俯拍过街自拍中的近景相机广告

Complete prompt

{
  "type": "image_generation_prompt",
  "language": "en",
  "style": "hyper-realistic cinematic selfie photography",
  "aspect_ratio": "9:16",
  "identity_preservation": {
    "use_reference_image": true,
    "strict_identity_lock": true,
    "alter_face": false,
    "alter_skin": false,
    "alter_hair": false,
    "alter_gender": false,
    "notes": "Preserve identical facial features, skin texture, hair, glasses, age, and gender from the uploaded reference image. No synthetic skin or sculptural look."
  },
  "subject": {
    "gender": "female",
    "capture_method": "selfie taken by the subject herself",
    "pose": {
      "selfie_arm": {
        "description": "one arm fully straight and completely extended upward holding the camera that takes the selfie",
        "visibility": "arm clearly visible, straight and dominant in frame",
        "camera_visibility": "the selfie camera device itself must NOT be visible in the frame"
      },
      "product_arm": {
        "description": "the other arm fully extended toward the camera holding the attached Canon camera",
        "importance": "product is closest to the camera and visually dominant"
      },
      "head": {
        "tilt": "slightly tilted toward the selfie camera"
      },
      "expression": "natural and relaxed facial expression"
    },
    "body_visibility": "full body visible from head to toe",
    "feet": "feet clearly touching the road surface"
  },
  "composition": {
    "perspective": "natural selfie perspective at chest height",
    "camera_angle": "extreme top-down angle, camera above the subject looking directly downward",
    "layer_depth": [
      "product (closest to camera)",
      "face",
      "full body",
      "city environment (background)"
    ]
  },
  "scale_and_perspective": {
    "effect": "forced perspective",
    "subject_scale": "the woman appears extremely giant",
    "buildings_scale": "buildings appear much smaller, reaching no higher than her knees",
    "dominance": "the subject visually dominates the entire scene",
    "realism": "inspiring scale while remaining physically believable"
  },
  "environment": {
    "location": "real urban intersection",
    "elements": [
      "pedestrian crosswalk",
      "road markings",
      "traffic signs",
      "cars",
      "bicycles",
      "pedestrians at realistic human scale"
    ],
    "setting": "ground-level urban environment"
  },
  "lighting": {
    "type": "natural daylight",
    "conditions": "clear or lightly cloudy sky",
    "shadows": "soft and realistic",
    "restrictions": "no fantasy or dramatic lighting"
  },
  "product_rules": {
    "usage": "use the uploaded Canon product exactly as provided",
    "distortion": "none",
    "logo": "unchanged",
    "appearance": "natural reflections and realistic highlights only"
  },
  "camera_quality": {
    "realism": "maximum photorealism",
    "depth": "clear separation of foreground, subject, and background",
    "artifacts": "none"
  },
  "constraints": [
    "No AI-art look",
    "No plastic or sculpted skin",
    "No distortion of face or body",
    "No extra limbs or incorrect anatomy",
    "No text or watermarks",
    "No visible selfie camera device"
  ],
  "output_goal": "Create a hyper-realistic cinematic selfie image of a woman using her exact reference identity, captured from an extreme top-down perspective in a real urban crosswalk, with forced perspective scale, natural daylight, and a Canon camera product prominently held toward the lens."
}

Make it your own

Use your own clear reference image. The source output is not necessarily the input, and original references may be unavailable.

Preserve the product shape, label, text and materials when using a reference. Change the scene and lighting deliberately.

Structure: [Subject] + [Composition] + [Light] + [Material or style] + [Text or details to preserve] + [Aspect ratio]

Open the gallery workbench ↗

Sources & attribution

Prompt author

@saniaspeaks_

Original source

Original post / source page ↗

Source collection

NanoBanana Trending Prompts · MeiGen.ai

Model

GPT Image (specific version not disclosed) (source label; not a current-version claim)

Source date

2026-01-10 · Original-post date recorded by the dataset; metric capture time is unknown.

Snapshot

2026-10-20

Evidence

455 archived likes / 16,829 archived views.Dataset archive; capture time unknown and live values not independently verified.

Content rights

CC BY 4.0 as declared by the source; attribution, exceptions and third-party rights apply..Retain attribution and source links. Upstream exceptions and third-party rights apply.

Examples were not regenerated by this site. Results vary with tools, model versions, references and editing rounds.Sources & methods