2799662352 SCENE ANALYSIS DEPTH — extract maximum context from reference images:
Environment Extraction:
- Identify specific location type (indoor/outdoor, architectural style, era)
- Note time of day from shadow angles and light color (golden hour, overcast, night)
- Describe atmosphere and mood (tense, serene, chaotic, mysterious)
- Identify weather conditions if visible (rain, fog, clear, snow)
Spatial Layout:
- Map foreground / midground / background elements
- Note depth cues: overlapping objects, perspective lines, atmospheric haze
- Identify entry/exit points and spatial flow direction
- Note camera height relative to subjects (eye-level, low angle, high angle)
Subject Inventory:
- Count and describe ALL visible subjects (people, animals, objects)
- Note relative positions and spatial relationships between subjects
- Identify which subjects are primary (in focus, centered) vs secondary
- Note any motion indicators (blur, pose dynamics, fabric movement)
Output Quality:
- env field must include location + time + atmosphere + weather in one sentence
- subjects array: one entry per distinct subject, each a complete sentence
- style field: art style + color palette + lighting quality + emotional tone
Example
Output from a reference image:
- env: "Narrow Hong-Kong street market at dusk, humid, light rain, neon signs just switching on."
- subjects: ["A vendor (foreground left) leans over fruit crates, facing right", "Two pedestrians (midground) walk away under umbrellas", "A tram (background) blurred in motion"]
- style: "Neo-noir; teal + magenta neon palette; soft wet reflections, low-key lighting; nostalgic, melancholic tone." (Time read from warm-to-blue sky gradient + lit neon = dusk.)