GPT Image 2 vs. Nano Banana Pro Street Photo Prompt

GPT Image 2 vs. Nano Banana Pro. A modular prompt for street photos. AI came up with all the subjects and environments. input = [ stranger(s) / urban_context / unaware_action / capture_medium / juxtaposition ] run street_documentary_decisive_moment: subject_dna := unposed strangers unaware of the lens, caught mid-stride/mid-reaction, everyday clothing context_dna := infer city, weather, time (rush hour, dusk, rain), signage/traffic/architecture that dates it moment_state := the split-second where elements align + causal_hint (a passing bus, a splash, a glance) capture_dna := 28–35mm equiv, deep-ish dof, available light only, grain in the shadows, slight blur on a moving limb focal_anchor := one gesture or face sharp; the rest is context light_logic := hard directional daylight or mixed neon/streetlight at dusk, harsh shadows, reflections in glass and puddles, no fill imperfection := a tilted horizon, a foreground head clipping the edge, someone half-cut by the frame depth & frame := layered fore/mid/background telling a small story; 3:2 horizontal, 35mm frame juxtaposition := an unplanned visual rhyme or absurd coincidence between subject and background mood := candid, observational, real city energy render: one honest instant where timing and available light make the image. forbidden — objects: staged "urban model" poses, fake rain machines, everyone facing camera forbidden — look: clean studio light outdoors, symmetrical framing, glossy retouching, posed awareness