Gemini AI photo prompts you can copy for trending looks

Woman at a rainy cafe window seat holding a ceramic cup in both hands
Man in a dark coat standing on a wooden jetty in thick dawn mist
Woman on a rolling library ladder pulling a book from a high shelf

Prompt hub7 prompt pages

The short answer

A Gemini AI photo prompt edits your own uploaded photo while keeping your face: it holds a keep line, then wardrobe, setting, light, camera angle and aspect ratio. You'll paste one of the 12 full prompts on this page into Gemini with one clear, front-facing photo, and Nano Banana, the Gemini image model that applies the edit, does the rest.

12 Gemini photo prompts to copy

12 to try first. Switch the model once. Every card follows.

Show every prompt for

Woman at a rainy cafe window seat holding a ceramic cup in both hands

Rainy cafe window portrait

Two light colours, one cool and one warm, give the frame depth without any editing effects.

Attach your photo first. “Rainy cafe window portrait” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic portrait of me sitting at a cafe window seat on a rainy afternoon, both hands around a ceramic cup, looking out through the glass. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: a narrow cafe with wooden tables, condensation on the glass and blurred umbrellas outside. Light: soft cool window light from the left with a warm lamp glow behind. Camera: portrait lens, shallow depth of field, eye level, shallow depth of field. Mood: quiet and unhurried. Composition: waist-up, subject on the right third. Aspect ratio 4:5. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic portrait of me sitting at a cafe window seat on a rainy afternoon, both hands around a ceramic cup, looking out through the glass. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: a narrow cafe with wooden tables, condensation on the glass and blurred umbrellas outside. Light: soft cool window light from the left with a warm lamp glow behind. Camera: portrait lens, shallow depth of field, eye level, shallow depth of field. Mood: quiet and unhurried. Composition: waist-up, subject on the right third. Output at 4:5. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic portrait of the person in my attached photo, sitting at a cafe window seat on a rainy afternoon, both hands around a ceramic cup, looking out through the glass. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: a narrow cafe with wooden tables, condensation on the glass and blurred umbrellas outside. Light: soft cool window light from the left with a warm lamp glow behind. Camera: portrait lens, shallow depth of field, eye level, shallow depth of field. Mood: quiet and unhurried. Composition: waist-up, subject on the right third. Aspect ratio 4:5. No text, letters, numbers, logos or watermarks.

How to use
  1. Copy it with the button above.
  2. Fill the blanks in {{double braces}} with your own details.
  3. Paste it under your photo in your image AI and generate.
  4. Then ask for one change at a time, not a new prompt.

Before you run “Rainy cafe window portrait” for the first time, the lesson on how to use photo edit prompts walks through copying, filling and sending it.

Golden hour rooftop portrait

Naming the light direction stops the model from adding a flat, front-lit glow.

Attach your photo first. “Golden hour rooftop portrait” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic portrait of me standing on a city rooftop terrace an hour before sunset, leaning on the railing and glancing back at the camera. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: a low skyline with unlit string lights and potted plants. Light: low golden sun from behind-left creating a warm rim light, soft sky fill on the face. Camera: standard lens, shallow depth of field, slightly low angle. Mood: easy and optimistic. Composition: three-quarter body, horizon on the upper third. Aspect ratio 4:5. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic portrait of me standing on a city rooftop terrace an hour before sunset, leaning on the railing and glancing back at the camera. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: a low skyline with unlit string lights and potted plants. Light: low golden sun from behind-left creating a warm rim light, soft sky fill on the face. Camera: standard lens, shallow depth of field, slightly low angle. Mood: easy and optimistic. Composition: three-quarter body, horizon on the upper third. Output at 4:5. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic portrait of the person in my attached photo, standing on a city rooftop terrace an hour before sunset, leaning on the railing and glancing back at the camera. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: a low skyline with unlit string lights and potted plants. Light: low golden sun from behind-left creating a warm rim light, soft sky fill on the face. Camera: standard lens, shallow depth of field, slightly low angle. Mood: easy and optimistic. Composition: three-quarter body, horizon on the upper third. Aspect ratio 4:5. No text, letters, numbers, logos or watermarks.

Four-frame photo booth strip

Listing four distinct expressions gives variety instead of four copies of one smile.

