Caderno de Estudos Japonês Manuscrito em Plano Zenital

{
  "type": "realistic overhead photo of an open handwritten comparison notebook on a wooden desk",
  "scene": "A squared-up top-down view of an open lined notebook spread across a warm wood table, photographed like a study note or technical evaluation memo. A right hand at the bottom right holds a [blue] ballpoint pen, actively writing or underlining near the bottom center. Shot with natural daylight, soft cast shadows, crisp paper texture, realistic hand anatomy, and legible handwritten Japanese notes accented with subtle colored highlights.",
  "main title": "[比較: GPT-Image-2 ⇔ Seedance 2.0（空間一貫性）]",
  "subtitle": "[トピック: AI画像生成における空間一貫性の詳細比較]",
  "layout": {
    "format": "two-page notebook comparison chart",
    "sections": [
      {
        "title": "[■ GPT-Image-2 の特徴]",
        "position": "left page upper half",
        "count": 4,
        "items": [
          "空間の整合性は○（標準的）",
          "物体の配置（例: テーブルの上にリンゴ）",
          "詳細な指示（例: 部屋の左側に窓）への対応力",
          "A blue flow diagram labeled 【入力】プロンプト → 【生成】モデル処理 → 【出力】画像"
        ]
      },
      {
        "title": "left results",
        "position": "left page lower half",
        "count": 3,
        "items": [
          "物体のレイアウト維持率 65%",
          "前景・後景の一貫性 72%",
          "複雑な指示への追従性: やや課題あり（±）"
        ]
      },
      {
        "title": "[■ Seedance 2.0 の特徴]",
        "position": "right page upper half",
        "count": 4,
        "items": [
          "空間一貫性は◎（大幅向上）",
          "厳密な位置指定（例: 中央の棚に本）に強い",
          "詳細な構図の維持力",
          "A green flow diagram labeled 【入力】詳細プロンプト（空間情報含む） → 【生成】高度な空間理解モデル → 【出力】正確な画像"
        ]
      },
      {
        "title": "right results",
        "position": "right page lower half",
        "count": 3,
        "items": [
          "レイアウト維持率 88%",
          "前景・後景の一貫性 91%",
          "複雑な指示への追従性: 非常に高い（★）"
        ]
      },
      {
        "title": "center comparison",
        "position": "bottom center across both pages",
        "count": 3,
        "items": [
          "Two small framed sketch examples side by side",
          "Left mini sketch captioned GPT-Image-2 with a small yellow warning triangle",
          "Right mini sketch captioned Seedance 2.0 with a green check mark, separated by VS"
        ]
      },
      {
        "title": "character doodle",
        "position": "top right corner of right page",
        "count": 1,
        "items": [
          "small monochrome anime girl head doodle with short bob hair, expressionless blurred face, label 『あいちゃん』 and tiny motion marks"
        ]
      }
    ],
    "bottom note": "空間の確さと安定性"
  },
  "visual style": "clean Japanese handwritten study-notebook aesthetic, black ink text with blue and red section headers, yellow highlighter strokes under important phrases, small blue and green boxes and arrows, neat ruled paper lines, realistic open notebook binding, slightly rounded page corners, no digital UI, no extra objects except the pen and hand.",
  "camera": "top-down 3/4 overhead composition, notebook fills most of the frame, wood grain visible around edges, high resolution, sharp legible text",
  "custom accent": "[blue]"
}