How to use these prompts in ChatGPT

  1. Open ChatGPT Images or a chat that supports image editing and upload your photo.
  2. Copy the prompt for the effect you need. Replace scene-specific words such as “mug,” “bookshop,” or “car” with details from your image.
  3. Make one change at a time. Name the target area, the intended edit, and the objects or details to preserve.
  4. Save the result and compare it with the original at full size. Retry from the original if the subject, text, or product details drift.

OpenAI says you can upload an image and describe edits, select an area, request text or transparency, and change aspect ratio. Its editor also warns that selection is not always precise and changes can extend beyond the selected area. Those limitations matter in the examples below. See OpenAI’s Images in ChatGPT help and its image prompting guide.

1. Remove the background

Use it for: Make a product cutout for a shop listing, slide, or design mockup.

Prompt used for this result
Edit Image 1 only. Isolate the exact cobalt-blue ceramic travel mug and black lid on a truly transparent background with clean edges and no white halo. Keep the mug, handle, lid and cream oval emblem recognizable. Remove the oak table and wall. Return a single product cutout, no text, no checkerboard pattern, no collage.
Original cobalt-blue travel mug on an oak table against an off-white wall
Input: original cobalt-blue travel mug on an oak table against an off-white wall. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 1: remove the background. The table and wall are gone. We checked the displayed WebP file and confirmed an alpha channel with both transparent and opaque pixels.
Edited result: The table and wall are gone. We checked the displayed WebP file and confirmed an alpha channel with both transparent and opaque pixels. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The table and wall are gone. We checked the displayed WebP file and confirmed an alpha channel with both transparent and opaque pixels.

Before you use it: Open the saved file over both light and dark backgrounds to catch pale edge halos.

2. Improve overall image quality

Use it for: Give an underexposed or soft photo a restrained first pass before a more specific edit.

Prompt used for this result
Edit Image 1 as a restrained photo enhancement. Improve overall clarity, exposure and tonal balance slightly while keeping the same adult woman's identity, pose, jacket, bookshop, passersby, framing and natural skin texture. Do not beautify, replace the face, add details, crop, or make the image look synthetic. Single edited photograph, no before/after collage.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 2: improve overall image quality. The portrait is slightly clearer and more balanced while its scene remains recognizable.
Edited result: The portrait is slightly clearer and more balanced while its scene remains recognizable. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The portrait is slightly clearer and more balanced while its scene remains recognizable.

Before you use it: Compare the face, hands, signage, and background objects at full size; “enhance” can quietly redraw them.

3. Replace a background

Use it for: Place a portrait in a new setting when a literal record of the original location is unnecessary.

Prompt used for this result
Edit Image 1. Replace only the bookshop and street background with a quiet sunlit garden terrace. Keep the exact fictional adult woman, her terracotta jacket, cream blouse, pose, expression, hands, proportions and framing recognizably the same. Match edge lighting and shadow direction naturally. Single edited photograph, no collage, no text.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.

What we got: The bookshop becomes a garden terrace and the subject remains visibly the same fictional person.

Before you use it: Inspect hair edges and light direction; treat the new background as a generated scene, not a documentary photo.

4. Correct color and exposure

Use it for: Neutralize a color cast without turning the scene into a stylized grade.

Prompt used for this result
Edit Image 1 as a subtle color correction. Neutralize the warm cast, gently open the shadows and protect the blue sea and red car from oversaturation. Keep the car, road, trees, pedestrian, geometry, framing and afternoon mood unchanged. No added objects, no dramatic cinematic grade, no collage or text.
Original red vintage compact car parked along a sunny coastal road
Input: original red vintage compact car parked along a sunny coastal road. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 4: correct color and exposure. The shadows and sky read more evenly; the red car and blue sea stay restrained.
Edited result: The shadows and sky read more evenly; the red car and blue sea stay restrained. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The shadows and sky read more evenly; the red car and blue sea stay restrained.

Before you use it: Look at white or gray surfaces and skin, if present, before judging color by the most saturated object.

5. Remove an unwanted object

Use it for: Clear a distraction such as a passerby from a casual image.

Prompt used for this result
Edit Image 1. Remove only the dark-backpack passerby at the far left edge and reconstruct the pavement, street and plants naturally. Keep the central fictional adult woman, all her features and clothing, bookshop, other visible background people, lighting and framing the same. Single edited photograph, no collage, no text.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 5: remove an unwanted object. The backpacked passerby disappears while the central portrait remains intact at normal viewing size.
Edited result: The backpacked passerby disappears while the central portrait remains intact at normal viewing size. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The backpacked passerby disappears while the central portrait remains intact at normal viewing size.

Before you use it: Inspect the repaired pavement and nearby plants for repeated textures or ghost outlines.

