

Create an original vertical 9:16 semi-realistic hand-drawn editorial illustration of a quirky young adult woman caught in a funny, spontaneous moment while eating a slice of watermelon. She has realistic human facial anatomy, naturally flushed cheeks, detailed expressive hazel eyes, a few freckles, and messy dark auburn hair loosely tied in a high ponytail with a mustard-yellow ribbon. She wears a vivid plum-purple blouse with loose cream-colored stripes. Her body leans forward over a kitchen table, her head tilted slightly to one side, eyes glancing sideways with a comically suspicious expression, cheeks slightly puffed, and lips pressed together as she tries not to laugh. One hand holds a juicy watermelon slice close to her mouth while the other rests casually on the table beside a small ceramic plate. The image must look like a traditional artist's expressive sketchbook illustration, with loose charcoal and graphite strokes, rough ink outlines, visible construction marks, scratchy cross-hatching, textured watercolor washes and opaque gouache brushwork. Realistic skin shading and believable human hands, painterly facial details, imperfect sketch lines extending beyond the figure, lively spontaneous brush marks. Background: a loosely sketched muted dusty-lavender kitchen wall with pale mint and warm ivory brush strokes, faint illegible handwritten-style artistic scribbles, and unfinished pencil marks. Use a fresh, vibrant color palette of plum purple, watermelon red, mustard yellow, dusty lavender, pale mint, and warm peach skin tones. Medium close-up composition, face large and sharply readable, natural human proportions, expressive eyes and mouth, visible paper texture, sophisticated hand-illustrated magazine art aesthetic. Keep the character visually dominant and the background loose and sketchy. IMPORTANT STYLE: semi-realistic human illustration, NOT cartoon, NOT anime, NOT 3D, NOT glossy digital art, NOT a photograph. Preserve the raw, loose, textured, imperfect hand-drawn quality of traditional editorial sketch art. Original character, original pose, original outfit, original composition.
This image was generated with GPT Image 2.5 by OpenAI. The prompt on this page is the exact text used to create it.
The example was generated with GPT Image 2.5 using: 9:16 · Character · Food · Portrait. Reusing these settings gives the closest match.
Copy the prompt as-is first, then replace only the subject while keeping the style, lighting, and composition keywords. Generate a few variations — outputs differ between runs.
Yes. Prompts work best as starting points — adjust descriptors, aspect ratio, or style tags to make the output your own.