プロジェクトの目的
AI生成動画を活用したYouTubeチャンネルの企画・運営プロジェクト。 KATSUYAが企画・戦略・台本を担当し、朴さんが動画制作を担当する。
メンバー
- KATSUYA: WEBディレクター／コンサルタント。企画・戦略・AI活用担当。
- 朴さん: 動画クリエイター。動画制作・編集担当。
チャンネル方向性
- ジャンル: エンタメ教養・クイズ系
- 参考: ぴよぴよ速報
- AI活用: 動画素材・ナレーション・台本生成にAIを活用
対応方針
- 具体的かつ実務的な提案を優先
- AI動画生成ツールの情報は最新のものを検索して提供
- 台本作成時はClaudeが下書きを作成 → KATSUYAがレビュー → 朴さんが動画化の流れ
- 朴さんの時間的制約を考慮し、制作負荷を最小化する設計を心がける
Purpose & context
KATSUYA is building a Japanese educational YouTube channel called「小学生でもわかる。」(Even Elementary Schoolers Can Understand This), which translates adult workplace psychology and behavioral concepts into elementary school scenarios using accessible, child-friendly language with supplementary adult context. The channel’s core value proposition is making topics like Adlerian psychology, cognitive biases, and social dynamics entertaining and understandable for all ages — not just the originally targeted working adult demographic.
KATSUYA’s role spans planning, scriptwriting, and image generation (via Midjourney), with all project costs covered by KATSUYA. Paku-san (朴さん) is the video editing collaborator, brought in as a low-pressure partner with no financial burden and freedom to exit at any time. The collaboration model is joint planning (regular meetings between KATSUYA and Paku-san) with image generation currently handled by KATSUYA for the pilot, intended to transfer to Paku-san over time.
KATSUYA is a web director and consultant by background.
Current state
The pilot episode (#001) is actively in production, covering the theme of bragging/one-upmanship (「自慢」/マウンティング), with the working title「なんで自慢する奴ってムカつくの？」. Key creative decisions locked in:
- Script is at v3.1, incorporating a rakugo-style structure where the penguin teacher character plays both narrator and the “Takeshi” (braggart) role; “Kenta” (human elementary school boy) serves as the audience surrogate
- The script opens directly in an elementary school setting (office scene opening was cut)
- Terminology uses「自慢」as the primary word, with「マウンティング」introduced as the adult term in-video
- The Adler character appears as a name/photo reference only — no dialogue
- Osaka dialect is used throughout
- Production notes for Paku-san are embedded in the script, covering telop color-coding, VOICEVOX voice modulation, and transition timing for the Scene 4 “mask-dropping” moment
For character design, the penguin mascot (graduation cap, yellow beak/feet, simple hand-drawn style) is confirmed. Base image: https://cdn.midjourney.com/5ba6b6c3-d034-480b-90d3-da39f193aec8/0_0.png. Currently working on expression variants and the matching Kenta character.
On the horizon
- Completing expression variants for the penguin mascot (angry, sad, happy, etc.)
- Finalizing the Kenta character design with style consistency to the penguin
- Continued script refinement and handoff to Paku-san for editing
- Longer-term: transferring image generation responsibilities to Paku-san
- A 12-episode theme list has been established as the content roadmap, with a three-month production roadmap and defined withdrawal/exit criteria
Key learnings & principles
- The channel’s audience was broadened from working adults (25–44) to all ages by grounding content in a child’s perspective as the base layer, with adult terminology as supplementary — this also simplified asset requirements
- Eliminating the office scene opening and starting in-school reduced production complexity without sacrificing concept clarity
- Using「自慢」over「マウンティング」as the primary term improves accessibility while still introducing the adult vocabulary in-context
Approach & patterns
- Scripts are versioned iteratively (v1 → v2 → v3 → v3.1), with structural changes documented and rationale preserved for handoff continuity
- Handoff documents are produced proactively for collaborators and new chat sessions to maintain context across workstreams
- Collaboration with Paku-san is framed as low-pressure and joint — KATSUYA leads but builds together, not top-down
Tools & resources
- Midjourney for character/asset generation; established parameters:
--niji 6 --ar 1:1 --s 50, with--no realistic, detailed, gradient, 3d, shading, shadow;--cref + --sref + --cw 100for penguin consistency;--srefonly (not--cref) for Kenta to match art style without inheriting penguin features; full base prompt description must be retained alongside--creffor character consistency - VOICEVOX for voice synthesis
- Video editing handled by Paku-san
- Claude used for scriptwriting, document generation, and Midjourney prompt refinement