html-explainer

15 Commits 1 Branch 0 Tags

2 Commits

Include renames

Author	SHA1	Message	Date
MohZhuang	3aec9f6701	feat(v1.2.3): 局部重渲开关、修片尾被截、QC 抽帧双判据渲染器 - 新增 --only <场景id>：改一句文案只重渲这一场（14 场 6194 帧全渲 20 分钟 → 末场 78 秒） - 新增 --mux-only：跳过截图，用现有帧重新合成 - 合成弃用 -shortest，改用 -af apad -t <renderSec>：音轨不足处补静音，成片长度由 layout 唯一决定（此前片尾留白连同末块字幕被按音轨长度切掉）体检 - qc_check.py 抽帧判据改「体积 + 内容」双条件：体积低于中位数 35% 时再看色数与标准差，避免极简画面（片头淡入起点、纯文字面）稳定误报 - 时长差改列「音轨 / layout 期望 / 成片」三个口径，并把「成片比音轨短」定为故障文档 - lessons.md 新增 35–45 条：入场动画的截止时刻、共享 kit.css、数字卷轴写法、结尾三类废句与「勾选框全空的待办清单」、极简帧信息重复、--only 的两个前提 - lessons.md 第 33 条同步修正：单场重渲已有开关、合成命令不再用 -shortest - workflow-guide.md 新增「改一句文案之后」子流程版本号同步至 1.2.3（SKILL.md / 双语 README / CHANGELOG / issue 模板）	2026-09-22 18:14:11 +08:00
MohZhuang	66b61002f2	chore: prepare html-explainer v1.2.2 for public release First public release of the html-explainer skill: deterministic HTML-to-MP4 explainer video pipeline for AI coding agents. Additions since v1.2.1: - scripts/lint_frames.py: pre-render audit of the eight frame-contract rules - scripts/peek_frame.mjs: single-scene screenshot preview (seconds, not minutes) - scripts/check_integrity.py: version/catalog/count consistency + stray-CDN scan - Bilingual README (README.md Chinese default, README.en.md English) - LICENSE (MIT), THIRD_PARTY_NOTICES.md, licenses/Apache-2.0.txt - CHANGELOG.md, CONTRIBUTING.md, .gitignore, .gitattributes - GitHub CI workflow and issue/PR templates Fixes and docs: - document lint_frames.py and peek_frame.mjs, previously absent from the file table - add the lint stage to the pipeline order - remove references to private client projects so the skill can ship publicly - record the full upstream attribution chain for the style catalog - package_skill.py now excludes repo-hygiene files and keeps licence files	2026-09-21 12:46:23 +08:00

Author

SHA1

Message

Date

MohZhuang

3aec9f6701

feat(v1.2.3): 局部重渲开关、修片尾被截、QC 抽帧双判据

渲染器
- 新增 --only <场景id>：改一句文案只重渲这一场（14 场 6194 帧全渲 20 分钟 → 末场 78 秒）
- 新增 --mux-only：跳过截图，用现有帧重新合成
- 合成弃用 -shortest，改用 -af apad -t <renderSec>：音轨不足处补静音，
  成片长度由 layout 唯一决定（此前片尾留白连同末块字幕被按音轨长度切掉）

体检
- qc_check.py 抽帧判据改「体积 + 内容」双条件：体积低于中位数 35% 时再看
  色数与标准差，避免极简画面（片头淡入起点、纯文字面）稳定误报
- 时长差改列「音轨 / layout 期望 / 成片」三个口径，并把「成片比音轨短」定为故障

文档
- lessons.md 新增 35–45 条：入场动画的截止时刻、共享 kit.css、数字卷轴写法、
  结尾三类废句与「勾选框全空的待办清单」、极简帧信息重复、--only 的两个前提
- lessons.md 第 33 条同步修正：单场重渲已有开关、合成命令不再用 -shortest
- workflow-guide.md 新增「改一句文案之后」子流程

版本号同步至 1.2.3（SKILL.md / 双语 README / CHANGELOG / issue 模板）

2026-09-22 18:14:11 +08:00

MohZhuang

66b61002f2

chore: prepare html-explainer v1.2.2 for public release

First public release of the html-explainer skill: deterministic HTML-to-MP4 explainer video pipeline for AI coding agents.

Additions since v1.2.1:

- scripts/lint_frames.py: pre-render audit of the eight frame-contract rules
- scripts/peek_frame.mjs: single-scene screenshot preview (seconds, not minutes)
- scripts/check_integrity.py: version/catalog/count consistency + stray-CDN scan
- Bilingual README (README.md Chinese default, README.en.md English)
- LICENSE (MIT), THIRD_PARTY_NOTICES.md, licenses/Apache-2.0.txt
- CHANGELOG.md, CONTRIBUTING.md, .gitignore, .gitattributes
- GitHub CI workflow and issue/PR templates

Fixes and docs:

- document lint_frames.py and peek_frame.mjs, previously absent from the file table
- add the lint stage to the pipeline order
- remove references to private client projects so the skill can ship publicly
- record the full upstream attribution chain for the style catalog
- package_skill.py now excludes repo-hygiene files and keeps licence files

2026-09-21 12:46:23 +08:00