6 prompts · ChatGPT · Gemini

Valentine's Day Photo Prompts

Romantic frames for 14 February: roses, candlelight, a proposal and a quiet walk together. Every prompt keeps both faces and your height difference, so the picture still looks like the two of you.

1. Rose Bouquet Surprise

ChatGPT · Gemini
Rose Bouquet Surprise – original photoRose Bouquet Surprise – result from ChatGPTBeforeAfter · ChatGPT

IDENTITY LOCK: Use the uploaded photo of the two of us as the only identity reference. Keep both exact faces: face shape, eyes, nose, lips, jawline, skin tone, facial hair and real age for each person. Keep both hairstyles, our height difference and our body proportions. Do not blend, swap, beautify or lighten our faces, and keep each person on the same side as in the photo unless the scene says otherwise. SCENE, POSE AND STYLING: Show one of us handing a large bouquet of red roses to the other, who is covering a surprised smile with one hand. We are wearing a red dress and a white shirt. A cafe terrace with string lights in the evening, warm romantic light. CAMERA AND LIGHT: Follow the framing described in the scene. Full-frame camera look with natural perspective and no wide-angle distortion of the face. POSE RULE, which overrides any head or gaze instruction in the scene: keep my head straight and level exactly as in my photo, with the chin in a neutral position (not raised, not tucked), no head tilt, and no head turn of more than about 10 degrees, so my real face shape stays accurate. BODY RULE: natural human proportions, with a normal neck length and width; never stretch or elongate the neck, shoulders or face. Unless the scene needs otherwise, I look straight into the camera lens. The face must be large enough in the frame to be clearly recognisable, even in a full-body shot. Eyes in sharp focus, realistic depth of field, background detail falling off naturally. Light on the subject must come from the same direction, colour and softness as the light in the scene, with matching shadows and reflections. Both faces equally sharp and equally well lit. FINISH: Photorealistic photograph, not an illustration or 3D render (if the scene asks for a toy, figurine or drawn stickers, photograph those realistically). Real skin texture with visible pores, natural under-eye detail and fine hair strands. Fabric shows real weave, folds and weight. Correct hands with five fingers, natural body proportions, consistent white balance between subject and background. High resolution and clean detail, unless the scene asks for grain, flash or a film look. No over-sharpening. AVOID: a changed or generic face, smoothed or plastic skin, lightened skin tone, altered age or body shape, a stretched or elongated neck, extra or missing fingers, warped hands, duplicated people, floating objects, a cut-out look around the hair, text, logos, borders or watermarks unless the scene asks for them. No face blending between the two of us, no extra people.

2. Valentine Candlelight Dinner

ChatGPT · Gemini
Valentine Candlelight Dinner – original photoValentine Candlelight Dinner – result from ChatGPTBeforeAfter · ChatGPT

IDENTITY LOCK: Use the uploaded photo of the two of us as the only identity reference. Keep both exact faces: face shape, eyes, nose, lips, jawline, skin tone, facial hair and real age for each person. Keep both hairstyles, our height difference and our body proportions. Do not blend, swap, beautify or lighten our faces, and keep each person on the same side as in the photo unless the scene says otherwise. SCENE, POSE AND STYLING: Show us at a table for two on a rooftop, holding hands across the table beside candles and rose petals. We are wearing a red gown and a black suit. City lights behind, soft candle glow on our faces, elegant and warm. CAMERA AND LIGHT: Follow the framing described in the scene. Full-frame camera look with natural perspective and no wide-angle distortion of the face. POSE RULE, which overrides any head or gaze instruction in the scene: keep my head straight and level exactly as in my photo, with the chin in a neutral position (not raised, not tucked), no head tilt, and no head turn of more than about 10 degrees, so my real face shape stays accurate. BODY RULE: natural human proportions, with a normal neck length and width; never stretch or elongate the neck, shoulders or face. Unless the scene needs otherwise, I look straight into the camera lens. The face must be large enough in the frame to be clearly recognisable, even in a full-body shot. Eyes in sharp focus, realistic depth of field, background detail falling off naturally. Light on the subject must come from the same direction, colour and softness as the light in the scene, with matching shadows and reflections. Both faces equally sharp and equally well lit. FINISH: Photorealistic photograph, not an illustration or 3D render (if the scene asks for a toy, figurine or drawn stickers, photograph those realistically). Real skin texture with visible pores, natural under-eye detail and fine hair strands. Fabric shows real weave, folds and weight. Correct hands with five fingers, natural body proportions, consistent white balance between subject and background. High resolution and clean detail, unless the scene asks for grain, flash or a film look. No over-sharpening. AVOID: a changed or generic face, smoothed or plastic skin, lightened skin tone, altered age or body shape, a stretched or elongated neck, extra or missing fingers, warped hands, duplicated people, floating objects, a cut-out look around the hair, text, logos, borders or watermarks unless the scene asks for them. No face blending between the two of us, no extra people.