6. Add text inside an image

Use it for: Mock up a social graphic or product headline when the words are short.

Prompt used for this result
Edit Image 1. Add the exact headline 'MORNING, MADE SIMPLE' in clean, legible dark navy sans-serif type in the empty wall area above and to the left of the cobalt-blue mug. Keep the mug, cream oval emblem, table, lighting and photograph unchanged. Spell the words exactly as quoted. Single edited photograph, no collage.
Original cobalt-blue travel mug on an oak table against an off-white wall
Input: original cobalt-blue travel mug on an oak table against an off-white wall. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 6: add text inside an image. The requested headline appears legibly across two lines above the mug.
Edited result: The requested headline appears legibly across two lines above the mug. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The requested headline appears legibly across two lines above the mug.

Before you use it: Read every letter and punctuation mark before publishing. For final web copy, accessible HTML text is often easier to correct and translate.

7. Replace an object

Use it for: Preview a different product or prop in the same photographic setting.

Prompt used for this result
Edit Image 1. Replace ONLY the cobalt-blue travel mug with a cream ceramic teapot of comparable height, with a short spout on the left and a handle on the right. Keep the teapot grounded at the same place on the oak table with a believable contact shadow. Preserve the off-white wall, tabletop grain, camera angle, daylight and framing. Remove the mug completely; do not leave its blue body, lid or emblem behind. One edited photograph, no text or collage.
Original cobalt-blue travel mug on an oak table against an off-white wall
Input: original cobalt-blue travel mug on an oak table against an off-white wall. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 7: replace an object. The travel mug is replaced by a cream teapot; the oak table and plain wall remain recognizable.
Edited result: The travel mug is replaced by a cream teapot; the oak table and plain wall remain recognizable. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The travel mug is replaced by a cream teapot; the oak table and plain wall remain recognizable.

Before you use it: Check the new object’s scale, contact shadow, and edges where the old object used to be.

8. Expand a photo to a new aspect ratio

Use it for: Adapt a landscape photo to a vertical story or phone screen without cutting off the subject.

Prompt used for this result
Edit Image 1 into a vertical 9:16 photograph by extending the existing coastal landscape above and below. Keep the red vintage car whole and recognizable at approximately the same relative scale, and retain the road, sea, tree and pedestrian. Add plausible sky and foreground road as needed; do not stretch the car or crop it. Single image, no borders, collage or text.
Original red vintage compact car parked along a sunny coastal road
Input: original red vintage compact car parked along a sunny coastal road. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 8: expand a photo to a new aspect ratio. The output is 941 × 1672 pixels, approximately 9:16, with the full car and much more sky and road.
Edited result: The output is 941 × 1672 pixels, approximately 9:16, with the full car and much more sky and road. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The output is 941 × 1672 pixels, approximately 9:16, with the full car and much more sky and road.

Before you use it: Expanded space is invented. Avoid presenting newly generated scenery as evidence of the location.

9. Add a realistic product shadow

Use it for: Ground a transparent product cutout on a clean studio surface.

Prompt used for this result
Edit Image 1, a transparent cutout of a cobalt-blue ceramic travel mug. Place this exact mug on a seamless very light gray studio floor and background, and create a soft realistic contact shadow cast to the lower right. Preserve the mug's blue finish, black lid, handle and blank cream oval emblem. Single product image, no text, no collage.
Transparent mug cutout from use 1, reused as the input for the product-shadow edit
Input: transparent mug cutout from use 1, reused as the input for the product-shadow edit. Credit: ChatGPT image edit by Grow AI Skills. Open full-size image.
Edited result for use 9: add a realistic product shadow. The cutout becomes a studio product image with a soft contact shadow.
Edited result: The cutout becomes a studio product image with a soft contact shadow. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The cutout becomes a studio product image with a soft contact shadow.

Before you use it: The shadow should begin at the base and follow a plausible light direction; a floating shadow weakens the mockup.

10. Change a photo into an illustration

Use it for: Create an editorial illustration while keeping the source composition.

Prompt used for this result
Edit Image 1 into a refined hand-drawn editorial illustration with colored pencil and restrained watercolor texture. Keep the central fictional adult woman's recognizable pose, terracotta jacket, cream blouse, bookshop doorway and basic composition. Draw as a coherent illustration rather than a photo filter. Single image, no text or collage.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 10: change a photo into an illustration. The scene reads as colored pencil and watercolor rather than a flat filter.
Edited result: The scene reads as colored pencil and watercolor rather than a flat filter. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The scene reads as colored pencil and watercolor rather than a flat filter.

Before you use it: Faces and small bookshop details are redrawn. Use the illustration as artwork, not as an exact visual record.

11. Blur the background

Use it for: Emphasize a subject in a busy portrait.

