Text to Image

Reference-Based Photo Editing for Candid Night Scene - Nano Banana Pro AI Prompt for Social Media Post — AI Image Prompt

An extremely detailed, structured JSON prompt for Nano Banana Pro instructing it to replicate the exact appearance of a person from a reference image and place them in a specific candid, low-quality, night-time café scene, focusing on realism and imperfect lighting. - AIPinMaker

Nano Banana 2Style TransferText to Image
Reference-Based Photo Editing for Candid Night Scene - Nano Banana Pro AI Prompt for Social Media Post
Text to Image

Prompt

{
  "task": "photo_editing",
  "instruction": "Replicate the exact appearance of the person from the reference image—including their face, hair, and the specific clothing they are wearing. Place this person into the pose and setting described below. The subject should be performing the actions (sitting, using phone, holding drink) while maintaining their visual identity and attire from the reference photo.",
  "photo_style": {
    "type": "candid_smartphone_photo",
    "aspect_ratio": "3:4",
    "time": "night",
    "device": "low-end Android phone",
    "shooting_style": "handheld, rushed, imperfect",
    "quality": "realistic, unpolished"
  },
  "subject": {
    "appearance_reference": "EXACT match of the person's face, features, and hair from reference image",
    "vibe": [
      "girl-next-door",
      "young",
      "fresh",
      "everyday",
      "not model-like",
      "not glamorous",
      "not influencer",
      "not styled"
    ]
  },
  "pose_and_action": {
    "seating": "sitting casually at a café table",
    "moment": "unplanned, unposed",
    "hands": {
      "one_hand": "holding a glass of drink",
      "other_hand": "using phone"
    },
    "attention": "focused on her phone, unaware of camera (candid)"
  },
  "clothing_and_appearance": {
    "attire": {
      "requirement": "MUST be an exact replica of the clothing (style, color, material) worn in the reference image",
      "fit": "retained from reference but adapted to the sitting pose"
    },
    "hair": {
      "style": "matches reference image",
      "detail": "flowing naturally in the night breeze"
    },
    "skin": {
      "look": "matches reference image",
      "restrictions": [
        "no beauty filter",
        "no retouching",
        "no AI-polished skin"
      ]
    }
  },
  "camera_angle_and_framing": {
    "angle": "slightly from the side and slightly behind",
    "perspective": "authentic candid",
    "framing": "medium-wide",
    "visibility": "upper body and part of table visible"
  },
  "lighting": {
    "type": "harsh and imperfect",
    "sources": [
      "rooftop café lights",
      "city street lights"
    ],
    "effects": [
      "uneven exposure",
      "blown highlights",
      "deep shadows",
      "imperfect white balance"
    ]
  },
  "focus_and_quality": {
    "focus": "slightly more on subject",
    "background_blur": "soft blur due to low light and camera limitations",
    "artifacts": [
      "visible digital noise",
      "grain",
      "slight motion blur",
      "imperfect focus accuracy"
    ]
  },
  "background":3f:T

Prompt breakdown

Subject
exact match of reference person's face, hair, and clothing as a young everyday girl-next-door type sitting at a café table
Style
candid smartphone photo, 3:4 aspect, low-end Android, handheld rushed imperfect, realistic unpolished quality
Lighting
harsh imperfect sources from rooftop café lights and city street lights creating uneven exposure, blown highlights, deep shadows, and imperfect white balance
Composition
slightly from the side and behind, medium-wide framing with upper body and part of table visible, authentic candid perspective
Mood
unplanned unposed moment with subject focused on phone and unaware of camera

Remix ideas

  • change the glass to a different colored drink while keeping the exact hand position
  • increase motion blur on the phone-holding hand to emphasize the rushed capture
  • shift the angle a few degrees more behind the subject to show less of the reference face

Reference images

Reference-Based Photo Editing for Candid Night Scene - Nano Banana Pro AI Prompt for Social Media Post reference
Text to Image

How to use this AI Image prompt template

  1. AiVideo Maker stepOne1
    Copy the prompt — grab this template’s prompt and negative prompt.
  2. iVideo Maker stepTwo2
    Pick a model — choose a recommended AI model for the best match.
  3. AiVideo Maker stepThree3
    Generate — open the studio with one click and create your result.

