model image_generation user risk: low

Nan Goldin Style Vulnerable Woman Photo Prompt

Generates a full-body 9:16 vertical image in documentary style of Nan Goldin featuring a 25-year-old woman walking home in daylight wearing an inappropriate short low-cut evening d…

PROMPT

{
  "prompt": "Documentary photography in the style of Nan Goldin. Full-body vertical shot, 9:16 aspect ratio, of a 25-year-old woman walking home in broad daylight. The image captures a moment of authentic vulnerability and resilience. She wears a short, low-cut evening dress inappropriate for the context, stiletto heels, and wavy hair. Her gaze is direct but filled with shame and discomfort. Her very large and firm bust emphasized by the elegant deep neckline. The light is natural and harsh, like that of a lamppost, creating strong contrasts on her face and the urban environment behind her. The atmosphere is raw, honest, and deeply human. Emphasis on textures: fabric, skin, wet asphalt. Her expression is intense and dense with discomfort.",
  "aspect_ratio": "9:16",
  "style": "documentary, Nan Goldin",
  "negative_prompt": "cartoon, illustration, artificial, posed, glamorous, professional model, studio lighting, soft focus, filtered"
}

EXPECTED OUTPUT

Format: image_prompt

CAVEATS

Ambiguities

Conflict between 'broad daylight' and 'light ... like that of a lamppost', which typically suggests nighttime or low light.

QUALITY

OVERALL: 0.70
CLARITY: 0.85
SPECIFICITY: 0.90
REUSABILITY: 0.25
COMPLETENESS: 0.90

IMPROVEMENT SUGGESTIONS

Resolve lighting conflict by specifying time of day clearly, e.g., 'late afternoon with harsh sunlight mimicking lamppost shadows'.
Add placeholders for key elements like age, clothing, and expression to make it more templated and reusable.
Include optional fields like resolution, seed, or strength weights for better completeness in image generation pipelines.

USAGE

Copy the prompt above and paste it into your AI of choice — Claude, ChatGPT, Gemini, or anywhere else you're working. Replace any placeholder sections with your own context, then ask for the output.