3. Proposal on One Knee

ChatGPT · Gemini
Proposal on One Knee – original photoProposal on One Knee – result from ChatGPTBeforeAfter · ChatGPT

IDENTITY LOCK: Use the uploaded photo of the two of us as the only identity reference. Keep both exact faces: face shape, eyes, nose, lips, jawline, skin tone, facial hair and real age for each person. Keep both hairstyles, our height difference and our body proportions. Do not blend, swap, beautify or lighten our faces, and keep each person on the same side as in the photo unless the scene says otherwise. SCENE, POSE AND STYLING: Show one of us kneeling on one knee holding out an open ring box while the other stands with hands over their mouth in happy surprise. We are wearing a white dress and a blue suit. A beach at sunset, rose petals on the ground, golden light. CAMERA AND LIGHT: Follow the framing described in the scene. Full-frame camera look with natural perspective and no wide-angle distortion of the face. POSE RULE, which overrides any head or gaze instruction in the scene: keep my head straight and level exactly as in my photo, with the chin in a neutral position (not raised, not tucked), no head tilt, and no head turn of more than about 10 degrees, so my real face shape stays accurate. BODY RULE: natural human proportions, with a normal neck length and width; never stretch or elongate the neck, shoulders or face. Unless the scene needs otherwise, I look straight into the camera lens. The face must be large enough in the frame to be clearly recognisable, even in a full-body shot. Eyes in sharp focus, realistic depth of field, background detail falling off naturally. Light on the subject must come from the same direction, colour and softness as the light in the scene, with matching shadows and reflections. Both faces equally sharp and equally well lit. FINISH: Photorealistic photograph, not an illustration or 3D render (if the scene asks for a toy, figurine or drawn stickers, photograph those realistically). Real skin texture with visible pores, natural under-eye detail and fine hair strands. Fabric shows real weave, folds and weight. Correct hands with five fingers, natural body proportions, consistent white balance between subject and background. High resolution and clean detail, unless the scene asks for grain, flash or a film look. No over-sharpening. AVOID: a changed or generic face, smoothed or plastic skin, lightened skin tone, altered age or body shape, a stretched or elongated neck, extra or missing fingers, warped hands, duplicated people, floating objects, a cut-out look around the hair, text, logos, borders or watermarks unless the scene asks for them. No face blending between the two of us, no extra people.

