Explore. Create. Go viral.

[Subject] A photorealistic side-by-side comparison graphic featuring two highly similar close-up portraits of the same young adult woman seated in the back seat of a car. She has a feminine presentation, fair skin, long straight black hair, vivid green eyes, softly defined brows, and a polished pretty face with subtle makeup. Sunglasses rest on top of her head in both panels, acting as a casual off-duty styling cue. She wears a black leather jacket over a black top. In the left panel she also wears a chunky gold chain necklace and smiles more openly, showing teeth; in the right panel the expression is more restrained and softly confident, with lips closed or slightly parted and a more editorial calm. Both versions show natural glossy skin, neat hair, and shallow car-interior framing. Keep exactly two portrait panels side by side, representing comparison versions of the same styled woman rather than two clearly different people. [Environment] Both portraits are set inside a car back seat with soft daylight coming through windows and a blurred urban or leafy exterior beyond the glass. The overall image is not a pure portrait but a comparison asset with a dark teal or blue border/background surrounding two rounded-rectangle vertical image panels. At the bottom center there is white Spanish text reading “Prompt: Chica en el asiento de atrás de un coche con una chaqueta de cuero” or equivalent wording. At the bottom of the left panel appears the label “SEEDREAM 4” in bold warm yellow-orange type, and at the bottom of the right panel appears the label “NANO-BANANA” in the same color. Preserve the split-screen layout, panel divider, text overlays, and car context within both panels. [Composition/Camera] The full image is a vertical social-media comparison card with two rounded portrait frames placed left and right, separated by a narrow vertical gap. Each panel is a tight medium close-up to close-up portrait framing the subject from upper chest to above the head. The left panel crop is slightly tighter and more front-facing; the right panel is slightly looser and angled. The sunglasses sit near the top of each panel, the leather jacket occupies the lower half, and the labels sit near the bottom over the images. Keep the full comparison-card composition, not a single portrait. Lens feel inside each panel should resemble a 50mm to 85mm equivalent portrait lens with shallow background blur. [Lighting] Soft natural daylight enters through the car windows in both panels, creating clean skin highlights and a lifestyle-beauty look. The left panel is a touch brighter and warmer, while the right panel is slightly more neutral and controlled, but both stay realistic and soft. No dramatic studio lighting, no harsh flash, no colored effects. The overall card background outside the panels remains dark and unobtrusive so the portraits dominate. [Style/Rendering] Realistic AI portrait comparison graphic for social media, not hand-drawn illustration, not screenshot of a web UI. The image should feel like a polished creator post comparing two model outputs under the same prompt. Preserve beautiful but believable face rendering, black leather texture, rounded panel corners, crisp typography labels, and clean design hierarchy. The palette centers on black leather, dark teal framing, natural skin tones, green eyes, and warm orange-gold text labels. The finish should be slick, modern, and comparison-oriented. [Detail constraints] Do not collapse the layout into one image, do not remove the Spanish prompt text, do not remove the model labels, do not replace the leather jacket, do not remove the sunglasses, and do not change the car interior into a different location. Keep the exact two-panel structure, maintain the small styling differences between the left and right portraits, and preserve the social-media comparison-card format precisely. Negative prompt: single portrait only, no split screen, no model labels, no prompt text, different outfits between panels, outdoor full-body shot, studio backdrop, no sunglasses, blonde hair, different eye color, low-resolution UI screenshot, visible website chrome, illustration, comic style, CGI render, random background clutter, no rounded panel corners, unreadable text, bright neon frame, four-panel layout, extra icons or buttons, missing leather jacket, missing car seat context. Suggested parameters: aspect ratio 4:5 vertical or similar portrait social-card ratio; each panel should feel like 50mm to 85mm portrait lens rendering; shallow background blur inside each portrait panel; 30 to 40 sampling steps; CFG or style strength 5.5 to 7.0 for realistic portrait models; sampler DPM++ 2M Karras or equivalent; seed suggestion 620571 or nearby values to preserve two similar-but-different outputs and text placement. Delta prompt strategy: 1) If the comparison layout collapses, append: two rounded-rectangle portrait panels side by side on a dark teal comparison card. 2) If the text disappears, append: white Spanish prompt text at bottom center and bold orange-yellow labels SEEDREAM 4 on left, NANO-BANANA on right. 3) If the woman looks too different between panels, append: same woman concept in both panels with slight variation in expression and styling, comparison format. 4) If the leather jacket changes, append: black leather jacket over black top in both portraits. 5) If the sunglasses vanish, append: dark sunglasses resting on top of the head in both panels. 6) If the car setting is lost, append: back seat of a car with soft daylight and blurred outside view through the window. 7) If the left/right mood becomes identical, append: left panel brighter and friendlier, right panel slightly calmer and more editorial. 8) If the image looks like a website screenshot, append: clean standalone social comparison card with no browser UI. 9) If the labels become unreadable, append: crisp bold model names in warm yellow-orange type near the bottom of each panel. 10) If panel corners or card styling drift, append: smooth rounded portrait frames with narrow central divider and dark surrounding border.