Flooded subway hero vs monster cinematic AI Image Prompt

Create a cinematic high-end photomanipulation scene in a flooded, derelict New York City subway station. Vertical 2:3 frame, mid-shot, low eye-level camera half submerged at waterline. In the foreground, show {argument name="hero description" default="a soaked male action hero with short wet black hair, dark tactical clothing, and tense body language"} wading chest-deep through murky green-black water, gripping a black assault rifle with both hands; the rifle has a long suppressor, a vertical magazine, tactical rails, and a bright white flashlight beam aimed toward the viewer. Place exactly three opaque face-obscuring rectangles: one large dark brown square covering the hero’s face, one large blurred brown square covering the creature’s face, and one smaller blurred brown square over the creature’s raised hand area. Behind the hero, slightly right of center and out of focus, show {argument name="pursuer creature" default="a bloated pale zombie-like humanoid woman with mottled diseased skin, exposed chest, stringy wet yellow-green hair, and one arm reaching forward"} emerging waist-deep from the floodwater, creating splashes and droplets. The setting should include exactly two subway trains, one partially visible along the left background edge and one silver train along the right side; exactly three major rust-stained support columns visible in depth; overhead fluorescent strip lights receding down the ceiling; one dark station sign reading “Eastbound” in the distant center-left; one smaller dark sign reading “Delancey Street” on a right column; and one glowing red exit sign in the background. Use {argument name="color grade" default="cold teal-green shadows with sickly fluorescent highlights and small red accents"}, heavy humidity, water droplets suspended in the air, reflections and ripples on the water surface, gritty rust, grime, peeling paint, cinematic suspense lighting, shallow depth of field, photorealistic horror-action style, ultra-detailed, dramatic contrast, no extra text, no watermark except a small subtle four-point sparkle icon at the bottom right.