Prompt Library

30 ChatGPT Photo Enhance Prompts to Fix Soft Phone Photos

30 copy-paste prompts

Upload [YOUR PHOTO] to ChatGPT (GPT-4o) and paste a prompt to sharpen, unblur, lift shadows, or upscale for print. Keep faces true.

In short: This page contains 30 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Hand-curated and tested by the AI Academy team.

By Louis Corneloup · Founder, Techpresso

Last updated September 2, 2026·Hand-curated & tested by the AI Academy team

Sharpen and Unblur

5 prompts

Gentle Sharpen a Soft Phone Pic

1/30

✨ What it does

ChatGPT gently sharpens the soft phone snapshot in [YOUR PHOTO] so faces and clothes read clearly without inventing new features. Compare it with the original, then download the clearer file if nobody looks swapped.

Enhance [YOUR PHOTO], a slightly soft phone snapshot. Gently sharpen existing edges in faces, hair, and clothing so the photo looks in focus, using only detail already present. Do not change the face, do not add objects, and do not modernize clothes. Keep identity, pose, and the whole scene identical. Avoid a crunchy or plastic finish. Natural, photographic result, original aspect ratio.

💡

Pro tip: Add 'enhance only detail already present' if the first pass looks too crisp or fake.

Unblur Handheld Motion Shake

2/30

✨ What it does

ChatGPT recovers edges on the motion-blurred subject in [YOUR PHOTO] so the handheld shake reads as a still frame. Compare pose and clothing with the original, then regenerate if it invented a sharper face than the blur supports.

Unblur [YOUR PHOTO], which has handheld motion blur from a slight camera shake. Recover the subject's edges and facial detail from the smear, reconstructing only what the existing blur implies. Do not change the face, do not add objects, do not invent a different pose, and do not modernize clothes. Keep clothing, background, and identity exactly as captured. Conservative, photographic correction, original aspect ratio.

💡

Pro tip: If the face is a smear, add 'stay conservative and do not invent unclear features.'

Recover a Missed-Focus Shot

3/30

✨ What it does

ChatGPT pulls the missed-focus subject in [YOUR PHOTO] into usable sharpness while leaving the background alone. Line it up with the original, then keep the take if the person still looks like themselves.

Enhance [YOUR PHOTO], where focus landed on the background instead of the subject. Bring the intended subject into usable sharpness using only the detail already in the frame. Do not change the face, do not add objects, and do not modernize clothes. Leave the background as it is. Keep identity, pose, and framing identical. Natural photographic focus correction, original aspect ratio.

💡

Pro tip: Name who should be sharp ('the person in front') so it does not sharpen the wrong plane.

Sharpen Edges Without Plastic Skin

4/30

✨ What it does

ChatGPT sharpens eyes, hair, and fabric in [YOUR PHOTO] while leaving pores and lines so skin does not turn plastic. Compare faces with the original, then print only if texture still looks real.

Sharpen [YOUR PHOTO] with restraint. Increase edge clarity on eyes, hairline, fabric, and background objects, but keep natural skin texture, pores, and lines. Do not change the face, do not airbrush, do not add objects, and do not modernize clothes. No halos, no crunchy contrast, no plastic finish. Faithful photographic sharpen, original aspect ratio.

💡

Pro tip: Add 'no skin smoothing or beautifying' because GPT-4o often over-retouches faces by default.

Conservative Camera-Shake Deblur

5/30

✨ What it does

ChatGPT deblurs short camera shake on [YOUR PHOTO] just enough to make faces readable without guessing new features. Compare it with the original, then keep the conservative version if identity still holds.

Deblur [YOUR PHOTO] from a short burst of camera shake. Recover usable outline and facial readability without inventing features that are not implied by the blur. Do not change the face, do not add objects, do not rewrite the scene, and do not modernize clothes. Keep clothes, lighting, and identity unchanged. Prefer a slightly soft honest result over a fake crisp one, original aspect ratio.

💡

Pro tip: Ask for 'slightly soft is acceptable' so it stops before inventing crisp eyes.

Upscale for Print

5 prompts

Upscale for an 8x10 Print

6/30

✨ What it does

ChatGPT upscales [YOUR PHOTO] into an 8x10 print file, rebuilding texture only from what is already visible. View it at print size next to the original, then send the file to a lab if the likeness holds.

Upscale [YOUR PHOTO] to a clean high-resolution file suitable for an 8x10 inch print. Reconstruct natural texture in faces, hair, and fabric only from what is already visible. Do not change the face, do not add objects, and do not modernize clothes. Keep pose, crop, and scene identical. Avoid invented pores or fake sharpness. Print-ready, original aspect ratio.

