Pick — AI photo restoration: repair, colorize, and upscale old photos on iPhone

What is Pick?

Pick is an iOS app that uses AI to restore old photos. Repair scratches and creases, colorize black-and-white prints, sharpen blurry shots, and upscale low-resolution originals to 4K — all from your phone, with professional-quality results and no editing skills required.

The use case is specific: your family's scanned albums, your grandmother's wedding photos, the Polaroid that survived a water-damaged box, the faded childhood portrait you want to reprint for a birthday. Photos that mean something, brought back to life.

The problem

Old photos deteriorate. Colors fade, surfaces scratch, emulsion cracks, detail smears. The family history you wanted to preserve sits in a shoebox and gets worse every year.

The options for fixing them used to be:

• Send them to a professional retoucher — $50-$200 per photo, 1-2 week turnaround
• Learn Photoshop's healing brush + clone stamp — a steep skill curve for a task you'll do rarely
• Use a consumer AI tool on a desktop — works, but requires scanning and desktop workflow
• Accept that they'll never look better — the quiet default for most people

The gap: a mobile, one-tap, high-quality restoration tool that works on a photo you just took with your phone's camera pointed at a print.

What it does

• AI photo restoration — repairs scratches, tears, spots, stains, and physical damage
• Colorization — adds realistic, period-appropriate color to black-and-white photos
• Enhance and unblur — sharpens out-of-focus shots, enhances facial detail
• Upscale to 4K — increases resolution of small or grainy originals for print
• Before/after view — side-by-side compare with the original
• Batch processing — restore multiple photos in one session

How it works

Pick uses a combination of on-device and cloud AI models. The restoration, colorization, and upscale steps are handled by a tuned pipeline that identifies what kind of damage is present (scratches vs. blur vs. fade) and applies the appropriate model. You select a photo from your camera roll or snap a new one of a print, tap the action you want, and the result comes back in a few seconds.

The cloud side uses short-lived, encrypted uploads for the processing step only. Photos are not stored, not used for training, not shared. You can read the privacy policy in-app before processing your first photo.

The before/after slider is there so you can judge the result honestly — AI restorations sometimes overreach (over-sharpening, unrealistic colorization), and Pick shows you what it did instead of trying to hide it. If a result isn't right, you re-run with different intensity settings.

Who it's for

• Family historians scanning and preserving old albums
• Anyone who inherited a box of family photos from a grandparent
• Genealogists digitizing archives
• Reprint / framing shops offering a restoration add-on
• People recovering photos from water, fire, or time damage

Real scenarios

Grandmother's 90th birthday. You want to print a framed photo of her from her twenties. The original is sepia, faded, with a tear across the corner. Pick repairs the tear, restores the faded areas, colorizes gently, and upscales the scan to print at 12×16 without blur.

A water-damaged album. A flood got into a photo album. Most photos are blurred, stained, or missing emulsion. Pick processes them in a batch — restoration + light colorization — and recovers enough detail to identify faces that were hard to read before.

Digitizing a grandparent's archive. You're scanning a drawer of 70-year-old prints with your phone camera (no flatbed scanner needed — phone cameras are now good enough). Pick cleans each one up automatically as you go. The project that would have taken a month of Photoshop takes an evening.

Why I built it

This one was personal. A family member asked me to "fix" an old scanned photo before framing it for a birthday — and the available tools were either expensive, slow, or required desktop skills they didn't have. The surprise was that the AI models available in 2025 are dramatically better than what was possible even two years ago. Restoration that used to require a retoucher's careful hand now runs in seconds from a phone.

The opportunity: wrap that capability in an interface a non-technical family member can use in under a minute, and make the privacy guarantees clean enough that people feel safe uploading their most personal photos.

Alternatives and how it compares

• Remini: closest competitor. Pick differs in pricing (a one-time unlock + pay-per-batch option vs. subscription), and in being more conservative with colorization defaults (we'd rather under-color than over-color).
• Photoshop Neural Filters: very capable, requires a CC subscription and desktop. Better for power users; worse for the mobile use case.
• Traditional retouchers: higher ceiling on quality, radically higher cost and slower.

FAQ

Are my photos used to train AI? No. Photos are uploaded only for the duration of processing, not stored, not shared, not used for training.

How realistic is the colorization? Good but not magic. The AI infers plausible colors — skin tones, grass, sky, clothing. For period-specific colors (a dress you know was red, not blue) you can re-run with a color hint.

Will it work on very damaged photos? It helps a lot, but there's a limit. If half the photo's emulsion is gone, Pick restores what's there — it doesn't invent what isn't.

Does it work offline? Light edits work offline. The heavy restoration and upscale steps require a network connection because the models are too large to ship on-device.

Try Pick

Get on the App Store →

Pick is shipped under DRISH LABS — see the full catalog for every other app.