
20 Nano Banana Photo Editing Prompts (Copy and Paste)
Copy 20 Nano Banana photo editing prompts for backgrounds, portraits, object removal, lighting, restoration, and precise edits without wasting credits.
The best Nano Banana photo editing prompts state what may change, what must stay unchanged, and how the new pixels should match the original photo.
Use this compact formula:
Use the uploaded photo as the source.
CHANGE ONLY: [one specific edit].
PRESERVE: [identity, pose, crop, product, text, objects, and other protected details].
MATCH: [perspective, light direction, shadows, reflections, grain, and depth of field].
AVOID: [the failure you do not want].Every prompt below is free to copy. Browsing, copying, and using the free Nano Banana prompt generator require no account or credits. Image generation is paid, and the editor shows the price before you run it.
If background replacement is your only task, the free Nano Banana background change prompt builder creates a focused preservation and scene-matching brief without uploading an image.
Quick answer: which prompt should I use?
| Your edit | Start with | Protect first | Common failure |
|---|---|---|---|
| Replace a background | Prompt 1 | face, pose, edges, camera | subject looks pasted in |
| Remove an object | Prompt 5 | nearby objects, texture, shadows | a blurry patch remains |
| Improve a portrait | Prompt 9 | identity, age, skin texture | plastic skin or a new face |
| Make a headshot | Prompt 10 | facial structure, hair, clothing | excessive beautification |
| Change clothing | Prompt 11 | identity, body, pose, hands | body shape changes |
| Fix lighting | Prompt 13 | geometry, colors, exposure intent | relighting changes the scene |
| Restore an old photo | Prompt 17 | identity and period details | invented modern details |
| Combine two photos | Prompt 19 | separate reference roles | faces, light, or scale conflict |
For most reference-based edits, start with Nano Banana 2 at 1K or 2K. Use Nano Banana Pro when the unresolved problem is complex typography or a precision-heavy final. Do not pay for 4K while the composition is still changing.
Background editing prompts
1. Replace the background and preserve the person
Use the uploaded portrait as the source. Replace only the background with a quiet modern hotel lobby at golden hour. Preserve the person's facial structure, skin tone, age, hairstyle, expression, clothing, pose, hands, body proportions, crop, and camera angle. Match the new background perspective and depth of field to the original camera. Add realistic contact shadows and warm reflected light around the subject's edges. Do not retouch the face, redesign the outfit, or add people.If the person looks pasted in, do not regenerate with more style words. Ask the model to correct edge light, contact shadows, color spill, perspective, and depth of field.
2. Create a clean catalog background
Isolate the main subject from the uploaded photo and place it on a seamless warm-white studio background. Preserve the complete silhouette, fine hair or product edges, transparent parts, colors, materials, and proportions. Add one soft grounding shadow directly beneath the subject. Center the subject without changing its scale or camera angle. No props, border, text, halo, or clipped edges.3. Change the sky only
Replace only the overexposed sky with a believable late-afternoon sky containing soft broken clouds. Keep the buildings, trees, people, horizon, crop, lens perspective, and every foreground detail unchanged. Match the original sun direction and add only the subtle color and reflection changes naturally caused by the new sky. Do not move the horizon or turn the scene into a dramatic fantasy sunset.4. Blur a distracting background
Keep the person, face, hair, clothing, pose, foreground objects, crop, exposure, and colors unchanged. Increase only the background depth-of-field blur to resemble a portrait made with an 85mm lens at a wide aperture. Preserve clean, realistic transitions around hair, shoulders, and transparent details. Do not blur the subject, invent bokeh lights, or apply beauty retouching.Object removal and cleanup prompts
5. Remove one unwanted object
Remove only the red bag on the floor. Reconstruct the floor material, grout lines, wall texture, and shadows that would naturally continue behind it. Keep every person, reflection, object, color, crop, and camera detail unchanged. The repaired area must have consistent texture and noise with no blur, repeated pattern, stain, or new object.Naming the object and its location is safer than saying “clean up the image.” A broad cleanup instruction gives the model permission to remove details you wanted.
6. Remove a background person
Remove only the person in the far-left background wearing a red jacket. Reconstruct the pavement, storefront, window reflection, and any occluded background geometry. Preserve the main subject, all other people, signage, lighting, crop, and perspective exactly as they are. Do not alter faces or duplicate nearby pedestrians.7. Remove glare from glasses
Reduce only the bright glare covering the subject's eyeglass lenses. Preserve the exact face, eye shape, eye color, gaze, glasses frame, skin texture, expression, hair, lighting, and crop. Reconstruct the visible eyes naturally through the lenses while retaining a subtle realistic lens reflection. Do not remove the glasses or beautify the face.8. Clean a product photo without redesigning it
Clean dust, fingerprints, and tiny surface smudges from the product while preserving its exact shape, dimensions, materials, texture, logo, label spelling, color, cap, seams, reflections, camera angle, and background. Keep natural material variation and realistic highlights. Do not smooth away edges, alter packaging text, add claims, or make the product look like a different model.For product-specific preservation rules, use the Nano Banana product photography prompt guide.