💡

Pro tip: State the exact size ('8x10 at 300 dpi') so it prioritizes the right amount of detail.

Enlarge a Tight 12MP Crop

7/30

✨ What it does

ChatGPT enlarges the tight 12MP crop in [YOUR PHOTO] so the small soft cut becomes a usable print. Zoom in on faces, then download the larger file if no one looks swapped.

Upscale [YOUR PHOTO], a tight crop from a 12MP phone file that now looks small and soft. Enlarge it to a usable print size, recovering edges without inventing people, objects, or a new face. Do not change the face, do not add objects, and do not modernize clothes. Keep the crop and composition exactly as uploaded. Honest enlargement, original aspect ratio.

💡

Pro tip: If you still have the uncropped original, upload that instead and crop after the upscale.

Print-Ready 5x7 from a Phone

8/30

✨ What it does

ChatGPT turns the phone snapshot in [YOUR PHOTO] into a clean 5x7 print file without rewriting the scene. Compare it with the original, then print if faces and clothes still match.

Prepare [YOUR PHOTO], a typical phone snapshot, as a print-ready 5x7. Upscale modestly, sharpen existing detail, and keep color and exposure close to the original. Do not change the face, do not add objects, and do not modernize clothes. Do not rewrite the scene. Clean, artifact-free file suitable for a drugstore print, original aspect ratio.

💡

Pro tip: Ask for 'no extra contrast punch' if you want the print to match the phone screen.

Framed Enlargement Without Fake Texture

9/30

✨ What it does

ChatGPT upscales [YOUR PHOTO] for a large framed print, adding only texture the file already supports. View it at wall size, then order the frame if detail still looks photographic.

Upscale [YOUR PHOTO] for a large framed enlargement. Increase resolution so faces stay readable on the wall, reconstructing texture only from genuine detail. Do not change the face, do not add objects, and do not modernize clothes. Avoid halos, invented pores, and over-sharpening that show at arm's length. Faithful wall print, original aspect ratio.

💡

Pro tip: Add 'avoid halos and over-sharpening' because those faults get loud at enlargement size.

Social Download to Print

10/30

✨ What it does

ChatGPT enlarges the social-app download in [YOUR PHOTO] and eases the extra softness from recompression. Compare it with the original upload, then print only if you cannot find a higher-resolution source.

Upscale [YOUR PHOTO], a photo re-downloaded from Instagram, iMessage, or another social app, into a printable file. Reduce the extra softness and compression from the download, then enlarge conservatively. Do not change the face, do not add objects, and do not modernize clothes. Keep the scene identical. Honest print from a social file, original aspect ratio.

💡

Pro tip: Always prefer the Camera Roll original over a re-download. Social apps throw pixels away.

Fix Phone Photos

5 prompts

Clean Up a Soft iPhone Snapshot

11/30

✨ What it does

ChatGPT gives the soft iPhone or Android snapshot in [YOUR PHOTO] a light sharpen and exposure tidy-up without a studio makeover. Compare it with the original, then download the everyday usable version.

Enhance [YOUR PHOTO], a typical soft iPhone or Android snapshot. Mildly sharpen, even the exposure a little, and reduce mild noise so it looks like a usable phone photo, not a retouched studio shot. Do not change the face, do not add objects, and do not modernize clothes. Keep identity, pose, and background identical. Natural phone-photo finish, original aspect ratio.

💡

Pro tip: Say 'keep the casual phone-photo look' so it does not turn a snapshot into a headshot.

Rescue a Screenshot of a Photo

12/30

✨ What it does

ChatGPT cleans screenshot pixels and extra softness from [YOUR PHOTO] so the captured image reads more like a real photo. Compare it with the screenshot, then regenerate if it invented detail the screen never had.

Enhance [YOUR PHOTO], which is a screenshot of a photograph rather than the original file. Reduce screen-pixel structure, slight moire, and extra softness from the capture, then recover usable photographic detail. Do not change the face, do not add objects, and do not modernize clothes. Do not invent a sharper original than the screenshot contains. Conservative rescue, original aspect ratio.

💡

Pro tip: If you can get the original Camera Roll file, stop and use that. Screenshots start with fewer pixels.

Fix a WhatsApp-Compressed Send

13/30

✨ What it does

ChatGPT eases WhatsApp blockiness and muddy color on [YOUR PHOTO] so the messenger send becomes usable again. Compare faces with the chat original, then download if compression artifacts are gone and identity held.