Related templates

Image-to-Image Clothing Swap Prompt
GPT Image 2

Image-to-Image Clothing Swap Prompt

Create a creative image of Image To Image Clothing Swap Prompt. Style: photorealistic. Composition: balanced and well-framed. Lighting: natural with cinematic mood. Category: photography. Reference: image-to-image-clothing-swap-prompt-2563.

inpaintphotorealisticcontrolnet
Text to Image
Realistic LEGO Food Miniature
GPT Image 2

Realistic LEGO Food Miniature

Transform only the food items (target object) in the image into tiny LEGO-style miniature sculptures (style) while keeping everything else completely unchanged and realistic. Preserve the original composition, lighting, camera angle, perspective, background, hands, table, people, utensils, and environment exactly as they are. The food should appear rebuilt from authentic LEGO bricks and pieces: visible studs, blocky geometry, plastic texture, tiny LEGO tiles, rounded brick edges, and miniature construction details. Keep the food instantly recognizable while translating it into a highly detailed LEGO miniature aesthetic. The LEGO food should match the original food’s colors, proportions, toppings, and arrangement. Maintain realistic shadows, reflections, depth of field, and interaction with the environment so the LEGO food feels naturally placed inside the real photograph. The contrast between the realistic world and the giant LEGO food sculptures should feel playful, surreal, and highly believable. Add 3-4 big emojis (extras) to the picture.39

foodphotorealisticinpaint
Text to Image
Mixed-Reality Comic Transformation
GPT Image 2

Mixed-Reality Comic Transformation

Convert the subject in this image into a 4:5 mixed-reality transformation with a fully realistic background and a realistically rendered face, combined with bold cartoon apparel (style), hand-drawn comic fire effects, and playful cartoon companions. Use a live-action hybrid composition with the subject preserved as the clear identity anchor while the stylized elements read as designed overlays inside one coherent scene. CHARACTER: Preserve the subject’s recognizable facial identity, hair state, and facial hair pattern if presented exactly, then re-render the face with realistic live-action AAA structure, natural skin response, and clean hybrid integration. Ensure high brightness, low contrast, medium saturation, low sharpness details. Dress the figure in exaggerated cartoon clothing with simplified seams, and graphic trims that feel intentionally designed rather than photographed. Add funny cartoon characters as supporting visual accents with expressive silhouettes, rounded forms, and lively attitude, while keeping the subject as the primary focal figure. The body should remain human and believable, but the clothing and accessories should carry comic exaggeration and strong graphic personality. FACE & ANATOMY: Keep the subject’s core facial proportions, expression logic, and likeness intact, rendered with realistic but controlled hybrid facial detail. Use natural eye depth, believable skin, while avoiding any plastic toy look or full caricature distortion. The body should read as live-action proportioned, with the stylization concentrated in the outfit, props, and surrounding comic elements. ENVIRONMENT: Restyle the exact same environment as a realistic setting with clear depth and natural perspective. Build the scene as a mixed-reality stage where comic fire drawings, playful animated accents, and graphic energy bursts sit visibly inside the real space without breaking spatial coherence. Keep the background believable, structured, and physically present, with the stylized elements integrated as deliberate designed interventions. RENDER PIPELINE: Use a mixed-reality rendering pipeline that combines realistic plate-like environment treatment with stylized 2D graphic overlays and hybrid character compositing. Preserve live-action shading on the face and background, then layer cartoon clothing, comic effect shapes, and illustrated fire forms with crisp contour control and clean edge separation. The final finish must feel like a polished real-world composite with intentional animated design language, not a pure cartoon render and not a standard photograph. COLOR & LIGHT: Use clean low contrast, and bright comic-highlight behavior against a realistic ambient base. Shadows remain natural in the background and face, while the stylized garments and effects carry punchy neon-like energy and high chroma separation.3a:T

inpaintphotorealisticprofile-pic
Text to Image
Convert image to night scene with warm lights
GPT Image 2

Convert image to night scene with warm lights

Create a cinematic image of Convert Image To Night Scene With Warm Lights. Style: photorealistic. Composition: balanced and well-framed. Lighting: natural with cinematic mood. Category: cinematic-video. Reference: convert-image-to-night-scene-with-warm-lights-2389.

moodylow-keyphotorealistic
Text to Image
Professional Image Redesign
GPT Image 2

Professional Image Redesign

Create a creative image of Professional Image Redesign. Style: photorealistic. Composition: balanced and well-framed. Lighting: natural with cinematic mood. Category: photography. Reference: professional-image-redesign-18366.

photorealisticultra-detailinpaint
Text to Image
Image Manipulation Prompt for Advertising Placement
Nano Banana 2

Image Manipulation Prompt for Advertising Placement

Create a creative image of Image Manipulation Prompt For Advertising Placement. Style: photorealistic. Composition: balanced and well-framed. Lighting: natural with cinematic mood. Category: photography. Reference: image-manipulation-prompt-for-advertising-placement-9151.

inpaintad-bannerphotorealistic
Text to Image

Explore more prompts

Browse more AI image and video prompts by category.

FAQ

How does the prompt keep the clothing identical to the reference?
It requires an exact replica of the style, color, and material from the reference image, only adapting the fit for the seated pose.
Why add visible digital noise and imperfect focus?
These artifacts replicate the limitations of a low-end Android phone shooting at night with handheld movement and poor lighting.