提示词
Goal: Create a single-page Japanese manga-style instructional comic about using a 3×3 image grid with GPT and seedance to generate advertising videos, narrated by a lazy but wise capybara. Make it look like a polished social media explainer comic with warm lighting, detailed animal rendering, and hand-drawn panel borders.
Canvas: Vertical 4:5 page, about 768×960 px, cream paper background with thin black outer border. Use a warm sepia-brown palette, soft shadows, cinematic illustration, semi-realistic capybara fur, manga speech bubbles, and clean readable Japanese typography.
Main character: A large chubby capybara with tan-brown fur, sleepy half-closed eyes, relaxed smug expression, and small dark paws. The capybara acts like a laid-back advertising strategist. Optional name: {argument name="character name" default="lazy capybara strategist"}.
Layout: Use exactly 6 comic areas: 1 large hero panel at the top, 4 smaller instructional panels in a 2×2 grid underneath, and 1 wide closing panel at the bottom.
Hero panel: The capybara sits at a wooden desk in a dim cozy office, one paw under its chin, in front of an open silver laptop. On the desk include exactly 4 visible desk items: a white mug labeled 「やる気ゼロ。」, a black smartphone, a spiral notebook labeled 「今日もやらない」 with a tiny capybara doodle, and a brown leather satchel at the right. Add a dramatic white spiky manga burst on the left with bold text: {argument name="burst quote" default="これ、地味にすごいやつやん…！"}. On the right side of the hero panel place a pinned beige reference poster titled 「引用元ポスト(要約)」 and main headline 「GPT5.5 × seedance 架空広告の作り方」. The poster contains exactly 9 thumbnail images in a 3×3 grid: red sports car, sneaker, perfume bottle, burger, luxury wristwatch, resort pool, coffee cup, smartphone on desk, tropical suitcase/beach scene. Under the grid, show four short numbered Japanese steps explaining: create images in a 3×3 grid, feed them to GPT to convert into a prompt, combine images plus prompt to generate video, and create variations with #higgsfield and MCP.
Instruction panel 1: Top-left small panel with blue circle number 「1」 and title 「何が便利か」. Show capybara from the chest up, thoughtful and unimpressed. Include one speech bubble saying 「3×3をまとめてGPTに食わせるから、一貫性のあるプロンプトがサクッとできる。」 Add a small visual diagram at the bottom: a tiny 3×3 color grid arrow to a document icon labeled 「プロンプト(設計図)」. Add small caption 「1枚ずつ考えるより手戻り減るで。」
Instruction panel 2: Top-right small panel with blue circle number 「2」 and title 「初心者が真似するなら」. Capybara raises one finger like giving advice. Include a speech bubble: 「まずは3×3で好きなジャンルの画像を集めてみる。無料素材や自作でOK。」 On the right show exactly 3 vertically stacked icons with arrows: a 3×3 grid labeled 「集める」, a brain labeled 「プロンプト化(GPT)」, and a play button labeled 「動画生成(seedance)」.
Instruction panel 3: Bottom-left small panel with blue circle number 「3」 and title 「注意点」. Capybara holds up two fingers with a serious sleepy expression. Speech bubble: 「権利・商用利用はちゃんと確認やで。素材の利用規約と、生成物の使いみちは自己責任。」 Along the bottom include exactly 3 warning icons with labels: yellow warning triangle 「権利確認」, document 「規約チェック」, lock 「自己責任」.
Instruction panel 4: Bottom-right small panel with blue circle number 「4」 and title 「小さく試す方法」. Capybara leans on the panel edge, explaining calmly. Speech bubble: 「いきなり全額を狙わなくてええ。3×3→1本だけ出してみる。それを見て微調整。小さく回して学ぶのがいちばん早い。」 On the right draw a circular workflow with exactly 4 steps and arrows: 「作る(3×3)」, 「出す(1本)」, 「見る(反応)」, 「調整(改善)」.
Closing panel: Wide bottom panel. Capybara lies sprawled on the desk, eyes closed, extremely lazy, with tiny snore text 「ふぅ…」. Large text on the left: {argument name="closing line" default="今日もやらないけど、ここだけ試すならアリ。"}. Add a small thought bubble: 「動いたもん勝ち…ワイは昼寝や。」 On the right place a taped yellow sticky-note checklist titled 「まとめ」 with exactly 4 unchecked items: 「3×3で設計図を作る」, 「自動で量産できる」, 「小さく回して学ぶ」, 「今日もやらない(大事)」, plus a paw-print mark.
Visual style: Detailed semi-realistic capybara, manga/comic instructional layout, warm office lighting, soft beige shadows, expressive Japanese lettering, blue numbered badges, hand-drawn ink outlines, slight paper texture, humorous but professional social media explainer aesthetic.
Constraints: Keep all Japanese text legible and in the locations described. Use exactly 6 panels, exactly 9 poster thumbnails, exactly 4 desk items, exactly 3 icons in panel 2, exactly 3 warning icons in panel 3, and exactly 4 workflow steps in panel 4. No photorealistic humans, no extra panels, no watermark, no English replacement for the visible Japanese labels.
桃子AI画廊Tao Image Gallery