Enhance [YOUR PHOTO], a photo crushed by WhatsApp or another messenger. Reduce blockiness, ringing, and muddy color from compression, then gently sharpen existing edges. Do not change the face, do not add objects, and do not modernize clothes. Keep the scene and identity identical. Usable messenger rescue, original aspect ratio.

💡

Pro tip: Ask the sender for the original file when you can. Messenger compression throws away a lot.

Recover a Heavy Digital-Zoom Crop

14/30

✨ What it does

ChatGPT firms up the mushy digital-zoom crop in [YOUR PHOTO] without pretending it was shot on a long lens. Compare it with the zoomed original, then keep the firmer crop if faces still match.

Enhance [YOUR PHOTO], a heavy digital-zoom or pinch-crop from a 12MP phone. Reduce the mushy look, recover edges that are still implied, and keep the crop exactly as framed. Do not change the face, do not add objects, and do not modernize clothes. Do not invent a telephoto original. Conservative zoom rescue, original aspect ratio.

💡

Pro tip: If the uncropped 12MP file still exists, upscale that full frame first, then crop.

Crop and Reframe a Busy Snap

15/30

✨ What it does

ChatGPT crops [YOUR PHOTO] tighter around the main subject and gently sharpens the new frame without adding people or objects. Compare the crop with the original, then download the reframed file if the scene is the same, just tighter.

Reframe [YOUR PHOTO] with a tighter, cleaner crop around the main subject. Improve composition by removing empty or distracting edges only. Do not change the face, do not add objects, do not invent people, and do not modernize clothes. Keep every remaining person and object identical. After the crop, gently sharpen so the new frame looks intentional. Natural photographic reframe.

💡

Pro tip: Say 'do not generate new background at the edges' so the crop trims, it does not outpaint.

Lighting and Exposure

5 prompts

Lift an Underexposed Indoor Shot

16/30

✨ What it does

ChatGPT lifts the underexposed indoor phone shot in [YOUR PHOTO] so faces become readable without flattening the room light. Compare it with the dark original, then download the lifted version if nothing new appeared.

Correct [YOUR PHOTO], an underexposed indoor phone photo. Lift shadows and midtones so faces are readable, without blowing highlights or changing the room's light color. Do not change the face, do not add objects, and do not modernize clothes. Keep pose and scene identical. Natural exposure recovery, original aspect ratio.

💡

Pro tip: Add 'keep the warm indoor tungsten color' if you do not want a cold, over-corrected white balance.

Recover a Backlit Subject

17/30

✨ What it does

ChatGPT opens the backlit silhouette in [YOUR PHOTO] so the face returns while the bright window or sky stays under control. Compare it with the original, then keep the filled take if the person still looks like themselves.

Fix [YOUR PHOTO], a backlit phone photo where the subject is a dark silhouette against a bright window or sky. Recover facial detail in the shadows while holding the bright background so it does not blow out further. Do not change the face, do not add objects, and do not modernize clothes. Keep the backlight direction. Balanced, photographic fill, original aspect ratio.

💡

Pro tip: Say 'hold the window highlights' so it does not turn the background into a white wall.

Rescue a Dark Night Phone Photo

18/30

✨ What it does

ChatGPT lifts the dark night phone photo in [YOUR PHOTO] just enough to see faces without turning the scene into daylight. Compare it with the original, then print if night color and identity still hold.

Enhance [YOUR PHOTO], a dark night phone photo. Lift exposure just enough to see faces and the scene, reduce the worst underexposure, and keep night color (street lights, indoor lamps) believable. Do not change the face, do not add objects, and do not modernize clothes. Do not turn night into daylight. Night-true recovery, original aspect ratio.

💡

Pro tip: Add 'keep it looking like night' so it does not apply a sunny grade.

Soften Harsh On-Camera Flash

19/30

✨ What it does

ChatGPT softens harsh on-camera flash on [YOUR PHOTO] so hot skin glare eases without a beauty filter. Compare faces with the original, then download if the flash look is gone and features stayed put.

Correct [YOUR PHOTO], a phone photo hit with harsh on-camera flash. Soften the hot glare on skin, recover a little shadow behind the flash, and even skin tone without beautifying. Do not change the face, do not add objects, and do not modernize clothes. Keep the scene and identity identical. Natural flash recovery, original aspect ratio.

💡

Pro tip: Add 'do not smooth pores' so flash correction does not become a retouch.

Lift Shadows, Keep Highlights

20/30

✨ What it does

ChatGPT lifts muddy shadows in [YOUR PHOTO] while protecting bright sky, lamps, and white shirts. Compare it with the original, then keep the rebalance if faces opened up and highlights did not blow.