Prompt used for this result
Edit Image 1 to create a shallow-depth-of-field portrait. Keep the fictional adult woman and her hair, face, jacket, hands and shoulder bag sharp. Gradually blur only the street, passersby and bookshop background into natural lens bokeh, as if photographed with a wide aperture. Preserve composition and plausible edges. One photograph, no collage.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 11: blur the background. The woman stays sharp while the street and shop soften into a stronger bokeh effect.
Edited result: The woman stays sharp while the street and shop soften into a stronger bokeh effect. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The woman stays sharp while the street and shop soften into a stronger bokeh effect.

Before you use it: Zoom into hair and shoulder edges; blur transitions can create halos.

12. Restore an old photograph

Use it for: Preview a cleaner version of a scratched family scan.

Prompt used for this result
Restore Image 1, a fictional damaged black-and-white family photo. Reduce scratches, dust, crease lines and fading; recover natural grayscale contrast and facial detail while preserving all three people's identities, ages, expressions, clothing and original framing. Do not colorize, glamorize or invent missing features. One restored photograph, no collage.
Original fictional scratched black-and-white family photograph of three adults
Input: original fictional scratched black-and-white family photograph of three adults. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 12: restore an old photograph. Scratches and the central crease are largely removed, and contrast improves.
Edited result: Scratches and the central crease are largely removed, and contrast improves. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: Scratches and the central crease are largely removed, and contrast improves.

Before you use it: Keep the original scan. Restored faces and clothing texture may contain inferred detail; do not overwrite archival evidence.

13. Retouch skin conservatively

Use it for: Make a light portrait touch-up without changing facial structure.

Prompt used for this result
Retouch Image 1 conservatively. Remove only a few temporary skin blemishes and reduce under-eye darkness slightly on the fictional adult woman. Keep her pores, freckles, expression, age, facial structure, hair, clothes, bookshop and framing. Avoid plastic skin or face replacement. Single edited photograph, no collage.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 13: retouch skin conservatively. The change is deliberately slight; the portrait still reads as the same person and scene.
Edited result: The change is deliberately slight; the portrait still reads as the same person and scene. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The change is deliberately slight; the portrait still reads as the same person and scene.

Before you use it: If the edit is barely visible, that can be preferable to a waxy face. Confirm consent before retouching a real person.

14. Add a product reflection

Use it for: Create a glossy hero treatment for a product concept.

Prompt used for this result
Edit Image 1 into a clean product photograph on a glossy dark surface. Keep the exact cobalt-blue ceramic mug, black lid, handle and cream oval emblem recognizable. Add a plausible softer reflection directly below the mug, aligned with its base and lighting. Single image, no extra products or text.
Original cobalt-blue travel mug on an oak table against an off-white wall
Input: original cobalt-blue travel mug on an oak table against an off-white wall. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 14: add a product reflection. The mug appears on a dark reflective surface with a softer inverted reflection.
Edited result: The mug appears on a dark reflective surface with a softer inverted reflection. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The mug appears on a dark reflective surface with a softer inverted reflection.

Before you use it: Check that the reflected handle and base align with the actual mug.

15. Combine two photos

Use it for: Build a concept composite from a subject photo and a scene photo.

Prompt used for this result
Use Image 1 as the woman reference and Image 2 as the coastal car scene. Make one believable photograph placing the same fictional adult woman in her terracotta jacket standing safely beside the parked red car, scaled naturally to the vehicle, while retaining the coastal road, sea and trees. Match sunlight and shadows. No duplicate people, no collage, no text.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Original red vintage compact car parked along a sunny coastal road
Input: original red vintage compact car parked along a sunny coastal road. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 15: combine two photos. The woman and car share one plausible coastal scene, but her pose and some vehicle details were regenerated.
Edited result: The woman and car share one plausible coastal scene, but her pose and some vehicle details were regenerated. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The woman and car share one plausible coastal scene, but her pose and some vehicle details were regenerated.

Before you use it: Use this for a concept image. Do not claim it documents a real event or exact product details.

16. Make a lifestyle product mockup

Use it for: Preview a product in a plausible home or desk setting.

Prompt used for this result
Edit Image 1 into a natural lifestyle product mockup. Place the same cobalt-blue ceramic travel mug with black lid, handle and blank cream oval emblem on a tidy wooden desk beside a closed notebook and a small plant. Keep the mug shape and product details clear and consistent, with believable scale and daylight. Single image, no slogans or collage.
Original cobalt-blue travel mug on an oak table against an off-white wall
Input: original cobalt-blue travel mug on an oak table against an off-white wall. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 16: make a lifestyle product mockup. The mug appears on a desk with a notebook and plant in soft daylight.
Edited result: The mug appears on a desk with a notebook and plant in soft daylight. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The mug appears on a desk with a notebook and plant in soft daylight.