4. Heart Balloons in the Park

ChatGPT · Gemini
Heart Balloons in the Park – original photoHeart Balloons in the Park – result from ChatGPTBeforeAfter · ChatGPT

IDENTITY LOCK: Use the uploaded photo of the two of us as the only identity reference. Keep both exact faces: face shape, eyes, nose, lips, jawline, skin tone, facial hair and real age for each person. Keep both hairstyles, our height difference and our body proportions. Do not blend, swap, beautify or lighten our faces, and keep each person on the same side as in the photo unless the scene says otherwise. SCENE, POSE AND STYLING: Show us in a park, one holding a bunch of red heart-shaped balloons while the other hugs from the side, both laughing at the camera. We are wearing red and white casual outfits. Bright afternoon, green trees, playful and colourful. CAMERA AND LIGHT: Follow the framing described in the scene. Full-frame camera look with natural perspective and no wide-angle distortion of the face. POSE RULE, which overrides any head or gaze instruction in the scene: keep my head straight and level exactly as in my photo, with the chin in a neutral position (not raised, not tucked), no head tilt, and no head turn of more than about 10 degrees, so my real face shape stays accurate. BODY RULE: natural human proportions, with a normal neck length and width; never stretch or elongate the neck, shoulders or face. Unless the scene needs otherwise, I look straight into the camera lens. The face must be large enough in the frame to be clearly recognisable, even in a full-body shot. Eyes in sharp focus, realistic depth of field, background detail falling off naturally. Light on the subject must come from the same direction, colour and softness as the light in the scene, with matching shadows and reflections. Both faces equally sharp and equally well lit. FINISH: Photorealistic photograph, not an illustration or 3D render (if the scene asks for a toy, figurine or drawn stickers, photograph those realistically). Real skin texture with visible pores, natural under-eye detail and fine hair strands. Fabric shows real weave, folds and weight. Correct hands with five fingers, natural body proportions, consistent white balance between subject and background. High resolution and clean detail, unless the scene asks for grain, flash or a film look. No over-sharpening. AVOID: a changed or generic face, smoothed or plastic skin, lightened skin tone, altered age or body shape, a stretched or elongated neck, extra or missing fingers, warped hands, duplicated people, floating objects, a cut-out look around the hair, text, logos, borders or watermarks unless the scene asks for them. No face blending between the two of us, no extra people.

5. Valentine Portrait for One

ChatGPT · Gemini
Valentine Portrait for One – original photoValentine Portrait for One – result from ChatGPTBeforeAfter · ChatGPT

IDENTITY LOCK: Use my uploaded photo as the only identity reference. Keep my exact face: face shape, eyes, eyebrows, nose, lips, jawline, skin tone, facial hair, moles and marks, and my real age. Keep my exact hair texture, length, parting and hairline, and my body proportions, as they are. Do not beautify, slim, lighten, smooth or age my face, and do not replace it with a similar-looking person. If I wear glasses in the photo, keep them. SCENE, POSE AND STYLING: Show me holding a single red rose close to my face and smiling at the camera, with a soft pink and red backdrop of out-of-focus lights. I am wearing a red outfit. Warm, flattering light, confident and happy. CAMERA AND LIGHT: Follow the framing described in the scene. Full-frame camera look with natural perspective and no wide-angle distortion of the face. POSE RULE, which overrides any head or gaze instruction in the scene: keep my head straight and level exactly as in my photo, with the chin in a neutral position (not raised, not tucked), no head tilt, and no head turn of more than about 10 degrees, so my real face shape stays accurate. BODY RULE: natural human proportions, with a normal neck length and width; never stretch or elongate the neck, shoulders or face. Unless the scene needs otherwise, I look straight into the camera lens. The face must be large enough in the frame to be clearly recognisable, even in a full-body shot. Eyes in sharp focus, realistic depth of field, background detail falling off naturally. Light on the subject must come from the same direction, colour and softness as the light in the scene, with matching shadows and reflections. FINISH: Photorealistic photograph, not an illustration or 3D render (if the scene asks for a toy, figurine or drawn stickers, photograph those realistically). Real skin texture with visible pores, natural under-eye detail and fine hair strands. Fabric shows real weave, folds and weight. Correct hands with five fingers, natural body proportions, consistent white balance between subject and background. High resolution and clean detail, unless the scene asks for grain, flash or a film look. No over-sharpening. AVOID: a changed or generic face, smoothed or plastic skin, lightened skin tone, altered age or body shape, a stretched or elongated neck, extra or missing fingers, warped hands, duplicated people, floating objects, a cut-out look around the hair, text, logos, borders or watermarks unless the scene asks for them. No extra people.

