Genealogy Photo Restoration: The Complete Family History Guide

Quick Answer

End-to-end guide to restoring photographs for family history and genealogy projects — digitization, AI restoration, preservation.

In 30 seconds:

• Digitize every photograph at 600 DPI TIFF before restoring — the scan is the permanent archival record.
• Identify the photo format first (tintype, ambrotype, cabinet card, gelatin silver) — each has distinct handling needs.
• Use AI restoration in batch for collections of 300+ images — manual work does not scale to family archives.
• Capture metadata alongside every scan: names, dates, locations, source — unlabeled photos are half-documents.
• Store TIFF masters on two physical drives plus cloud, with JPEGs for sharing — follow the 3-2-1 backup rule.

Best AI tool for this job: ArtImageHub — one-time $4.99, no subscription. Processes in 30 seconds, no signup, handles cracked ambrotypes, faded tintypes, water-stained gelatin silver prints, torn family portraits.

When to hire a specialist instead: Daguerreotypes, glass-plate negatives, or one-of-a-kind 19th-century heirlooms that require physical conservation before any digital work.

---

Genealogy work and photo restoration arrived at the same place from two different directions. The family historian wants a name and a date attached to a face. The photo restorer wants a face clear enough to recognize. Put them in the same room and they discover they have been chasing the same object: a photograph that survives long enough and in good enough condition for the next generation to say, "that is Great-Grandma Ida in 1908."

I have worked on roughly eight thousand family-history collections over the last decade, and the honest truth is that the restoration step is not the hard part. The hard part is the decision-making before you ever touch a scanner — what to prioritize, what to keep as an original, what to discard after digitizing, how to annotate things so the archive is still legible to your grandchildren. This guide is how I approach a genealogy collection end-to-end, from the cardboard box somebody hands me to the shareable digital archive that gets distributed to forty cousins at a reunion.

If you already have scans and just want to run them through restoration, skip to the old photo restoration workflow and come back for the organizational sections later. If you have a shoebox of tintypes and no idea where to start, stay here. The upstream decisions matter more than any single restoration pass.

Why Genealogy Photos Are a Different Project Than "Old Photos"

A generic old-photo restoration assumes one image on a scanner at a time. Genealogy is different in four specific ways, and if you approach it with the single-image mindset you will either waste weeks of effort or lose information you can never recover.

1. Scale. A typical family-history collection is 300 to 2,000 images across multiple generations, not a single print. The per-image process has to be fast enough that 1,000 units is a weekend, not a year.
2. Provenance. The image without the caption is half a document. Who is in the photo, when, where, and why — this is the actual historical record. A beautifully restored photo of an unidentified person is pretty; an annotated, medium-quality scan of the same photo is useful.
3. Mixed media and eras. You will have tintypes from the 1870s next to Kodachromes from the 1960s next to inkjet prints from 2002, all in the same album. The restoration approach differs for each.
4. Stakeholders. Unlike a personal photo project, a genealogy collection belongs to an extended family. Decisions about originals — keep, redistribute, donate — involve aunts you have not spoken to in five years.

The decisions that come out of these four differences shape everything downstream. Batch scanning with metadata discipline beats slow perfectionism. Dating by photographic process beats guessing by clothing. A shared cloud archive beats the "one uncle has the good box" problem. And — this one is important — restoring a scan does not replace preserving the original. The original carries physical evidence (paper stock, photographer's imprint, handwritten notes on the reverse) that a digital restoration cannot capture.

The Four Kinds of Heirloom You Will Find in the Box

Before you do anything, sort the box into four buckets. Each has a different workflow.

Cased Images (1840s–1860s)

Daguerreotypes, ambrotypes, and tintypes in hinged leather or thermoplastic cases. These are one-of-a-kind — there is no negative, the case is the object. Never remove a daguerreotype from its case; the silver surface tarnishes in minutes once exposed. Scan these through the protective glass as-is, and treat the case itself as part of the artifact. The techniques I cover in vintage photo repair techniques go into the handling specifics, and the process-dating in how to identify old photo formats will help you sort tintypes from ambrotypes when the labels have worn off.

