6 AI Prompts That Can Restore Old Photos in Seconds

A practical guide to breathing life back into faded, torn, and time-worn photographs, no professional software or editing skills required.

Old photographs are time capsules. They hold the way your grandmother used to smile, the texture of a coat your grandfather always wore, a childhood birthday party you’ve nearly forgotten. But time is cruel to physical prints. Fading, yellowing, scratches, tears, water damage, slowly, the memories erode.

For decades, professional photo restoration meant hiring a skilled graphic designer and spending significant money. That changed almost overnight. Today, AI tools like ChatGPT, Google Gemini, and similar platforms can restore, colorize, sharpen, and rejuvenate old photographs using nothing more than plain written instructions, which we call prompts.

The secret is knowing how to write those prompts. Vague instructions give vague results. The 6 prompts in this guide have been crafted with precision, pulling together the best techniques from image restoration, professional photography, and AI communication. Each one targets a specific problem old photos tend to have, and each one can be used by anyone, with no technical background whatsoever.

How to use these prompts

• Open ChatGPT, Google Gemini, or any capable AI image tool on your phone or computer.
• Upload your old photo by clicking the attachment or image icon in the chat interface.
• Copy one of the prompts below and paste it into the message box.
• Hit send and let the AI work. Review the result, and if needed, try a second prompt to refine it further.
• Download the restored image directly from the AI’s response.

The complete one-shot restoration

If you only try one prompt from this entire guide, make it this one. I designed the Complete One-Shot Restoration prompt to address every common problem old photographs suffer from in a single instruction. It covers sharpness, color fading, physical damage, noise, grain, and resolution, all in one go. 

This prompt works exceptionally well when you have an old photo but are not sure what specifically is wrong with it. Rather than guessing which individual problem to tackle, you let the AI analyze the whole image and apply a full suite of corrections. The instruction to preserve identity is particularly important; it keeps the AI from improving a face in ways that make the person unrecognizable. For first-time users, this prompt is both the safest and most impressive place to begin.

The prompt: 

“Restore this photograph comprehensively and professionally. Analyze every visible form of damage or degradation, including scratches, tears, creases, stains, fading, discoloration, blur, noise, and grain and correct each one carefully. Sharpen soft edges and facial features naturally, without over-processing. Repair or reconstruct any missing areas by intelligently matching the surrounding textures, colors, and tones. Where colors have faded, restore them to vibrant, realistic values that suit the original era of the photo. Adjust brightness and contrast so that all areas of the image are clearly visible, with no blown-out highlights or lost shadow detail. Preserve the complete facial identity and natural likeness of every person in the image without alteration. Upscale the final output to the highest possible resolution — Full HD or above — with a clean, photo-realistic finish that looks like a professionally restored print.”

Professional portrait revival

Portrait photographs deserve a specific type of treatment. The goal is not just to fix damage but to produce an image that looks as if it were taken by a professional photographer using high-end equipment.

This prompt does exactly that by referencing top-tier camera systems. Mentioning specific cameras, such as the Canon EOS R5 or Sony A1, is not random; it is a deliberate technique that helps the AI understand the visual standard you are aiming for. These cameras are known for producing images with exceptional skin-tone accuracy, natural depth of field, and crisp yet non-clinical sharpness.

The result of this prompt is often the most striking of the six, particularly for old family portraits where the subject’s face is the emotional core of the image. The AI rebuilds skin texture, eye clarity, hair detail, and clothing definition with careful precision, while the overall image takes on the warmth and dimensionality of a modern professional photoshoot. It is the closest thing to sending a photo back in time with a better camera.

The prompt: 

“Restore and enhance this portrait photograph to the visual quality standard of a professional studio shoot captured on a Canon EOS R5 mirrorless camera. Remove all visible damage including scratches, tears, grain, noise, blur, and discoloration. Restore accurate, natural skin tones with lifelike texture — pores, fine lines, and natural variations should be visible but refined. Sharpen the eyes to achieve a natural sparkle and clarity without artificial enhancement. Define hair strands and clothing fabric naturally. Do not alter the person’s actual facial structure, age, or identity in any way — restore what is already there, do not recreate or beautify. Correct the color balance so that whites are white, shadows retain detail, and the overall palette feels warm and naturally lit. Upscale to a minimum of 4K resolution with smooth tonal gradations and a finish consistent with a high-end professional portrait print.”

Colorizing black and white photos

One of the most emotionally powerful things you can do with an old photograph is colorize it. When a black-and-white photo suddenly shows the blue of someone’s eyes, the warm brown of a wooden kitchen table, or the dusty green of a military uniform, it transforms from a historical document into something that feels alive and present. 

Colorization is also one of the most technically delicate operations to get right. Done poorly, it looks garish and artificial. This prompt prioritizes authenticity over drama. It asks the AI to use knowledge of the era’s typical colors, natural lighting conditions, and real-world skin tones to apply colors that feel genuinely plausible rather than invented. 

The prompt:

“Colorize this black and white photograph with careful attention to historical accuracy and natural realism. Use your knowledge of the photo’s approximate era to apply period-appropriate colors to clothing, furnishings, vehicles, and environmental elements, for example, muted earth tones and wool fabrics for early 20th century, or brighter synthetic colors for mid-century modern periods. Apply realistic, natural skin tones to all people in the image, with subtle variations for light sources and shadow. Preserve all shadows, highlights, and tonal depth from the original monochrome image; do not flatten the lighting when adding color. Ensure the sky, foliage, and architectural elements receive colors that are plausible for the scene’s location and season. The final colorized image should feel as though the original photographer had a color camera, not as though a digital painting was applied on top of a photo. Clean up any existing scratches, grain, or damage during the process and output in the highest available resolution.”