Attach your photo first. “Four-frame photo booth strip” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic vertical four-frame photo booth strip of me in four frames: serious, laughing, mid-sentence, then looking away. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: a curtained booth with a plain pale wall. Light: direct on-camera flash, slightly harsh with a soft falloff. Camera: wide-angle fixed booth camera, close crop. Mood: playful and imperfect. Composition: four stacked frames with thin white borders. Aspect ratio 9:16. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic vertical four-frame photo booth strip of me in four frames: serious, laughing, mid-sentence, then looking away. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: a curtained booth with a plain pale wall. Light: direct on-camera flash, slightly harsh with a soft falloff. Camera: wide-angle fixed booth camera, close crop. Mood: playful and imperfect. Composition: four stacked frames with thin white borders. Output at 9:16. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic vertical four-frame photo booth strip of the person in my attached photo, in four frames: serious, laughing, mid-sentence, then looking away. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: a curtained booth with a plain pale wall. Light: direct on-camera flash, slightly harsh with a soft falloff. Camera: wide-angle fixed booth camera, close crop. Mood: playful and imperfect. Composition: four stacked frames with thin white borders. Aspect ratio 9:16. No text, letters, numbers, logos or watermarks.

Magazine cover portrait with a made-up title

A fictional masthead keeps the look of a cover without copying a real publication.

Attach your photo first. “Magazine cover portrait with a made-up title” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic magazine cover portrait of me looking straight at the camera with a calm half-smile, one shoulder slightly forward. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: a plain warm grey studio backdrop. Light: broad soft light from the upper left, subtle fill on the right, clean catchlights. Camera: portrait lens, moderate depth of field, chest-up, tack-sharp eyes. Mood: confident and understated. Composition: space above the head for a masthead; the masthead reads HARBOUR in a tall serif and there is no other text. Aspect ratio 3:4. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic magazine cover portrait of me looking straight at the camera with a calm half-smile, one shoulder slightly forward. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: a plain warm grey studio backdrop. Light: broad soft light from the upper left, subtle fill on the right, clean catchlights. Camera: portrait lens, moderate depth of field, chest-up, tack-sharp eyes. Mood: confident and understated. Composition: space above the head for a masthead; the masthead reads HARBOUR in a tall serif and there is no other text. Output at 3:4.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic magazine cover portrait of the person in my attached photo, looking straight at the camera with a calm half-smile, one shoulder slightly forward. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: a plain warm grey studio backdrop. Light: broad soft light from the upper left, subtle fill on the right, clean catchlights. Camera: portrait lens, moderate depth of field, chest-up, tack-sharp eyes. Mood: confident and understated. Composition: space above the head for a masthead; the masthead reads HARBOUR in a tall serif and there is no other text. Aspect ratio 3:4.

Neon street at night, caught mid-step

Practical lights only prevents the unnatural fill that makes night shots look composited.

Attach your photo first. “Neon street at night, caught mid-step” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic cinematic street portrait of me walking along a wet city street at night, caught mid-step and looking slightly off camera. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: a narrow street with shopfronts, steam from a vent and reflections in puddles. Light: pink and teal neon signs with practical lights only, reflected on the wet pavement. Camera: wide-angle lens, shallow depth of field, slight motion blur in the background, sharp subject. Mood: cinematic and a little lonely. Composition: full body in the left third with leading lines from the road. Aspect ratio 16:9. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic cinematic street portrait of me walking along a wet city street at night, caught mid-step and looking slightly off camera. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: a narrow street with shopfronts, steam from a vent and reflections in puddles. Light: pink and teal neon signs with practical lights only, reflected on the wet pavement. Camera: wide-angle lens, shallow depth of field, slight motion blur in the background, sharp subject. Mood: cinematic and a little lonely. Composition: full body in the left third with leading lines from the road. Output at 16:9. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic cinematic street portrait of the person in my attached photo, walking along a wet city street at night, caught mid-step and looking slightly off camera. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: a narrow street with shopfronts, steam from a vent and reflections in puddles. Light: pink and teal neon signs with practical lights only, reflected on the wet pavement. Camera: wide-angle lens, shallow depth of field, slight motion blur in the background, sharp subject. Mood: cinematic and a little lonely. Composition: full body in the left third with leading lines from the road. Aspect ratio 16:9. No text, letters, numbers, logos or watermarks.

Autumn park bench reading portrait

A candid action (looking up from a book) reads more natural than a posed smile.

