
↔Portrait Face Recovery
Tap to compare face recovery.
Use a ready-made old photo to new photo AI prompt to repair scratches, recover faded colors, sharpen faces, and recreate damaged memories with natural detail. Copy a prompt, upload your photo, and restore it online with Media.io AI.
Free credits on signup.
Explore five restoration examples, drag the comparison slider, or tap any card to switch between the original photo and the AI-restored result.
BeforeAfter
Recover facial clarity, reduce blur, and preserve the original identity, age, expression, pose, and background.

↔Tap to compare face recovery.

↔Tap to reveal repaired damage.

↔Tap to compare restored color.

↔Tap to reveal recovered scenery.

↔Tap to compare scan cleanup.
Interaction: drag the large slider, choose a restoration, or tap each card to toggle Before and After.
Turn prompt ideas from ChatGPT, Gemini, or Reddit into practical restorations. Media.io helps repair visible damage while keeping faces, clothing, composition, and historical character recognizable.
Use an old photo restoration prompt to remove scratches, dust, stains, tears, grain, faded areas, and small missing details. AI handles multiple defects together, reducing the need for time-consuming manual retouching.
Paste a ChatGPT old photo prompt, refine a Gemini restoration instruction, or adapt a prompt shared on Reddit. Media.io gives you an online image workflow where those instructions can be tested and improved quickly.
Detailed prompts can tell AI to retain facial proportions, age, expression, hairstyle, clothing, and pose. This helps the restoration look like the same person instead of an unrelated modern portrait.
After repairing damage, continue with color restoration, face enhancement, background cleanup, and higher-resolution output. The connected Media.io workflow is useful for family albums, genealogy projects, memorials, and archival sharing.
Choose a scanned family portrait, faded wedding photo, torn print, blurry childhood picture, or black-and-white image. For better results, use the clearest scan available.
Copy an old photo to new photo prompt below. Describe the damage to repair and specify what must stay unchanged, including identity, age, expression, clothing, pose, composition, and historical style.
Generate the restored version, compare it with the original, and refine the prompt if facial details or colors change too much. Download the result or continue enhancing it with Media.io.
Start with the prompt that matches your photo. Each old photo prompt can be used as a ChatGPT or Gemini prompt, then customized with the subject, time period, damage, and preferred output.
Best for family portraits with scratches, dust, fading, grain, and minor tears.
Prompt: Restore this old photograph by removing scratches, dust, stains, cracks, grain, and faded areas. Repair damaged edges and reconstruct small missing details. Sharpen faces and clothing naturally while preserving the exact identity, age, expression, pose, original composition, and historical character. Produce a clean, realistic, high-resolution restoration.
Use this prompt to add natural color without losing the original period look.
Prompt: Colorize this vintage black-and-white photo with realistic, historically appropriate skin tones, hair, clothing, objects, and background colors. Preserve the original face, age, expression, lighting direction, composition, and film character. Avoid oversaturation, modern styling, plastic skin, or invented accessories.
Improve soft faces while limiting unwanted identity changes.
Prompt: Enhance this blurry old portrait by recovering natural detail in the eyes, eyebrows, lips, hair, skin, and clothing. Reduce blur and noise while preserving the exact facial proportions, identity, age, expression, hairstyle, pose, and background. Do not beautify, reshape, de-age, or replace the person.
Reconstruct ripped edges, missing corners, and damaged backgrounds.
Prompt: Reconstruct the torn and missing areas of this photograph using the visible image as reference. Repair ripped edges, cracks, folds, stains, and missing corners. Continue clothing, background, lighting, and textures consistently while preserving every visible person and the original composition.
Recover washed-out prints from older film cameras and albums.
Prompt: Revive this faded color photograph by restoring balanced contrast, accurate white balance, natural saturation, and fine texture. Remove yellowing, color casts, dust, and mild grain. Preserve the original people, clothing, environment, lighting mood, and nostalgic film appearance.
Prepare a restored memory for albums, frames, or family gifts.
Prompt: Create a print-ready restoration of this old photo. Repair visible damage, recover facial and fabric detail, improve tonal range, and upscale the image cleanly. Keep natural texture and subtle film grain. Preserve identity and historical authenticity; avoid over-sharpening, halos, artificial skin, or newly invented details.
A strong old photo to new photo AI prompt names the visible damage, the restoration goal, and the details that must remain unchanged. Include scratches, dust, fading, tears, blur, grain, or missing areas as needed. Then request preservation of identity, age, expression, clothing, pose, composition, lighting, and historical character. This gives the model clearer boundaries and reduces unwanted redesigns.
Yes. ChatGPT can help you write or refine an old photo restoration prompt, while Media.io provides the image workflow for generating and comparing the restored result. Copy the prompt, customize it for the damage in your photo, and state clearly that the person's identity and original composition must be preserved. You can then adjust the wording if the first output changes too much.
The best prompt for old photo restoration in Gemini is specific rather than simply saying “make this photo new.” Describe each correction, such as removing scratches, repairing tears, restoring faded tones, sharpening faces, and reconstructing small missing details. Add constraints such as “preserve the exact identity, age, expression, clothing, pose, framing, and period style” and “avoid beautification, face replacement, modern accessories, or invented details.”
Reddit prompts can be useful starting points, but results vary with the image, model, and wording. Check whether a shared prompt explains what should remain unchanged, not only what should be enhanced. Avoid copying instructions that aggressively beautify, de-age, replace faces, or modernize clothing. Test the prompt on a duplicate file and compare the output carefully with the original photo.
AI can repair many scratches, tears, faded areas, and small missing regions, and it can add plausible color to black-and-white photos. However, heavily damaged or hidden details may be inferred rather than recovered exactly. To preserve faces, request the exact identity, facial proportions, age, expression, hairstyle, and pose to remain unchanged. Generate more than one version and verify important family or historical details before printing or archiving the result.