Repairing physical damage

Physical trauma, like water stains, deep creases from being folded in a wallet, or actual tears, can make a photo feel lost forever. This prompt treats the AI like a digital surgeon. It specifically instructs the tool to sample surrounding pixels to heal gaps and smooth out rough patches left by time and handling.

It goes beyond just fixing the image; it asks the AI to intelligently fill in the blanks. If a corner of a jacket is missing or a crease runs through a background, the AI uses its knowledge of shapes and patterns to rebuild those elements seamlessly.

The prompt: 

“Act as a professional photo conservator and repair all physical damage in this image. Specifically, target and remove deep creases, water stains, surface scratches, and frayed edges. Rebuild any missing portions of the composition by intelligently sampling the surrounding environment and textures. Ensure the repairs are invisible and do not distort the original subjects. The final image should be smooth, clean, and appear as though it has never been damaged.”

Lighting and exposure correction

Old photos are often either blown out (too bright) or muddy (too dark). This usually happened because of poor flash bulbs or developing errors. This prompt asks the AI to re-light the scene. It digs into the shadows to uncover hidden details and dials back the highlights to reveal textures previously invisible.

Think of this as an HDR (High Dynamic Range) upgrade for a 50-year-old photo. It balances the exposure so that everything from the dark corners of a room to the bright sky outside a window looks perfectly exposed.

The prompt:

“Optimize the lighting and contrast of this image to reveal hidden details. Brighten underexposed or shadowy areas, especially faces, without introducing grain. Simultaneously, recover details in overexposed or ‘washed out’ sections. Adjust the overall contrast to add depth and dimension, making the subjects stand out clearly. The final output should have a balanced, naturally lit appearance as if it were professionally lit in a studio.”

The deep detail recovery prompt

This is the most comprehensive and ambitious prompt in the guide. The Deep Detail Recovery Prompt is designed for photographs that are in the worst condition, severely faded, heavily damaged, blurry, and possibly quite small to begin with. It combines every restoration technique covered in the previous five prompts into a single unified instruction, creating a complete pipeline from damage assessment all the way through to final finishing. Think of it as the master restoration session that leaves nothing on the table.

What makes this prompt different from simply combining the others is the sequencing and emphasis on layered restoration. It asks the AI to work through the image systematically: first assessing and cataloging the damage, then repairing structural issues, then recovering tonal and color information, then enhancing fine detail, and finally applying a finishing grade.

Each stage builds on the last. The explicit instruction to preserve identity while also performing the most aggressive possible enhancement reflects a key tension in AI photo restoration: the AI will do things better if given license to be thorough, but identity must remain sacrosanct. This prompt navigates that tension directly. For truly precious photographs in dire condition, this is the one to use.

The prompt:

“Perform a full, multi-stage deep restoration of this photograph, treating it as a professional archival restoration project. Work through the following stages systematically: 

Stage 1 — Damage Assessment: Analyze the entire image and identify every form of degradation present, including physical damage (tears, scratches, folds, stains), chemical deterioration (fading, yellowing, discoloration), optical issues (blur, grain, noise), and resolution limitations. 

Stage 2 — Structural Repair: Reconstruct all physically damaged or missing areas using intelligent context-aware inpainting, ensuring all repairs are completely seamless and match the surrounding visual content. 

Stage 3 — Tonal and Color Restoration: Restore the full dynamic range of the image — recover shadow detail, reduce blown-out highlights, and rebalance the overall exposure. Correct all color casts and restore accurate, natural colors throughout, particularly for skin tones, foliage, sky, and fabrics. 

Stage 4 — Detail Enhancement: Apply targeted sharpening to all areas of the image, with greatest precision on human faces. Restore fine details including hair, eyelashes, skin texture, clothing weave, and background elements. Reduce all noise and grain while preserving natural photographic texture. 

Stage 5 — Super-Resolution Upscaling: Upscale the fully restored image to the highest resolution available — targeting 4K or above — using AI super-resolution to reconstruct fine detail consistent with the image content. 

Stage 6 — Final Finishing: Apply a warm, natural finishing grade that makes the image feel beautiful and lifelike. Introduce a very subtle vignette and ensure color richness and tonal depth throughout. Throughout all stages, preserve the complete identity and likeness of every person in the photograph without alteration. Deliver the final result as a pristine, print-quality image that looks professionally restored.”

Memory is worth the effort

Every old photograph sitting in a drawer right now was once someone’s most cherished new memory. A wedding. A graduation. A Tuesday afternoon that felt ordinary at the time. The people in those photographs were real, their lives were vivid and full of color, and they deserve to be seen clearly.

These six prompts are tools, but what they serve is something much older and more human than technology. They serve remembrance. They serve the quiet act of not letting people disappear from our families’ stories just because the print has faded. Pick the photo that matters most. Start with Prompt 1. See who comes back.

For more AI prompt ideas, check out our guide to the best Google Gemini prompts for job searching, resume writing, and career planning in 2026.

The post 6 AI Prompts That Can Restore Old Photos in Seconds appeared first on eWEEK.