Before you use it: If the mug represents a real SKU, verify dimensions, lid shape, materials, and label against an approved product photograph.

17. Increase the output resolution

Use it for: Make a small image easier to use in a larger layout.

Prompt used for this result
Edit Image 1, a deliberately low-resolution 362 by 272 pixel product photo. Reconstruct a cleaner, higher-resolution version of the same cobalt-blue ceramic travel mug on the same wood table with off-white wall. Preserve the black lid, handle and cream oval emblem. Increase detail without inventing letters or changing the mug design. Single image, no collage.
Low-resolution copy of the original mug photograph, prepared as a 362 by 272 pixel input
Input: low-resolution copy of the original mug photograph, prepared as a 362 by 272 pixel input. Credit: Derived from our generated image by Grow AI Skills. Open full-size image.
Edited result for use 17: increase the output resolution. The 362 × 272 pixel input produced a 1447 × 1087 pixel output, almost four times the width and height. The model reconstructed texture rather than recovering hidden source detail.
Edited result: The 362 × 272 pixel input produced a 1447 × 1087 pixel output, almost four times the width and height. The model reconstructed texture rather than recovering hidden source detail. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The 362 × 272 pixel input produced a 1447 × 1087 pixel output, almost four times the width and height. The model reconstructed texture rather than recovering hidden source detail.

Before you use it: Pixel count alone does not prove fidelity. Compare logo edges, small print, and material texture to a higher-quality original if one exists.

18. Create a cinematic light treatment

Use it for: Test a mood shift for a campaign image or story visual.

Prompt used for this result
Edit Image 1 with a cinematic yet credible evening light treatment. Keep the fictional adult woman, her terracotta jacket, cream blouse, pose, face, bookshop and framing recognizable. Shift the scene to a soft blue-hour environment with a warm bookstore glow and balanced skin tones; retain natural shadows and do not add props. Single edited photograph, no collage.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 18: create a cinematic light treatment. The daylight portrait becomes a blue-hour scene with warm bookstore light.
Edited result: The daylight portrait becomes a blue-hour scene with warm bookstore light. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The daylight portrait becomes a blue-hour scene with warm bookstore light.

Before you use it: The time of day is fabricated. Avoid using the edit where lighting carries factual meaning.

19. Remove your own watermark

Use it for: Clean a mark that you placed on your own synthetic or licensed image.

Prompt used for this result
Edit Image 1, our own synthetic mug photograph with our own demo watermark. Remove ONLY the gray words 'GROW AI SKILLS DEMO' from the upper-right off-white wall. Reconstruct the wall texture and lighting in that small text area. Keep the cobalt-blue travel mug, black lid, blank cream emblem, handle, wood table, camera angle, framing and all other areas visually unchanged. One clean product photograph, no words, no collage.
Synthetic blue travel mug photograph with a Grow AI Skills Demo watermark on the upper-right wall
Input: synthetic blue travel mug photograph with a Grow AI Skills Demo watermark on the upper-right wall. Credit: ChatGPT image edit by Grow AI Skills. Open full-size image.
Edited result for use 19: remove your own watermark. The gray watermark is gone and the mug, table, and camera view remain recognizable. This is a visual edit, not a pixel-identical restoration.
Edited result: The gray watermark is gone and the mug, table, and camera view remain recognizable. This is a visual edit, not a pixel-identical restoration. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The gray watermark is gone and the mug, table, and camera view remain recognizable. This is a visual edit, not a pixel-identical restoration.

Before you use it: This worked on a plain wall. In two earlier attempts on a busier coastal scene, the model changed the car and road, so compare the whole frame. For the cleanest source, use your unwatermarked master.

20. Turn a photo into a pencil sketch

Use it for: Create monochrome portrait artwork from a photograph.

Prompt used for this result
Edit Image 1 into a detailed graphite pencil sketch on warm white paper. Keep the central fictional adult woman's face, expression, hair, jacket, hands, bookshop doorway and overall composition recognizable. Use monochrome graphite shading and confident line work, not a color illustration or photo filter. Single image, no words or collage.
Original fictional portrait of a woman in a terracotta jacket outside a bookshop
Input: original fictional portrait of a woman in a terracotta jacket outside a bookshop. Credit: Original synthetic image: Grow AI Skills / ChatGPT image generation. Open full-size image.
Edited result for use 20: turn a photo into a pencil sketch. The portrait becomes a graphite-style drawing with visible pencil shading.
Edited result: The portrait becomes a graphite-style drawing with visible pencil shading. Credit: ChatGPT image edit and editorial review by Grow AI Skills. Open full-size image.

What we got: The portrait becomes a graphite-style drawing with visible pencil shading.

Before you use it: Treat it as a new interpretation; small facial and background details are redrawn.