Rebalance [YOUR PHOTO] by lifting muddy shadows while protecting already-bright highlights (sky, lamps, windows, white shirts). Recover face detail in the dark areas only. Do not change the face, do not add objects, and do not modernize clothes. Keep the original contrast character. Controlled shadow lift, original aspect ratio.

💡

Pro tip: Name the bright area to protect ('keep the window from blowing out') for a cleaner lift.

Noise and Compression

5 prompts

Clean Night ISO Noise

21/30

✨ What it does

ChatGPT cleans night ISO speckle on [YOUR PHOTO] while protecting eyes, hair, and fabric so the frame does not go waxy. Compare it with the noisy original, then download if likeness and texture survived.

Denoise [YOUR PHOTO], a night phone photo with heavy ISO noise and colored speckles. Reduce luminance and chroma noise while keeping real detail in eyes, hair, and fabric. Do not change the face, do not add objects, and do not modernize clothes. Leave a little natural grain so it still looks like a photograph. Night denoise, original aspect ratio.

💡

Pro tip: Ask to 'keep a faint grain' so the denoise does not look like a plastic render.

Undo WhatsApp Blockiness

22/30

✨ What it does

ChatGPT smooths WhatsApp 8-pixel blocks and ringing on [YOUR PHOTO] without inventing a cleaner camera original. Compare it with the chat file, then regenerate if faces gained features the blocks never showed.

Repair compression on [YOUR PHOTO], which was sent through WhatsApp and shows 8-pixel blocks, ringing, and smeared color. Smooth the blocks and recover edges that are still implied. Do not change the face, do not add objects, and do not modernize clothes. Do not invent a higher-quality original. Messenger cleanup, original aspect ratio.

💡

Pro tip: Work from the largest messenger size you can export. Tiny chat thumbs have almost nothing left.

Remove JPEG Save Artifacts

23/30

✨ What it does

ChatGPT reduces mosquito noise and banding from the over-saved JPEG in [YOUR PHOTO] without rewriting the scene. Compare it with the original file, then download the cleaner JPEG if nothing extra appeared.

Clean [YOUR PHOTO], a JPEG that has been saved or forwarded too many times. Reduce mosquito noise, banding, and muddy edges from repeated compression. Do not change the face, do not add objects, and do not modernize clothes. Keep the scene identical. Recover only what the file still supports. JPEG cleanup, original aspect ratio.

💡

Pro tip: If you have an HEIC or original Camera Roll copy, upload that instead of the forwarded JPEG.

Fix Screenshot Pixelation

24/30

✨ What it does

ChatGPT removes screen pixels and stair-stepping from the screenshot in [YOUR PHOTO] so it reads more like a photograph. Compare it with the screenshot, then keep it only if you cannot obtain the original file.

Enhance [YOUR PHOTO], a screenshot of a photo that shows screen pixels, slight color banding, and a soft, stair-stepped look. Remove the screen structure and recover a more photographic surface. Do not change the face, do not add objects, and do not modernize clothes. Do not invent detail the screenshot never contained. Screenshot cleanup, original aspect ratio.

💡

Pro tip: A screenshot of a photo is a last resort. Ask for the original before you print.

Denoise a High-ISO Indoor Party

25/30

✨ What it does

ChatGPT reduces high-ISO color noise in the indoor party shot [YOUR PHOTO] while leaving faces and clothes unchanged. Compare it with the original, then download the quieter frame if everyone still looks like themselves.

Denoise [YOUR PHOTO], an indoor party or restaurant phone photo shot at high ISO. Reduce colored noise in shadows and skin while keeping faces, expressions, and clothing exactly as they are. Do not change the face, do not add objects, and do not modernize clothes. Keep the warm indoor light. Usable party photo, original aspect ratio.

💡

Pro tip: Count the people in the prompt so no one gets merged or dropped in a crowded room.

Face and Fine Detail

5 prompts

Sharpen a Face, Keep the Identity

26/30

✨ What it does

ChatGPT sharpens eyes, brows, and mouth in [YOUR PHOTO] without changing age, bone structure, or expression. Compare the face with the original, then regenerate if it invented a different person.

Sharpen the face in [YOUR PHOTO] so eyes, brows, and mouth read clearly, using only existing detail. Do not change the face, do not alter age, bone structure, or expression, do not add objects, and do not modernize clothes. No beautify, no slimmer features, no new teeth or makeup. Identity-true face sharpen, original aspect ratio.

💡

Pro tip: Add 'same person, same age, same expression' if the first result looks like a cousin.