Card-Mounted Prints (1860s–1910s)

Cartes de visite, cabinet cards, and their larger cousins. These are albumen or gelatin prints glued to a card stock that often carries the photographer's studio mark on the reverse. The mark is a goldmine — photographer names and addresses can be cross-referenced to city directories to pin down a date window to within three or four years. Always scan both sides. Do not trim or remove the print from its mount under any circumstance; the mount is evidence.

Loose Paper Prints (1920s–1990s)

The working bulk of most family collections. Black-and-white silver gelatin through the 1950s, color C-prints from there forward. These are what most people mean when they say "old family photos." They respond well to AI restoration and scan quickly at batch rates. Our guide on how to enhance old family portraits is the specific workflow for the face-forward shots in this era, which tend to dominate what survives.

Slides, Negatives, and Digital (1950s onward)

Kodachrome and Ektachrome slides through the 1970s, then color negatives and consumer digital. Slides and negatives require a scanner that handles transparencies — not all flatbeds do — and they hold far more information than the paper prints made from them. When you find a box of negatives, treat it as a primary source, not a duplicate. Many family photos that survive only as faded 3×5 prints have a perfectly exposed negative sitting in an envelope nearby.

Dating a Photo When There Is No Caption

Undated photos are the norm, not the exception, and half the genealogy payoff is pinning them down. You do not need to be a photohistorian — you need a small, reliable set of dating cues.

• Process. The photographic process itself dates a photo to a window. Daguerreotypes: 1840–1860. Tintypes: 1855–1900 (with a long tail into the 1930s at carnivals). Cabinet cards: 1866–1920. Kodachrome: 1935 onward, with the characteristic red saturation a telltale. The how to identify old photo formats reference has the process-by-process field guide.
• Mount style. Cabinet card mount colors and edge treatments follow strict fashion cycles. Beveled gold edges on thin cream stock: 1880s. Dark green or maroon with a matte finish: 1890s. Embossed scalloped edges: 1900–1910.
• Photographer's imprint. Cross-reference the studio address against historical city directories (available through most local historical societies and several free genealogy databases).
• Clothing and hairstyle. Less reliable than process, because rural and immigrant communities lagged fashion by five to fifteen years. Use as a tiebreaker, not a primary signal.
• Photographic paper stock. Postwar Kodak borders with scalloped edges and month-year dates printed on the white frame are a gift — you can read the exact month of printing from the border code.

Dating narrows your restoration approach too. A 1910 albumen print has different failure modes than a 1975 C-print, and AI restoration tuned for the color-faded modern era will misread the warm sepia of the albumen as damage to correct rather than intended tone.

Cataloging Before You Scan

This is the step people skip, and it is the step that turns the project from a hobby into a usable archive. You need a catalog. Not fancy — a spreadsheet is fine — but it has to exist before the scanning volume gets past about 50 photos.

My catalog columns, in order of importance:

1. Unique ID (GEN-0001, GEN-0002…). Zero-padded so they sort correctly.
2. Source (which relative, which album, which box).
3. Estimated date (decade if you do not know the year).
4. People identified. Use full names, not nicknames, even if it feels redundant.
5. Location if known.
6. Event if known (wedding, christening, vacation, reunion).
7. Notes on damage / handling constraints.
8. Physical storage location after you are done.

The unique ID becomes the filename for every scan (GEN-0001.tif for the scan, GEN-0001.jpg for the restored web version). File names are forever; spreadsheet columns can evolve. If you invest in clean filename discipline on day one, you can migrate the spreadsheet into a proper database later without renaming a thing.

For the scanning workflow itself, my how to digitize old photos walkthrough covers equipment, color targets, and handling. The scale-specific version — how to batch a box of 400 — is in how to digitize a large photo collection, which is where the metadata discipline really earns its keep.

Scanning: Resolution Choices for Family-History Work

Genealogy has a different resolution calculus than single-photo restoration, because you want the archive to outlive the current monitor, the current print size fashion, and your current software. Overscan now, because rescanning 800 prints in ten years is not going to happen.