Attach your photo first. “Autumn park bench reading portrait” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic portrait of me sitting on a park bench with fallen leaves around, reading a paperback and looking up as if interrupted. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: a path lined with maples, leaves mid-fall, a few out-of-focus walkers far behind. Light: soft overcast light with warm bounce from the orange leaves. Camera: portrait lens, shallow depth of field, slightly side-on. Mood: reflective and cosy. Composition: seated, three-quarter view, leaves in the foreground. Aspect ratio 3:4. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic portrait of me sitting on a park bench with fallen leaves around, reading a paperback and looking up as if interrupted. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: a path lined with maples, leaves mid-fall, a few out-of-focus walkers far behind. Light: soft overcast light with warm bounce from the orange leaves. Camera: portrait lens, shallow depth of field, slightly side-on. Mood: reflective and cosy. Composition: seated, three-quarter view, leaves in the foreground. Output at 3:4. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic portrait of the person in my attached photo, sitting on a park bench with fallen leaves around, reading a paperback and looking up as if interrupted. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: a path lined with maples, leaves mid-fall, a few out-of-focus walkers far behind. Light: soft overcast light with warm bounce from the orange leaves. Camera: portrait lens, shallow depth of field, slightly side-on. Mood: reflective and cosy. Composition: seated, three-quarter view, leaves in the foreground. Aspect ratio 3:4. No text, letters, numbers, logos or watermarks.

Woman on a rolling library ladder pulling a book from a high shelf

Library ladder portrait

A low angle with strong vertical lines makes the space feel grand and the subject small and at ease.

Attach your photo first. “Library ladder portrait” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic portrait of me standing on a rolling library ladder, pulling a book from a high shelf and smiling down. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: an old wood-panelled library with tall shelves and green-shaded desk lamps. Light: warm tungsten lamps with cool window fill from the right. Camera: wide-angle lens, shallow depth of field, from slightly below. Mood: curious and cosy. Composition: full height, shelves leading toward the vanishing point. Aspect ratio 3:4. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic portrait of me standing on a rolling library ladder, pulling a book from a high shelf and smiling down. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: an old wood-panelled library with tall shelves and green-shaded desk lamps. Light: warm tungsten lamps with cool window fill from the right. Camera: wide-angle lens, shallow depth of field, from slightly below. Mood: curious and cosy. Composition: full height, shelves leading toward the vanishing point. Output at 3:4. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic portrait of the person in my attached photo, standing on a rolling library ladder, pulling a book from a high shelf and smiling down. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: an old wood-panelled library with tall shelves and green-shaded desk lamps. Light: warm tungsten lamps with cool window fill from the right. Camera: wide-angle lens, shallow depth of field, from slightly below. Mood: curious and cosy. Composition: full height, shelves leading toward the vanishing point. Aspect ratio 3:4. No text, letters, numbers, logos or watermarks.

Man in a dark coat standing on a wooden jetty in thick dawn mist

Harbour jetty in morning mist

Fog does the background work, so the model has less to invent and the face stays accurate.

Attach your photo first. “Harbour jetty in morning mist” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic portrait of me standing on a wooden jetty at dawn in thick mist, hands in pockets, looking out at the water. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: moored fishing boats fading into fog, rope coils and wet planks. Light: diffuse pale light, low contrast, a faint warm glow on the horizon. Camera: portrait lens, shallow depth of field with long-lens compression. Mood: still and calm. Composition: wide shot, subject small on the left, jetty leading to the right. Aspect ratio 4:5. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic portrait of me standing on a wooden jetty at dawn in thick mist, hands in pockets, looking out at the water. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: moored fishing boats fading into fog, rope coils and wet planks. Light: diffuse pale light, low contrast, a faint warm glow on the horizon. Camera: portrait lens, shallow depth of field with long-lens compression. Mood: still and calm. Composition: wide shot, subject small on the left, jetty leading to the right. Output at 4:5. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic portrait of the person in my attached photo, standing on a wooden jetty at dawn in thick mist, hands in pockets, looking out at the water. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: moored fishing boats fading into fog, rope coils and wet planks. Light: diffuse pale light, low contrast, a faint warm glow on the horizon. Camera: portrait lens, shallow depth of field with long-lens compression. Mood: still and calm. Composition: wide shot, subject small on the left, jetty leading to the right. Aspect ratio 4:5. No text, letters, numbers, logos or watermarks.

Cobblestone square with a vintage bicycle

A prop with a clear job (the bicycle) gives the pose a reason, which looks less staged.

