GPT Image prompt

Use the uploaded portrait as the strict identity reference

Use the uploaded portrait as the strict identity reference is a free 349-word prompt written for GPT Image. It opens with “Use the uploaded portrait as the strict identity reference.” and works from a reference photo you upload, so the result keeps the subject you already have. The text below is the original, unedited version — no sign-up, no rewriting.

Prompt breakdown

What this specific prompt locks in, read directly from its text. Nothing here is a generic template — each value below comes from the words in the prompt.

Model: GPT Image
Opening instruction: “Use the uploaded portrait as the strict identity reference.”
Aspect ratio: 4:5
Reference photo: Required — the prompt edits an uploaded image.
Placeholders: None — the prompt is ready to paste as written.

What this prompt asks you to preserve

These clauses are what stop the edit from changing the subject. Keep them in the prompt when you customise it.

“Preserve the person’s face, facial structure, eyes, nose, lips, skin tone, hairstyle, hair color, age impression, gender presentation, and overall recognizable vibe.”
“Keep their original gender presentation and create a clean studio-photo look.”
“Keep it funny and relatable, not exaggerated.”

How to use this prompt

1Open GPT Image in the image tool you already use.
2Upload the reference photo first. This prompt opens with “Use the uploaded portrait as the strict identity reference.”, so the subject is expected to come from that image.
3Paste the prompt exactly as written. Do not summarise it: the ordering of the clauses is what keeps the composition stable.
4Generate once, then change a single phrase per attempt. Start with the 4:5 framing rather than rewriting the whole prompt.

What to change first

The framing is set to 4:5. Change that value to get a different crop without touching the rest of the prompt.
Keep the preservation clauses intact, starting with “Preserve the person’s face, facial structure, eyes, nose, lips, skin tone, hairstyle, hair color, age impression, gender presentation, and overall recognizable vibe.”. Those are what stop the edit from changing the subject.
The opening clause — “Use the uploaded portrait as the strict identity reference.” — fixes the subject; the rest is styling. Swap the subject and the style stays; swap the style words and the subject stays.

Notes for GPT Image

Back to categories All GPT Image prompts

GPT Image prompt

Use the uploaded portrait as the strict identity reference

Prompt breakdown

What this specific prompt locks in, read directly from its text. Nothing here is a generic template — each value below comes from the words in the prompt.

Model: GPT Image
Opening instruction: “Use the uploaded portrait as the strict identity reference.”
Aspect ratio: 4:5
Reference photo: Required — the prompt edits an uploaded image.
Placeholders: None — the prompt is ready to paste as written.

What this prompt asks you to preserve

These clauses are what stop the edit from changing the subject. Keep them in the prompt when you customise it.

“Preserve the person’s face, facial structure, eyes, nose, lips, skin tone, hairstyle, hair color, age impression, gender presentation, and overall recognizable vibe.”
“Keep their original gender presentation and create a clean studio-photo look.”
“Keep it funny and relatable, not exaggerated.”

How to use this prompt

1Open GPT Image in the image tool you already use.
2Upload the reference photo first. This prompt opens with “Use the uploaded portrait as the strict identity reference.”, so the subject is expected to come from that image.
3Paste the prompt exactly as written. Do not summarise it: the ordering of the clauses is what keeps the composition stable.
4Generate once, then change a single phrase per attempt. Start with the 4:5 framing rather than rewriting the whole prompt.

What to change first

The framing is set to 4:5. Change that value to get a different crop without touching the rest of the prompt.
Keep the preservation clauses intact, starting with “Preserve the person’s face, facial structure, eyes, nose, lips, skin tone, hairstyle, hair color, age impression, gender presentation, and overall recognizable vibe.”. Those are what stop the edit from changing the subject.
The opening clause — “Use the uploaded portrait as the strict identity reference.” — fixes the subject; the rest is styling. Swap the subject and the style stays; swap the style words and the subject stays.

Notes for GPT Image

More GPT Image prompts

Whimsical childrens storybook illustration in a cute kawaii cartoon

Subject and interaction Create a realistic Polaroid-style candid

Late-1990s summer travel editorial as a matched two-image diptych

Cinematic automotive billboard poster, Honda Civic in matt black

Back to categories All GPT Image prompts

Prompt

349 words

Use the uploaded portrait as the strict identity reference. Preserve the person’s face, facial structure, eyes, nose, lips, skin tone, hairstyle, hair color, age impression, gender presentation, and overall recognizable vibe. Create a high-quality surreal mixed-style portrait featuring: 1. The uploaded person as a realistic human subject 2. A mysterious simple black line-doodle figure physically teasing or bothering them in a funny way The real person must remain realistic, attractive, natural, and recognizable — not cartoon or anime. Keep their original gender presentation and create a clean studio-photo look. The doodle character should be a minimalist black hand-drawn stick/line figure: - bold black outlines only - simple head and body - tiny dot eyes or blank face - no identity, color, shading, clothing, or realistic details - flat graphic style contrasting with the realistic person Choose one random playful “bullying” interaction each time, such as: - gently grabbing arms, legs, wrists, or shoulders - dragging or lifting them awkwardly - wrapping around or trapping them - carrying them like an object - squeezing them into a silly pose - covering their mouth playfully - making them sit or lie awkwardly The interaction should feel mischievous, annoying, absurd, cute, and meme-like — never violent, scary, sexual, or harmful. The person’s reaction should naturally match: annoyed, confused, tired, embarrassed, surprised, unamused, helpless, or mildly frustrated. Keep it funny and relatable, not exaggerated. Choose a simple clean outfit that fits the person: casual clothing, sweatshirt, knitwear, T-shirt, loungewear, minimal fashion, or neat everyday style. Use a plain white, gray, or neutral studio background. Avoid clutter and distractions. Composition: - vertical 4:5 or 9: 16 - full-body or 3/4-body framing - show both characters clearly - doodle figure must physically interact with the person Style goals: realistic photography + flat black doodle art, strong contrast, surreal meme humor, clean polished social-media aesthetic, realistic anatomy, strong identity match. Negative prompt: gore, blood, injury, horror, abuse, sexual content, creepy monster, realistic creature, detailed doodle face, colored doodle, messy background, cartoon human, anime style, weak resemblance, extra limbs, bad anatomy, floating body, no contact, text, watermark.

More GPT Image prompts

Where to modify this prompt

The editor guide pinpoints the clauses that carry the subject, style and framing. Change these first.

Subject or product
Use the uploaded portrait as the strict identity reference
Replace it with your own subject, product or character, and keep the camera and lighting clauses unchanged.
Reference image
the uploaded portrait
Swap in your own image, or remove this clause if you want a text-only generation.
Aspect ratio
4:5
Change it to 4:5, 1:1 or 16:9 for the placement you need.
Camera and lens
full-body
Try 35mm, 50mm or 85mm, and keep one lens choice per prompt.
Style
anime
Change it to editorial, anime, 3D render or vintage, but keep one main style.