — supplements the upstream `embedded-captions` skill. Applies ON TOP of it; do not expect it folded into the upstream skill.**

What this does: Explains why this internal technique skill coexists with the marketplace skills/ directory—the repo's internal layer (.claude/skills/) applies patches here without altering upstream packages.

Invocation timing (trigger conditions) is specified in the description: when adding captions to talking-head or launch videos, deciding whether a phrase is drop / rail / embed, laying out compositions that will host captions (and NOT reserving a keep-out band), and centering compositions to the true frame center under captions.

.claude/skills/captions-overlay/SKILL.md

Load when adding captions/subtitles to a talking-head or launch video, when deciding whether a phrase should be dropped, ride the verbatim rail, or be promoted to a scarce embedded climax, when laying out a composition that will carry captions (do NOT reserve a keep-out band), or when centering a composition on the true frame center under captions.

What this does: The trigger terms serve as a 'when you must load' checklist, covering model classification, composition layout, and center alignment tasks.

2Core Capabilities

01Spoken sentence tri-state classification (drop / rail / embed) and presentation formats for each

02Embed scarcity control (≤1/beat, non-adjacent, ≥1 beat interval, at most one apex)

03Overlay composition law (true center y=H/2, no keep-out band, full-bleed to bottom permitted)

04Soft avoidance criteria for bottom ~80px center zone (only critical small text yields)

05Dispatching Standard vs. Cinematic modes (former for verbatim readability, latter only for purely cinematic intents)

06Acceptance path: no machine gate; readability of captions over content is visually judged via finalize snapshot QA

07Defines default caption rendering appearance (single line, bottom-centered, ~5-8% canvas height, overlay layer)

3External Dependencies

Kind	Dependency / Endpoint
package	上游 skill 依赖（被本教义叠加）：embedded-captions（rail+embed 模型出处）与 product-launch-video 的场景/finalize 代理（约束 #13）
cli	已退役的 captions 检查入口（仅在文中被点名为「不再存在」的检查名，本目录不含该脚本）

4Recommended Governance & Evidence

Suggested Controls for Safe Operation

For Agent Runtime

Can be auto-loaded

For Administrators

Register source and version

Evidence & Triggers (File Paths & Excerpts)

SKILL.md: Pure prompt directives and rule text; no script execution or external side effects

Audit Notes: Evaluated against the six-tier grading criteria (evaluation target = skill directory itself): the entire directory contains only 1 markdown file (SKILL.md, 6,095 bytes), with no scripts, no executable code blocks required to run, no URLs/domains (zero hits in full-text URL scan), no process.env/credential/keychain access, and no local disk-write commands. It only defines three types of criteria: 'captions are an overlay, compose on true center, keep embeds scarce', delegating verification to human review. The attack surface is zero, so it is graded Green.

5Second-Pass Verification

[ok] 是否真的没有任何可执行物（green 档的关键前提） — rglob 全量列举：只有 `.claude/skills/captions-overlay/SKILL.md` 一个文件；无 scripts/、无 assets/、无 references/。全文无 shebang、无 ```bash 块。
[ok] 是否含隐藏的网络外发（域名/端点） — `https?://` 在本文件零命中；无 curl/npx/CLI 调用语句。被点名的 `captions.mjs keepout` 是「已退役」的历史检查名，本目录并不含该文件（全量清单已证）。
[discrepancy] meta.official_desc 是否与 frontmatter 原文一致 — 本 SKILL.md 的 description 是**未加引号的 plain scalar**，正文里含 `constraint #13 (...)`——YAML 把空格后的 `#` 当注释起点，`yaml.safe_load` 因此只解析到 `…embedded-captions and constraint`，丢掉后半段（解析长度 643 vs 原文该行 782 字符）。为避免展示层出现半句话，本 JSON 按**原文逐字**保留完整 description（含 `#13 (captions overlay, keep-out band retired) … Applies ON TOP of embedded-captions.`），该串在源文件中逐字可命中。此为上游 frontmatter 的书写瑕疵，非本 skill 的行为差异。
[ok] 「验收靠人眼」是否被误读为「无验收」 — 原文两条并存：`There is no machine keep-out gate … Finalize snapshot QA judges caption-over-content legibility visually.` 与底部 ~80px 的软避让规则。结论：有验收，但验收方式从机器闸门改为快照目视——第一遍表述准确。
[ok] 与上游 embedded-captions 的关系是否有事实依据 — SKILL.md 开篇 blockquote 与 description 双重声明 ON TOP 关系；引用的 rail+embed 与 constraint #13 均标注了出处（embedded-captions / product-launch-video scene agent），未见与上游冲突的改写。

6Verdict & Scenarios

Turns the subjective 'should content yield to captions' intuition into a verifiable geometric proposition

Quantifies scarcity rules into actionable constraints (≤1/beat, non-adjacent, ≥1 beat interval), shutting down the common mistake of 'embedding entire sentences'

Explicitly declares changes in acceptance methodology (machine gate retired → snapshot visual QA), preventing false assumptions about automated checks

Narrows the avoidance scope down to a single soft rule for 'critical small text', preserving full-bleed composition freedom

Clearly states upstream relationships and trigger conditions, avoiding redundant or contradictory rules with embedded-captions

Best for: Best for: Determining 'what to drop, what stays on the rail, and which word earns an embed' when adding captions to talking-head or product launch videos; laying out compositions meant to host captions (ensuring no secret keep-out bands, aligning to true frame center); providing a reference doctrine against the common rework mistake of 'shifting content upward for captions'. Intended to be loaded on top of /embedded-captions and /product-launch-video.

Not for: Not suitable for: (1) Scenarios looking for an executable workflow or tooling—contains no scripts, templates, or commands; (2) Scenarios requiring machine-verifiable caption safe-zone checks—the keepout gate has been retired, use snapshot QA or custom checks instead; (3) Pure subtitle burn-in without composition decisions—use upstream embedded-captions directly; (4) Scenarios needing multi-language or subtitle format (SRT/VTT) specifications, which this doctrine does not cover.

Install Agent-Ready

① Verified Mirror 2026-09-28

Option A · Direct DownloadDownload captions-overlay.tar.gz

sha256: 4983e91a90bd624b…

Option B · Copy Guide for Agent

Agents automatically download and verify mirrors from this JSON guide.

② Upstream GitHub

GitHub Repo ↗

Pinned to commit d2f0bc7f34.