Attach your photo first. “Cobblestone square with a vintage bicycle” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic street portrait of me walking a vintage bicycle with a flower basket across a cobblestone square, glancing toward the camera. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: an old square with shuttered windows and cafe chairs, late morning. Light: soft late-morning sun with open shade on the face. Camera: standard lens, shallow depth of field, street-photography framing. Mood: light and nostalgic. Composition: full body, bicycle crossing the frame diagonally. Aspect ratio 4:5. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic street portrait of me walking a vintage bicycle with a flower basket across a cobblestone square, glancing toward the camera. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: an old square with shuttered windows and cafe chairs, late morning. Light: soft late-morning sun with open shade on the face. Camera: standard lens, shallow depth of field, street-photography framing. Mood: light and nostalgic. Composition: full body, bicycle crossing the frame diagonally. Output at 4:5. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic street portrait of the person in my attached photo, walking a vintage bicycle with a flower basket across a cobblestone square, glancing toward the camera. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: an old square with shuttered windows and cafe chairs, late morning. Light: soft late-morning sun with open shade on the face. Camera: standard lens, shallow depth of field, street-photography framing. Mood: light and nostalgic. Composition: full body, bicycle crossing the frame diagonally. Aspect ratio 4:5. No text, letters, numbers, logos or watermarks.

Colourful fruit market stall

Warm bounce from nearby colours gives skin natural warmth without a filter.

Attach your photo first. “Colourful fruit market stall” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic candid portrait of me browsing a colourful fruit market stall, holding up a peach to inspect it. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: wooden crates of fruit, hand-lettered price tags and striped awnings. Light: open shade with warm bounce from the awnings. Camera: wide-angle lens, shallow depth of field, candid framing. Mood: lively and warm. Composition: waist-up, hands and fruit in the lower third. Aspect ratio 4:5. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic candid portrait of me browsing a colourful fruit market stall, holding up a peach to inspect it. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: wooden crates of fruit, hand-lettered price tags and striped awnings. Light: open shade with warm bounce from the awnings. Camera: wide-angle lens, shallow depth of field, candid framing. Mood: lively and warm. Composition: waist-up, hands and fruit in the lower third. Output at 4:5. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic candid portrait of the person in my attached photo, browsing a colourful fruit market stall, holding up a peach to inspect it. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: wooden crates of fruit, hand-lettered price tags and striped awnings. Light: open shade with warm bounce from the awnings. Camera: wide-angle lens, shallow depth of field, candid framing. Mood: lively and warm. Composition: waist-up, hands and fruit in the lower third. Aspect ratio 4:5. No text, letters, numbers, logos or watermarks.

Train window seat daydream

A profile view with a soft background keeps the face easy to hold steady.

Attach your photo first. “Train window seat daydream” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic portrait of me seated by a train window watching the countryside pass, chin resting on one hand. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: an empty carriage with patterned seats and a small table. Light: side light flickering through trees, golden patches on the seat and sleeve. Camera: standard lens, shallow depth of field, shot from across the aisle. Mood: dreamy and unhurried. Composition: profile three-quarter, window on the left, scenery softly blurred. Aspect ratio 4:5. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic portrait of me seated by a train window watching the countryside pass, chin resting on one hand. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: an empty carriage with patterned seats and a small table. Light: side light flickering through trees, golden patches on the seat and sleeve. Camera: standard lens, shallow depth of field, shot from across the aisle. Mood: dreamy and unhurried. Composition: profile three-quarter, window on the left, scenery softly blurred. Output at 4:5. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic portrait of the person in my attached photo, seated by a train window watching the countryside pass, chin resting on one hand. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: an empty carriage with patterned seats and a small table. Light: side light flickering through trees, golden patches on the seat and sleeve. Camera: standard lens, shallow depth of field, shot from across the aisle. Mood: dreamy and unhurried. Composition: profile three-quarter, window on the left, scenery softly blurred. Aspect ratio 4:5. No text, letters, numbers, logos or watermarks.

Glass greenhouse portrait

Diffused daylight prevents hard shadows, which are the usual cause of odd-looking faces in edits.

Attach your photo first. “Glass greenhouse portrait” edits a photo you upload, so it won't work without one.