Portrait and headshot prompts
For LinkedIn, resume, executive, speaker, team-directory, and controlled-retouching briefs, use the complete Nano Banana professional headshot prompt guide.
9. Natural portrait retouch
Retouch this portrait conservatively. Preserve the person's exact facial structure, age, skin tone, eye shape, nose, mouth, jawline, expression, hairstyle, and distinguishing features. Remove only temporary blemishes and stray lint. Reduce under-eye darkness slightly while keeping pores, fine lines, freckles, and natural asymmetry. Maintain the original light, color, crop, and background. No face reshaping, skin whitening, airbrushing, makeup, or plastic texture.No prompt can guarantee biometric identity. Compare the result with the source, especially around the eyes, jaw, teeth, ears, and hairline.
10. Professional headshot from a casual portrait
Use the uploaded photo as the identity reference. Preserve the same facial structure, skin tone, age, hairstyle, expression, body proportions, and recognizable features. Change the clothing to a tailored navy blazer over a plain light shirt. Replace the background with a soft neutral-gray studio backdrop. Use an eye-level shoulder-up crop, gentle key light, subtle fill, realistic skin texture, and natural posture. Do not slim the face, whiten teeth, add jewelry, or create a beauty-filter look.11. Change clothing only
Preserve the person's identity, face, hair, age, expression, body shape, pose, hands, camera angle, background, and lighting. Change only the clothing to a dark green linen overshirt over a plain cream T-shirt. Make the garments follow the existing body position with believable seams, folds, texture, and occlusion. Do not change anatomy, hairstyle, accessories, or skin.12. Correct a closed-eye photo
Open only the subject's closed eyes using the person's visible facial features as the reference. Preserve the exact face, eyelid shape, eye spacing, gaze direction, expression, age, skin texture, light, and all other pixels. Make the eyes relaxed and consistent with the camera angle. Do not enlarge the eyes, change eye color, reshape the face, or add makeup.When identity matters across several edits, reuse the same reference and the same preservation block. The character consistency guide explains that workflow in detail.
Lighting and color prompts
13. Fix flat indoor lighting
Improve only the lighting in this indoor portrait. Preserve the subject's identity, face, skin tone, clothing, pose, room layout, object positions, crop, and camera perspective. Add soft directional window light from camera left, gentle fill on the shadow side, and realistic contact shadows. Keep highlights below clipping and retain natural skin texture. Do not replace the room or change the time of day.14. Match the color of a reference photo
Use image 1 as the photo to edit and image 2 only as a color-grading reference. Preserve every subject, face, object, material, text element, composition, crop, and lighting direction from image 1. Transfer only the overall white balance, contrast curve, shadow color, highlight color, and saturation character from image 2. Do not copy people, objects, scenery, or composition from image 2.15. Turn daytime into blue hour
Transform this street photo from midday to natural blue hour. Preserve the exact architecture, signs, vehicles, people, road layout, camera position, and crop. Darken the sky gradually, illuminate existing windows and streetlights, and add restrained cool ambient light with realistic warm practical lights. Keep all existing text unchanged. No rain, neon signs, new buildings, or fantasy colors.16. Repair a dark, noisy phone photo
Improve the visibility of this low-light phone photo while preserving the scene, people's identities, expressions, positions, clothing, colors, crop, and nighttime atmosphere. Lift shadow detail conservatively, reduce color noise, correct the white balance, and recover natural edge detail. Keep realistic fine grain and avoid halos. Do not turn night into day, smooth faces, invent details, or oversharpen.Restoration and composition prompts
17. Restore an old family photograph
Restore this scanned family photograph while preserving every person's identity, age, expression, pose, clothing, hairstyle, and the original historical setting. Remove only dust, scratches, stains, fold marks, and scanning discoloration. Recover restrained contrast and fine detail without inventing facial features or modernizing clothing. Retain a subtle period-appropriate film texture. Produce a restored black-and-white version.For colorization, make a second pass only after approving the structural restoration:
Colorize this approved restoration with restrained, historically plausible colors. Preserve all faces, clothing details, objects, lighting, texture, and composition exactly. Use natural skin tones and period-appropriate material colors. Do not change facial features, add makeup, modernize garments, or remove the original film character.18. Expand the canvas for a landscape crop
Extend this vertical photo into a horizontal 16:9 composition. Keep the original image centered and preserve every original pixel, subject, face, object, perspective, and light direction. Continue the environment naturally to the left and right using consistent geometry, texture, depth, and atmospheric perspective. Do not stretch, resize, or reposition the original subject, and do not add new focal objects.19. Put a person into a reference environment
Use image 1 only for the person's identity, clothing, and body proportions. Use image 2 only for the environment, camera perspective, and lighting. Place the person naturally in the environment from image 2 at a believable scale. Preserve the face, hair, age, outfit, and body shape from image 1. Match contact shadows, color spill, light direction, focal depth, and grain to image 2. Do not blend identities or copy people from the environment reference.20. Final cleanup without redesign
Treat this as a technical finishing pass, not a redesign. Preserve the subject, identity, anatomy, composition, crop, camera angle, colors, text, logos, object positions, and background. Correct only small edge artifacts, inconsistent shadows, accidental duplicates, isolated texture errors, and visible generation seams. Keep natural skin and material texture. Do not add, remove, restyle, or reposition anything else.How to repair a failed photo edit
Do not keep submitting the same prompt. Name the visible failure and narrow the next edit.