Recover Hair and Eyelash Detail

27/30

✨ What it does

ChatGPT recovers faint hair, eyelash, and fabric detail in [YOUR PHOTO] without inventing a new hairstyle. Hunt for extra strands or lashes, then download if only implied detail returned.

Recover fine detail in [YOUR PHOTO] around hair strands, eyelashes, and fabric weave, using only what is faintly present. Do not change the face, do not add objects, do not invent a new hairstyle, and do not modernize clothes. Keep identity and the scene identical. Subtle, photographic micro-detail, original aspect ratio.

💡

Pro tip: Say 'reconstruct only faintly visible strands' so it does not draw a salon haircut.

Keep Skin Texture, No Beautify

28/30

✨ What it does

ChatGPT clarifies [YOUR PHOTO] while keeping wrinkles, pores, and freckles so nobody gets a beauty filter. Compare skin with the original, then print the honest version if they still look their age.

Enhance clarity in [YOUR PHOTO] but keep every wrinkle, pore, freckle, and skin mark. Do not airbrush, do not change the face, do not add objects, and do not modernize clothes. Sharpen only enough for a usable photo. The person must look their real age. Honest skin, original aspect ratio.

💡

Pro tip: Spell out 'no skin smoothing' on every face prompt. Models default to a glamour pass.

Keep Every Face in a Group

29/30

✨ What it does

ChatGPT sharpens the group in [YOUR PHOTO] so every face is readable without swapping or merging anyone. Count the people against the original, then download once each person still looks like themselves.

Enhance [YOUR PHOTO], a group phone photo. Sharpen and even exposure so every face is readable. Do not change the face, do not merge or swap people, do not add objects, and do not modernize clothes. Keep the exact head count, positions, and expressions. Faithful group enhance, original aspect ratio.

💡

Pro tip: Write the head count in the prompt ('six people') so likeness does not drift in a crowd.

Conservative Soft Selfie Rescue

30/30

✨ What it does

ChatGPT recovers usable clarity on the soft front-camera selfie in [YOUR PHOTO] without slimming or beautifying. Compare it with the original, then keep the rescue if it still looks like you.

Enhance [YOUR PHOTO], a slightly soft front-camera selfie. Recover a usable level of facial clarity without beautifying, slimming, or changing identity. Do not change the face, do not add objects, and do not modernize clothes. Keep the selfie angle, background, and expression. Conservative, still-you result, original aspect ratio.

💡

Pro tip: Add 'do not slim the face or change the jaw' because selfie enhances often reshape.

Free tool

Image Prompt Builder

Build a rich copy-paste prompt for Midjourney, DALL-E, or Stable Diffusion.

Try it free →

Frequently Asked Questions

Open ChatGPT with GPT-4o (or a current image-capable model), upload your phone photo, then paste a prompt and replace [YOUR PHOTO] with a reference to the uploaded image. ChatGPT returns a sharper, cleaner, or larger version in seconds. Pick the prompt that matches the fault (motion blur, missed focus, night noise, WhatsApp compression, backlight) rather than a generic enhance.

It should, if you stay conservative. Every prompt here tells GPT-4o not to change the face, not to add objects, and not to modernize clothes. Image models still rebuild the picture, so faces can drift, especially on heavy blur or tiny crops. Always compare the result with the original. If the person looks like a relative instead of themselves, regenerate and add 'same person, same age, same expression.'

Upload the original Camera Roll or gallery file, preferably HEIC or the full-resolution JPEG, not a screenshot of the photo and not a copy re-downloaded from WhatsApp, iMessage, or Instagram. Screenshots and messenger sends have already thrown pixels away. If the original is gone, use the screenshot or WhatsApp prompts and stay conservative. Do not expect a 12MP crop or a screenshot to print like a camera original.

A clean, uncropped recent phone photo often prints well at 5x7 and can look acceptable at 8x10 after an upscale prompt. A tight 12MP crop, a WhatsApp send, or a screenshot should stay smaller, closer to 4x6, or you will see mush and invented texture. Specify the target size in the prompt, view the result at print size, and reject any version that invented pores, teeth, or a different face.

This page is for soft, noisy, dark, or compressed phone photos you want to make usable: unblur, sharpen, upscale, lift exposure, reduce ISO noise or WhatsApp blocks. The photo restoration page is for damaged vintage prints (tears, scratches, fading, water stains, missing corners). If the photo is an old torn print, use restoration. If it is last weekend's blurry phone pic, stay here. For black-and-white to color, use the colorize prompts.

Browse All Prompt Collections

Popular prompt collections

AI Photo Editing Prompts ChatGPT Jailbreak Prompt Best ChatGPT Prompts Funny Hinge Prompts ChatGPT Caricature Prompts