Gemini

If no photo is attached or you can't edit images, say so in one sentence and stop. Using the uploaded photo of me as the reference, create a photorealistic portrait of me standing among tall plants in a glass greenhouse, gently touching a leaf and looking at it. Keep my face, skin tone, hairstyle and body shape exactly as in the uploaded photo, with no beauty filter, smoothing or slimming. Wardrobe: {{outfit}}. Setting: a steamy glass roof, terracotta pots and a misting hose. Light: bright diffused daylight with soft green bounce and a light haze. Camera: standard lens, shallow depth of field. Mood: fresh and serene. Composition: three-quarter body, plants framing both sides. Aspect ratio 3:4. No text, logos or watermark.

Nano Banana

If no photo is attached or you can't edit images, say so in one sentence and stop. Edit the attached photo of me into a photorealistic portrait of me standing among tall plants in a glass greenhouse, gently touching a leaf and looking at it. Preserve my exact face, facial features, skin tone and hairstyle; change only the wardrobe ({{outfit}}), the setting and the lighting. Setting: a steamy glass roof, terracotta pots and a misting hose. Light: bright diffused daylight with soft green bounce and a light haze. Camera: standard lens, shallow depth of field. Mood: fresh and serene. Composition: three-quarter body, plants framing both sides. Output at 3:4. No text, letters, numbers, logos or watermarks.

ChatGPT image

If no photo is attached or you can't edit images, say so in one sentence and stop. Create a photorealistic portrait of the person in my attached photo, standing among tall plants in a glass greenhouse, gently touching a leaf and looking at it. The face must match the reference photo exactly: same features, skin tone and hair, with no smoothing or reshaping. Wardrobe: {{outfit}}. Setting: a steamy glass roof, terracotta pots and a misting hose. Light: bright diffused daylight with soft green bounce and a light haze. Camera: standard lens, shallow depth of field. Mood: fresh and serene. Composition: three-quarter body, plants framing both sides. Aspect ratio 3:4. No text, letters, numbers, logos or watermarks.

Gemini photo packs by group

Seven occasion packs sit under this hub, from image generation to a polaroid couple photo, and each pack holds its own prompts to copy.

  • Group of adults in white and denim walking barefoot along the waterline at golden hour

    Image pack21 prompts

    Family photo prompts

    21 AI scenes that set the layout and keep each face, tips for a big family split, and 8 poses…

  • Four small figures holding hands on a wooden footbridge in a golden autumn forest

    Image pack20 prompts

    Family picture of 4 prompts

    20 layouts with height order set, from a diamond to 2 by 2 on steps, 4 for a family of 3, and…

  • Couple with wet hair laughing on a beach towel in an instant film frame

    Image pack22 prompts

    Polaroid couple photo prompts

    22 flash scenes that keep both faces, with film tint and a white frame in each, and a fix if…

  • Smiling man in a navy tee and grey overshirt against a plain light-grey wall

    Image pack20 prompts

    LinkedIn profile picture prompts

    18 jobs framed for the circle crop, a fix for your old photo, a 4:5 post to match, and a dress…

  • Restored 1970s colour print of two adults on a park picnic blanket

    Image pack22 prompts

    Old photo restoration prompts

    22 fixes for scratches, fading, tears, stains and folds, each with a line that keeps faces…

  • Steaming stone-grey ceramic mug on a worn oak table in warm morning light

    Image pack25 prompts

    Gemini image generation prompts

    25 scene, product and art ideas in full sentences, a rule for text in an image, and a fix if…

  • Couple walking hand in hand through a golden wheat field at sunset

    Image pack22 prompts

    Couple photo prompts

    22 scenes that keep both faces in 1 photo, with light and ratio set, a way to join 2 shots…

How do you use a Gemini AI photo prompt?

To use a Gemini AI photo prompt, upload one clear photo of yourself in the Gemini app, paste the prompt and change one field per turn, such as the outfit or the light. Here's how:

  1. Pick one clear, front-facing photo of yourself in even light, and don't use a filtered one. A sharp reference gives the best likeness, so it's worth 2 or 3 minutes to find one.
  2. Upload it in the Gemini app, then paste a prompt from the cards. You'll fill in the outfit field first, since it's the one field every card shares.
  3. Check the face first, since it's the detail people notice. If it's drifted, reply keep my face, skin tone, hair and expression exactly as in the uploaded photo and ask again.
  4. Change one field per follow-up, such as the colour of the top or the time of day, so you'll know what caused what.

