How to Scan Old Family Photos (Phone or Flatbed, Done This Weekend)

The first photo I pulled out of my grandfather's shoebox was a 3.5-by-5-inch print of a man leaning against a railway carriage, 1948 by the penciled date on the back. The print had curled into a tube over forty years in a dry attic, the emulsion had developed a faint amber cast, and the pencil note on the reverse read, in his handwriting, "Walter, before the transfer." None of that showed up in a casual glance at the front. All of it mattered.

That is the case for scanning in one sentence: the physical print is deteriorating right now, and the information on it, including the writing on the back, exists nowhere else. The bad news is that the internet has turned digitizing into a research project involving archival gloves and 600 DPI debates. The good news is that a shoebox of 500 photos is a weekend of work, not a sabbatical.

Your phone covers most of the box

A modern phone camera resolves far more detail than a 1950s snapshot contains. A 3.5×5 print scanned at 600 DPI is 2,100 by 3,000 pixels, about 6 megapixels, and any phone made in the last six years shoots 12 megapixels or more. The camera is not the bottleneck. Technique is.

Use a dedicated capture mode rather than the plain camera where you can. Google's PhotoScan, free on both Android and iOS, takes several overlapping shots per print and stitches them computationally, which is how it removes the glare that otherwise ruins a glossy print. On iPhone, Apple's built-in document scanning works too; it lives in Notes and Files and surfaces in other apps depending on your iOS version. Whichever you use, the setup matters more than the app:

• Indirect daylight, near a window, flash off
• The print on a plain matte surface in a contrasting color
• Shot straight down, print filling the frame, tap to lock focus
• One print per shot until you get fast; batch cropping costs more time than it saves

Photograph the backs as a separate image, always. The back of a print is often more informative than the front: pencil names, dates, a photographer's studio stamp with a city, a stamp box that dates the paper stock itself. My Walter photo is only identifiable because I turned it over.

When the flatbed is worth the setup

A flatbed scanner justifies its setup in three situations. Tiny prints, the kind smaller than a matchbook, because enlargement is the whole point and you want every available pixel. Prints that have faded into a heavy magenta or orange cast, where a scanner's software can often restore tonal range better than a single camera exposure. And anything you plan to reprint large, since the printing options you choose later determine the resolution worth capturing now.

On the resolution question, the numbers resolve it quickly. Printing at original size needs about 300 DPI, which is roughly the limit of what inkjet paper and human eyesight resolve at reading distance. A 3.5×5 print at 300 DPI is 1,050 by 1,500 pixels, plenty for a reprint. The calculation changes when you enlarge. Blowing that same print up to 8×10 means the pixels get stretched 2.7 times across, and 300 DPI at output would require 2,400 by 3,000 pixels from the scan. Scan at 600 DPI and you capture 2,100 by 3,000, which covers the enlargement. So: 300 DPI for reprinting what you have, 600 DPI for anything small that descendants might want on a wall.

One caution on scanner software. If you own an old Canon or Epson flatbed whose driver no longer installs on your current computer, VueScan is the standard workaround. It talks to thousands of long-discontinued scanners directly and is a one-time purchase rather than a subscription.

Storage math, so the project does not die on a full disk

Assume 500 prints at 600 DPI, saved as high-quality JPEG. Each file lands around 4 to 6 MB, so the full session is 2 to 3 GB. That fits in the free tier of Google Photos or any 32 GB USB drive with room to spare. If you scan to TIFF for maximum quality, multiply by four or five: the same 500 photos becomes roughly 10 GB, which is still trivial on a 256 GB laptop but stops being trivial on a phone.

The habit that matters is not the volume, it is the duplication. Keep full-resolution originals in one folder and never edit them; edit copies. A practical minimum is two copies on two devices, one of them off-site. Cloud storage counts as off-site. Digital archives survive by being boring and redundant, not by being curated.

The failure modes worth knowing in advance

Curled prints will not sit flat, and forcing one against scanner glass can crack old emulsion. Flatten them under a stack of books for a day first. If a print stays stubbornly curved, photograph it slightly curved rather than pressing it flat; a shadow at the edge beats a cracked image.

Magnetic albums, the self-adhesive kind with a peel-back plastic sheet, are the worst object in most family collections. The adhesive yellows with age and the plastic off-gasses, which is why prints in them often show orange stains. Lift prints gently with dental floss worked under a corner, or a thin spatula. If a print genuinely resists, photograph the whole album page as-is. A slightly degraded capture beats a torn print.

Handling advice has also drifted from folklore. Most conservators now recommend clean, thoroughly dried hands over gloves. Cotton gloves reduce your grip, which is exactly how prints get dropped, and the fibers catch on flaking edges. If you want gloves, thin nitrile ones give grip and keep finger oils off the emulsion. What actually damages prints is humidity, not skin contact, so the box belongs in a living space with stable temperature, not the attic.

Faded color needs a light touch. Those 1970s orange shifts are partly dye degradation in the print itself, and aggressive auto-correction produces skin tones that look radioactive. Correct a copy, keep the original file untouched, and accept that some casts are permanent.

Naming while you still remember

Name files at the point of scanning, roughly by content. 1948-walter-railway.jpg is searchable in ten years; IMG_4021.jpg is not. You will not remember what a batch of numbered files contains, and neither will whoever inherits the drive. This is the stage where scanning hands off to sorting, and the fuller workflow is covered in our guide to organizing old family photos.

Turning scans into something people will actually read

A folder of 500 well-named scans is still just a shoebox with better lighting. The value appears when faces get names and names get connected, which is where a photo-first tree tool fits. Photo Family Tree runs entirely in your browser. You upload the scans, click each face to record who it is, and the tree assembles the relationships around them. Two properties make it suited to this specific job. Nothing is uploaded to a server, so the scans never leave your device, which matters for family photos you are not ready to circulate. And the export is one high-resolution PNG rather than a project file, so you can send it to a sibling and get corrections back in the same week. If you want to see the range of what a finished tree looks like before you start, there are 19 starting points on the templates page.

That correction loop is the fastest way to finish an archive. My grandfather labeled maybe forty of the five hundred prints. My aunt, looking at the first tree draft, corrected six of them, including proving that the man at the railway carriage was not Walter at all but Walter's older brother, who died in 1951 and whose name I had never heard until that phone call.

The weekend, concretely

Saturday morning, flatten the curled prints under books and scan the whole stack, front and back, however imperfectly. Saturday afternoon, name the files and make a second copy. Sunday, sit with the oldest relative you can reach, put the scans on a shared screen, and write down every name they give you while they are still on the line. Handle the originals once, return them to the box, and do not try to organize during the capture pass. By Sunday night the shoebox exists in two places, and the names exist in one that does not fade.