# V1 Reference Analysis: `v2EN9GVEeMc`

Source: YouTube `https://www.youtube.com/watch?v=v2EN9GVEeMc`

Local research assets:
- Audio: `storage/research/v2EN9GVEeMc/audio.m4a`
- Video: `storage/research/v2EN9GVEeMc/video.mp4`
- Frame sheet: `storage/research/v2EN9GVEeMc/frames/contact_sheet.jpg`
- Working transcript: `storage/research/v2EN9GVEeMc/transcript.txt`

## What The Reference Does Well

- It behaves like a live blackboard lesson, not a slide recital.
- The teacher persona is a modern junior-high Chinese teacher: direct, exam-aware, sometimes conversational, and always oriented around what students should remember.
- The board is not decorative. It carries the thinking path: title, author, source, key vocabulary, sentence structure, rhetorical devices, and the main reading axis.
- The lesson starts from why the text matters, then goes into word-by-word reading. That makes the classical text feel approachable before details appear.
- The key analytical frame is stable: observation creates material; imagination turns material into `物外之趣`.

## V1 Direction

V1 should keep the existing `childhood-wonder-video` pipeline, but change the teaching experience:

- From `古典圖文朗讀` to `現代黑板講解`.
- From one `text` field to three layers:
  - `text`: original article line shown on the board.
  - `boardNotes`: short chalkboard anchors for vocabulary and concepts.
  - `narration`: teacher-style spoken explanation.
- From generic storybook prompts to classroom-compatible prompts with a modern teacher silhouette, green chalkboard, and the story scene.
- From reading the original classical text aloud to explaining it in student-facing Mandarin.

## Content Policy

The transcript is used only for structural analysis. V1 does not reproduce the reference teacher's script or exact long phrasing. The generated lesson uses public-domain source text and newly written explanations.

## Implementation Notes

Updated script:

- `apps/api/src/scripts/produceChildhoodWonderVideo.ts`

Key changes:

- Added a `TEACHER_ROLE` and `TEACHING_AXIS`.
- Expanded each segment with `title`, `focus`, `boardNotes`, and `narration`.
- Revised image prompts to use a modern classroom-teacher character direction without copying the exact real person.
- Reworked SVG slides into a chalkboard layout with white/yellow chalk hierarchy.
- TTS now uses `segment.narration` instead of only the classical sentence.

Expected effect:

- Longer and more useful narration.
- More exam/lesson-like visual organization.
- Better match to the reference video's blackboard teaching mode while preserving the existing V1 generation command and output folder.