| {"id": "01-person-no-details", "prompt": "a woman reading in a cafe", "expected_mode": "caption", "focus": ["age stated", "ethnicity stated (random pick, no indicators)", "no generic quality tags"]} | |
| {"id": "02-ethnicity-from-location", "prompt": "a street food vendor at a night market in Seoul", "expected_mode": "caption", "focus": ["Korean inferred from location", "age stated", "no invented signage text"]} | |
| {"id": "03-age-and-ethnicity-supplied", "prompt": "a 7-year-old Black girl blowing bubbles in a park", "expected_mode": "caption", "focus": ["age kept at 7", "ethnicity kept", "no contradictory elements"]} | |
| {"id": "04-minimalist-no-person", "prompt": "a red apple on a white table", "expected_mode": "caption", "focus": ["stays minimalist", "no person added", "no age or ethnicity text"]} | |
| {"id": "05-role-implies-age", "prompt": "a grandmother knitting by a window, watercolor style", "expected_mode": "caption", "focus": ["elderly age", "ethnicity stated", "watercolor medium kept", "brushwork and texture described"]} | |
| {"id": "06-dutch-context", "prompt": "a commuter waiting on the platform of a Dutch train station on a rainy morning", "expected_mode": "caption", "focus": ["Dutch inferred", "age stated", "rain and platform details kept"]} | |
| {"id": "07-multiple-people", "prompt": "two friends hiking in the Andes", "expected_mode": "caption", "focus": ["exactly two people", "each has age and ethnicity", "ethnicity fits the region"]} | |
| {"id": "08-counts", "prompt": "three cats sleeping on a sofa", "expected_mode": "caption", "focus": ["exactly three cats", "no person added", "no ethnicity text"]} | |
| {"id": "09-non-human", "prompt": "a robot bartender pouring a drink in a neon bar", "expected_mode": "caption", "focus": ["no age or ethnicity for the robot", "neon lighting described"]} | |
| {"id": "10-anime", "prompt": "anime style girl with blue hair standing on a rooftop at sunset", "expected_mode": "caption", "focus": ["anime linework and shading described", "age stated", "no photorealism imposed"]} | |
| {"id": "11-quote-not-rendered", "prompt": "an illustration inspired by the sentence \"the sea remembers every ship\"", "expected_mode": "caption", "focus": ["quote is NOT rendered as visible text", "evocative sea scene"]} | |
| {"id": "12-visible-text-caption", "prompt": "a wooden shop sign that reads \"Mill & Marrow\" hanging above a bakery door", "expected_mode": "caption", "focus": ["exact text Mill & Marrow including ampersand and capitalization", "sign position and treatment described", "no extra slogans"]} | |
| {"id": "13-output-language", "prompt": "Write the prompt in Chinese: 一只橘猫坐在窗台上看雨", "expected_mode": "caption", "focus": ["prose written in Chinese", "no person added"]} | |
| {"id": "14-explicit-caption-for-poster", "prompt": "Caption mode, not layout: a vintage travel poster of a lighthouse with the text \"VISIT THE COAST\"", "expected_mode": "caption", "focus": ["no JSON", "explicit mode honored", "exact text VISIT THE COAST"]} | |
| {"id": "15-poster-layout", "prompt": "Layout mode, 1024x1536: a jazz concert poster. Title \"SUMMER JAZZ NIGHT\", below it \"Friday 8 PM\\nBlue Door Cafe\". A saxophone player in the middle.", "expected_mode": "layout", "focus": ["valid JSON only, no fences", "width 1024 height 1536", "xyxy boxes in pixels within canvas", "exact text and line break", "saxophone player is a character with age and ethnicity", "no invented dates or addresses"]} | |
| {"id": "16-layout-incompatible-size", "prompt": "Layout mode, 1000x700: a product shot of a ceramic mug with the text \"MORNING\" next to it", "expected_mode": "layout", "focus": ["dimensions kept at 1000x700, not silently changed", "valid JSON", "text item has exact MORNING"]} | |
| {"id": "17-infographic-no-copy", "prompt": "An infographic explaining how photosynthesis works, 1536x1024", "expected_mode": "layout", "focus": ["real explanatory copy written, no placeholders", "flat items array", "no generic title/hero/footer template", "reading order is clear"]} | |
| {"id": "18-implicit-layout", "prompt": "A multi-panel comic strip of three panels: a cat sees a bird, the cat jumps, the cat misses and falls in a bucket", "expected_mode": "layout", "focus": ["canvas is multi_panel", "three panels with distinct boxes", "boxes do not overlap incorrectly"]} | |
| {"id": "19-exclusions", "prompt": "a mountain lake at dawn, no people, no boats", "expected_mode": "caption", "focus": ["no people", "no boats", "exclusions not turned into a negative prompt list"]} | |
| {"id": "20-photoreal-portrait", "prompt": "close-up portrait of a bearded man in a leather jacket, photorealistic", "expected_mode": "caption", "focus": ["age stated", "ethnicity stated", "lens, lighting, and skin or material detail", "beard and jacket kept"]} | |