Object-Oriented Prompt for Composition and Lighting Study
This is an extremely detailed, structured YAML-style prompt designed for the Nano Banana Pro AI, focusing on logical composition and visual flow. It uses object-oriented concepts like 'class definition' to logically define elements, ensuring the AI understands complex instructions like '12 placements' and an 'S-curve visual path' for generating high-contrast concept art. The goal is to move beyond random generation ('gacha') to deliberate visual 'architecture'.
Prompt
# Composition & Lighting Study Generator
# Project: "{argument name="Project Title" default="Zodiac Parade"}" - {argument name="Illustration Theme" default="Emotional Night Illustration"}
# Purpose: コントラストの高いコンセプトアートを作成し、全体の「流れ」と「配置」を確定させる
# ▼ 1. プロジェクトのゴール定義
project_goal:
type: "Atmospheric Concept Art / Lighting Study"
# ここで「S字の光の道」を視覚的な最優先事項としてAIに認識させる
visual_priority: "Establish a strong S-curve visual path using light."
canvas_configuration:
aspect_ratio: "9:16"
# 色塗りに惑わされないよう、あえてモノクロやセピアのスケッチとして出力指示
rendering_mode: "Monochrome or Sepia Sketch with Strong Lighting Bloom."
camera_angle: "High Angle (Cinematic overview)."
# ▼ 2. 視線誘導ロジック（ここが重要）
visual_guidance_logic:
# 視線のスタートとゴールを言語化する
path_of_light: "視線は『左下（少女）』から始まり、水面の灯籠を辿って『右上（奥）』へ抜ける。"
# 明暗のルールを決めることで、画面が散らかるのを防ぐ
contrast_strategy:
- "Darkest Areas: River surface (reflecting lights)."
- "Brightest Areas: Lanterns (Light Sources)."
# テキストを入れるための「左上」の余白を確保
negative_space: "Top-Left (Dark Sky) should be kept empty for text placement."
# ▼ 3. オブジェクトのクラス定義
scene_elements:
# --- 境界線 (Terrain) ---
- element: "Terrain_Divider"
visual: "Diagonal separation."
# ここで「陸」と「川」を明確に分けないと、キャラが水没したり浮いたりする
details: "Bottom-Left is the Snowy River Bank (Land). Top-Right is the flowing River (Water)."
# --- 人物 (Character) ---
- element: "Girl_Silhouette"
position: "Bottom_Left (On the snow)"
pose: "Crouching/Sitting, facing the river."
# 初期段階では「顔」よりも「着物のボリューム感（塊）」を重視
style: "Rough sketch style, showing the volume of the Kimono."
# --- 灯籠の配置 (Lanterns) ---
# ▼「12個」と一言で書くより、グループ分けしたほうがAIは数を守りやすい
- element: "Lantern_Group_Water (Past/Present)"
count: "7 lanterns (1 in hand + 6 on water)."
arrangement: "Forming a beautiful S-Curve flowing towards top-right."
- element: "Lantern_Group_Land (Future)"
count: "5 lanterns."
arrangement: "Clustered neatly, waiting to be released."
rendering_style:
- "Chiaroscuro (Strong contrast between light and dark)."
- "Focus on Composition balance and Visual Flow."
Open in Hitboard Original post