| Failure | Repair instruction |
|---|---|
| Face changed | “Restore the face from the original reference; change no facial geometry.” |
| Subject looks pasted in | “Match edge light, color spill, contact shadow, grain, and focal depth.” |
| Removal left a smear | “Reconstruct the specific material and continue its lines and texture.” |
| Clothing changed the body | “Restore the original body proportions, pose, hands, and silhouette.” |
| Scene was redesigned | “Revert all non-target regions; change only [named target].” |
| Skin looks plastic | “Restore pores, fine lines, natural asymmetry, and restrained sharpening.” |
| Text or logo mutated | “Restore the exact source artwork; do not regenerate or reinterpret it.” |
| Too many details drifted | Return to the source image and make one edit instead of repairing the derivative. |
Research on general-purpose image restoration reports that concise prompts with explicit fidelity constraints can produce a better balance between reconstruction accuracy and perceived quality. That supports a practical rule: add boundaries before adding adjectives. See the 2026 preprint Can Nano Banana 2 Replace Traditional Image Restoration Models?.
A credit-conscious editing workflow
- Start from the cleanest available source, not a screenshot of a screenshot.
- Make one structural edit at a time.
- Put protected details in a reusable
PRESERVEblock. - Work at 1K first; move to 2K after the edit direction is correct.
- Repair a local failure with a narrow instruction instead of regenerating everything.
- Return to the original source if several details have drifted.
- Use 4K only for the approved final.
- Use Pro only when a specific precision problem justifies the higher cost.
| Working stage | Suggested starting point | Cost here |
|---|---|---|
| Cheap concept | Nano Banana · 1K | 4 credits |
| Reference edit draft | Nano Banana 2 · 1K | 8 credits |
| Working edit | Nano Banana 2 · 2K | 12 credits |
| Approved 4K final | Nano Banana 2 · 4K | 18 credits |
| Precision-heavy 2K edit | Nano Banana Pro · 2K | 16 credits |
Google's own guidance recommends being specific, describing the complete scene, and refining through conversation rather than treating every attempt as a fresh start. See Google's official Nano Banana prompting tips.
Before uploading a personal photo
- Do not upload identity documents, medical records, financial documents, intimate images, or photos you do not have permission to use.
- Remove location metadata when it is not needed.
- Crop out unrelated people and private background details.
- Keep the original file; AI output is a derivative, not a backup.
- Review faces, hands, text, logos, and factual details before publishing or selling the result.
The prompts on this page improve instruction clarity; they do not guarantee pixel-perfect identity, factual restoration, or production-ready text.
Build your own editing prompt
Use the free Nano Banana prompt generator to turn a specific editing goal into a structured CHANGE, PRESERVE, MATCH, and AVOID brief. It runs locally and consumes no generation credits.
When the prompt is ready:
カテゴリー
その他の投稿

25 Nano Banana Prompts for Editing, Product Photos, Text, and Characters
Copy practical Nano Banana prompts for photo editing, product scenes, character consistency, posters, and 4K finals—plus a workflow that saves credits.


Nano Banana 2 vs Nano Banana Pro: Which Should You Use?
Compare Nano Banana 2 and Nano Banana Pro for editing, text, infographics, speed, 4K output, and credit cost. Get a task-by-task recommendation.


Nano Banana Character Consistency Prompts: 15 Practical Templates
Keep a person or character recognizable across poses, scenes, expressions, and story panels with 15 Nano Banana consistency prompts and a low-cost workflow.