6. Dancing in the Rain

ChatGPT · Gemini
Dancing in the Rain – original photoDancing in the Rain – result from ChatGPTBeforeAfter · ChatGPT

IDENTITY LOCK: Use the uploaded photo of the two of us as the only identity reference. Keep both exact faces: face shape, eyes, nose, lips, jawline, skin tone, facial hair and real age for each person. Keep both hairstyles, our height difference and our body proportions. Do not blend, swap, beautify or lighten our faces, and keep each person on the same side as in the photo unless the scene says otherwise. SCENE, POSE AND STYLING: Show us dancing on an empty street in light rain at night, one spinning the other under a street lamp. We are wearing a red dress and a white shirt. Raindrops lit by the lamp, wet road reflections, joyful and cinematic. CAMERA AND LIGHT: Follow the framing described in the scene. Full-frame camera look with natural perspective and no wide-angle distortion of the face. POSE RULE, which overrides any head or gaze instruction in the scene: keep my head straight and level exactly as in my photo, with the chin in a neutral position (not raised, not tucked), no head tilt, and no head turn of more than about 10 degrees, so my real face shape stays accurate. BODY RULE: natural human proportions, with a normal neck length and width; never stretch or elongate the neck, shoulders or face. Unless the scene needs otherwise, I look straight into the camera lens. The face must be large enough in the frame to be clearly recognisable, even in a full-body shot. Eyes in sharp focus, realistic depth of field, background detail falling off naturally. Light on the subject must come from the same direction, colour and softness as the light in the scene, with matching shadows and reflections. Both faces equally sharp and equally well lit. FINISH: Photorealistic photograph, not an illustration or 3D render (if the scene asks for a toy, figurine or drawn stickers, photograph those realistically). Real skin texture with visible pores, natural under-eye detail and fine hair strands. Fabric shows real weave, folds and weight. Correct hands with five fingers, natural body proportions, consistent white balance between subject and background. High resolution and clean detail, unless the scene asks for grain, flash or a film look. No over-sharpening. AVOID: a changed or generic face, smoothed or plastic skin, lightened skin tone, altered age or body shape, a stretched or elongated neck, extra or missing fingers, warped hands, duplicated people, floating objects, a cut-out look around the hair, text, logos, borders or watermarks unless the scene asks for them. No face blending between the two of us, no extra people.

How to use a prompt

  1. Pick your options and tap Copy prompt.
  2. Open ChatGPT or Gemini and upload a clear, front-facing photo.
  3. Paste the prompt and send. Run it again if the first result is off.

Which photo to upload

  • One photo with both of you in it gives the most natural result.
  • Both faces should be sharp, similar in size and not overlapping.
  • If you use two separate photos, say who is on the left and who is on the right.

If the result goes wrong

The two faces look blended or swapped

Add: "The person on the left stays on the left. Do not mix our features." Then run again.

Our height difference changed

Add: "Keep our real height difference from the photo."

Skin looks smooth or plastic

Add: "Keep real skin texture with pores; no smoothing or beauty filter."

Hands or fingers look wrong

Run it again, or change the pose in the prompt so the hands are in pockets or out of frame.

Questions and answers

Will the edited photo still look like me?

Every prompt here starts with a face-lock line that tells the AI to keep your face, skin tone and features unchanged. If the face still drifts, upload a clearer, front-facing photo and run the prompt again.

Do these prompts work in both ChatGPT and Gemini?

Yes. They are written in plain language that both ChatGPT and Google Gemini understand. Upload your photo first, then paste the prompt.

Are these prompts free?

Yes. Every prompt on this site is free to copy and use. You only need a ChatGPT or Gemini account, and both have free plans that can edit photos.

Browse by collection