Every card follows one pattern: the keep line, then fields for wardrobe, setting, light, camera angle and aspect ratio. Gemini's image model, Nano Banana, applies the edit, and each card's model tabs hold a Nano Banana version and a ChatGPT version. For more Nano Banana ideas, visit the Nano Banana prompts hub, or browse the full AI image prompts library.

How do you stop Gemini changing your face in a photo edit?

To stop Gemini changing your face, start the prompt with the words "keep my face, skin tone and hair exactly as in the photo" and ask for one change per turn.

Face drift happens when Gemini redraws the whole picture, so any part the prompt doesn't name can move. These habits hold the face in place:

  • Start every prompt with the keep line. It's the first thing each card on this page says, and it won't work as an afterthought.
  • Ask for one change at a time. Setting, outfit and light together is plenty, and that's enough to judge the result.
  • Pick soft light and simple backgrounds, since hard shadows and busy scenes tend to distort faces.
  • Regenerate instead of patching. A fresh try often fixes a face that's drifted.

Which photo should you upload to Gemini?

The photo to upload to Gemini is a sharp, front-facing shot in even daylight with no sunglasses, where your face fills about a third of the frame.

  • Show one person only, so Gemini knows whose face to keep.
  • Skip beauty filters and cropped foreheads, since the keep line can't hold a face the photo doesn't show clearly.
  • Responsible use is the rule on every card here: use only photos of yourself or of people who agreed, keep minors out of any edit shown publicly, and don't make convincing images of real public figures.

Say an image is AI-edited when you share it, and keep the original photo, since you'll want it for the next edit.

Gemini photo editing prompts by look

This page has 12 Gemini photo editing prompts across 5 looks, including cinematic, retro film, studio, travel and golden hour, each with the keep line and an aspect ratio.

  • Cinematic: rainy cafe window, neon street at night and train window seat.
  • Retro film: four-frame photo booth strip and cobblestone square with a vintage bicycle.
  • Studio: magazine cover with a made-up title, library ladder and glass greenhouse.
  • Travel: harbour jetty in morning mist and colourful fruit market stall.
  • Golden hour and autumn: rooftop portrait and park bench reading portrait.

Here's the spread: ratios run from 9:16 for the photo booth strip to 16:9 for the neon street, and 10 of the 12 cards use 4:5 or 3:4. Pick a model tab and all 12 cards change with it.

Gemini photo prompt packs by occasion

There are 7 Gemini photo prompt packs by occasion: image generation from text, family photos, a family of 4, a LinkedIn profile picture, old photo restoration, a couple photo and a polaroid couple photo. Each pack is its own page of prompts, so this hub isn't a long list of every occasion.

No photo to start from? A picture written from words alone is what Gemini image generation prompts covers. Holiday cards, three generations or siblings together sit on prompts for family photos. One frame with two adults and two children needs its own pose order, which you'll find in Gemini prompts for a family picture of 4.

A head-and-shoulders crop that reads well in LinkedIn's small circle is the job of Gemini prompts for a LinkedIn profile picture. Faded, torn or black and white prints get repaired with old photo restoration prompts. Instant-film frames for two people, white border and flash included, are on Gemini Polaroid couple photo prompts. Two faces from one photo get their own pack: the Gemini prompt for a couple photo.

Gemini AI photo prompt FAQ

Still stuck? Build it in Prompt Studio. It'll score each part as you go.

Is Gemini photo editing free?

Yes, Gemini photo editing is free to try in the Gemini app with a Google account, with a daily image limit that depends on your plan. Paid plans raise the limit, and Google lists the current numbers on its Gemini Apps help page.

Do Gemini photo prompts work in ChatGPT?

Yes, Gemini photo prompts work in ChatGPT image editing with the same keep line, and every card carries a ChatGPT tab next to the Nano Banana one. It's worth repeating the keep line and regenerating once or twice, if a face drifts.

How many photos can Gemini edit in a day?

The number of photos Gemini edits in a day depends on your Google plan, and Google lists the current limit on its Gemini Apps help page. The limit changes, so check that page instead of trusting a number that's gone stale here.

Can I use a Gemini photo prompt on a friend's picture?

No, don't use a Gemini photo prompt on a friend's picture without their consent. Edit only your own photos or photos the people in them agreed to, and keep minors out of any edit you'll share publicly.

Make your own in Prompt Studio

Answer a few questions. The prompt's scored as it's built. It's free, and nothing leaves your browser.