Common Mistakes When Scanning Family Photos

The back of an old print is the single richest source of dating and context information you have access to: handwritten dates, names, locations, occasions, and faded developer timestamps. If you only scan the front, you throw all of that away — permanently, because you’re probably not going to feed every print through the scanner a second time.

Always scan both sides, even when the back looks blank. Many scanners have an auto-detect mode that only saves the back when text is visible — turn that off and force-save every back. Light pencil notes and faint timestamps are exactly the ones an auto-detector misses, and they’re the ones AI dating tools can still read. See our scanning best practices guide for the exact settings on the most common feed scanners.

300 DPI looks fine on screen at the print’s original size. Then your grandkid wants to enlarge a 4×6 of their grandmother to hang on a wall, or a restoration tool needs more pixels to repair a tear, or an AI upscaler tries to read the timestamp in the corner. At 300 DPI there’s nothing left to work with.

Scan at 600 DPI. Storage is cheap, hard drives keep getting bigger, and the prints themselves are not getting any younger. The professional archival standard is 600 DPI for prints and 1200–2400 DPI for negatives or slides. You only get one chance to do this with each photo — do it at the resolution you’ll wish you’d picked in 20 years.

PNG is a great format for screenshots. It is a poor format for archival photo scans, and many photo libraries treat PNG files as second-class citizens — some don’t even read or write the EXIF date fields you’ll need later to put photos in chronological order.

Use JPG for most families — small files (about 2 MB each at 600 DPI), universal metadata support, and indistinguishable from the original print at high quality settings. Use TIFF if you plan to do further editing — lossless, larger (about 20 MB each), still excellent metadata support. Skip PNG.

Every speck of dust on the glass becomes a black spot on every photo for the next hour. Every smear becomes a streak. You won’t notice while you’re scanning at speed — you’ll notice when you’re looking at the photos six months later and every print from the third stack has the same fingerprint in the upper left.

Wipe the glass with a microfiber cloth and a small amount of isopropyl alcohol before each session. Use a soft brush or a rocket blower to dust the prints themselves — especially black-and-white prints, which show dust most ruthlessly. For feed scanners, run a cleaning sheet through every few hundred photos.

Feed scanners like the Epson FastFoto and Kodak Scanza are fast, which makes it tempting to load them like a printer paper tray and walk away. Then two prints stick together, the scanner skips one, and that 1973 birthday photo simply never appears in your archive — and you have no way of knowing it’s missing.

Load smaller batches (5–10 prints), fan them slightly so they don’t cling, and watch the count of saved files match the count of prints you fed in. It’s slower per session, but you don’t lose photos you’ll never recover.

If batch one is named scan_0001.jpg, batch two is named IMG_0001.jpg, and batch three is named fastfoto_2024_05_09_0001.jpg, you have just made every downstream tool — including AI dating, including front/back pairing, including future you sorting on a Sunday afternoon — significantly harder.

Pick one naming convention and use it for the entire project. A practical one: <album-or-box-label>_<sequence>.jpg, e.g. Jul89Sept90_0001.jpg. The album label gives you a known time window per batch, the sequence number keeps front/back pairs together, and you can recognize at a glance which physical box a file came from.

If you dump the entire shoebox into one stack and feed it through in random order, you have lost a huge amount of context that was free for the taking: photos that were stored next to each other in an album are usually from the same event, the same trip, or the same year. AI dating tools (and your future self) can use that adjacency as a strong signal.

Scan one album, sleeve, or grouping at a time. Keep the order the prints were stored in. Save each grouping to its own folder. The original physical organization is the closest thing to ground truth you’ll ever get.