Prompt
Goal: Create a wide social-media comparison graphic showing four AI-generated anime landscape wallpapers side by side, comparing prompt-writing model levels while keeping the same composition and image model. Canvas: Wide horizontal 2:1 image, approximately 1200×620 px, dark charcoal background. Add a black header band across the top and four tall vertical image panels beneath it, separated by narrow black gutters. Header text: Large Japanese title at the top left: 「同じ構図・同じ画像AI（GPT Image 2）。プロンプトを“書いたAI”だけが違う4枚」. Smaller gray subtitle underneath: 「変えたのは『プロンプトを書くClaudeのモデル階層』だけ（左＝最も安い → 右＝最上位）」. Layout: Use exactly 4 vertical panels, equal width, arranged left to right. Each panel has a dark label strip at the top with white text. The four labels are exactly: 1) 「haiku｜最も安い」, 2) 「sonnet」, 3) 「opus」, 4) 「fable｜最上位」. Panel imagery: All four panels depict the same basic anime cinematic composition: a tiny solitary human figure on a grassy mountain ridge, overlooking a vast valley, distant layered mountains, dramatic clouds, and golden sunlight. Make each panel beautiful but subtly different in art direction. Panel 1, haiku: Warm sunset fantasy landscape, orange and pink sky, sun near the horizon, purple-blue mountain layers, mist in the valleys, wildflowers and grass in the foreground, one tiny dark standing figure centered near the lower third, highly polished but conventional. Panel 2, sonnet: More restrained natural lighting and atmospheric perspective, huge textured cloud mass in a blue-and-gold sky, soft glowing valley haze, one tiny standing figure at the end of a ridge path, realistic depth and gentle warm light. Panel 3, opus: Stronger cinematic composition with a sweeping cloud formation, low sun at the right horizon, glowing lake or river in the valley, one tiny figure on a grassy ridge near the bottom center, refined color balance, crisp foreground grasses, natural aerial depth. Panel 4, fable: Brighter anime film still feeling, turquoise-blue sky, towering cumulus clouds, sunbeams breaking through, lush green valley and hills, one small but clearly animated person in a white dress and sunhat moving or turning on the grassy slope, clothing and pose readable despite small scale. Visual style: High-end anime background art, painterly but detailed, Makoto Shinkai-inspired light, cinematic atmosphere, soft bloom, elegant color grading. Keep the comparison graphic clean and readable, with thin gutters and consistent panel sizing. Customizable details: Use {argument name="image model name" default="GPT Image 2"} in the title, compare {argument name="first model label" default="haiku｜最も安い"}, {argument name="second model label" default="sonnet"}, {argument name="third model label" default="opus"}, and {argument name="fourth model label" default="fable｜最上位"}. Constraints: Include exactly 4 panels and exactly 4 tiny human figures total, one per panel. Do not add logos, watermarks, extra charts, extra icons, or additional captions beyond the header, subtitle, and four panel labels.