• Standard 4×6 to 5×7 prints: 600 DPI minimum, 800 DPI if you have any intent to crop. This gives you enough data to print at 2× original size later, which is what happens when somebody wants a framed enlargement at a memorial service.
• Small formats (wallet size, cabinet card, tintype): 1200 DPI. You want to pull the face out of a 2×3 inch cabinet card for a portrait crop.
• Negatives and slides: 2400–3200 DPI optical (not interpolated). The film has the resolution; the scanner just has to capture it. A decent flatbed with transparency adapter works; a dedicated film scanner is better if you have hundreds.
• Archival master files: uncompressed TIFF, 48-bit color, even for black-and-white. The best resolution for scanning old photos has the per-format decision tree with printing-size math.

Two practical notes. First, scan both sides of any mounted print or anything with visible handwriting on the reverse. The handwriting is genealogy data, not ephemera. Second, keep the scanner glass cleaner than you think you need to — at 1200 DPI, a single piece of lint is a two-pixel black line across your ancestor's face, and you will not see it on the preview.

The Restoration Triage: What Actually Needs Fixing

A 500-image collection does not need 500 restorations. It needs 500 scans, 150 light-touch auto restorations, and 20 careful hand-worked restorations. Learning to tell which bucket each photo belongs in saves you a hundred hours.

Green pile — auto-restore and move on:

• Light fading, overall contrast loss, slight color cast.
• Minor dust and surface dirt.
• Moderate graininess from fast film.

Yellow pile — needs more attention:

• Significant color fading, especially the pink-shift on 1970s prints.
• Creases and minor tears not crossing faces.
• Localized staining, tidemarks, or foxing.
• Noticeable color shift such as magenta or orange cast on color prints from the dye-instability era.

Red pile — hand work or professional conservation:

• Tears across faces.
• Missing emulsion patches.
• Water damage with emulsion bonding — the full triage is in the ultimate guide to water damage photo restoration and the at-home version in restore water damaged photographs at home.
• Heavy silvering on the mirror-like edges of old black-and-white prints.
• Photos of extreme personal or historical importance that justify the extra time.

I run the green pile through old photo restoration in batch mode, review output thumbnails at 100%, flag anything the auto pass got wrong, and move on. The yellow pile gets individual review — usually auto plus a 2–3 minute tweak per image. The red pile I reserve for a dedicated session, because the cognitive overhead of switching between batch mode and detail work is real.

The AI Restoration Pass: What to Expect on Family Photos

Modern AI restoration is genuinely good at the failure modes that dominate family archives. It is also overconfident on a specific set of failures that genealogists care about, and you need to know where to second-guess it.

What AI Handles Well in Genealogy Workflows

• Overall tonal recovery from faded prints. A flat gray 1960s snapshot comes back with plausible midtones and shadow detail.
• Face sharpening on group photos. The face-aware pipeline understands where eyes and mouths belong, so it restores portrait structure reliably. My how to enhance group photo faces guide has the specific workflow for the common "twelve relatives in a front-porch row" shot.
• Color-cast correction on 1970s–1990s C-prints. The pink-shift problem has a consistent chemistry, and the models were trained on it extensively.
• Dust and scratch cleanup on modest-damage prints where the emulsion is intact.
• Reading the explainer on how AI photo restoration works will save you frustration — understanding what the model does lets you predict when it will succeed.

What AI Gets Wrong on Genealogy Photos Specifically

• It does not know who your ancestors are. Face restoration on a blurry 1920s wedding portrait produces a face that looks plausible for the era, not necessarily the actual person's features. If there is a sharper reference photo of the same person, scan both and compare.
• It does not preserve period-correct tone. A warm sepia albumen print is not a faded black-and-white; the sepia is the intended tone. Left to defaults, AI will neutralize it. Turn off aggressive color correction for pre-1910 material.
• It invents edges and hands. If a photo has a partially cropped hand on a relative's shoulder, the model may extend it plausibly — but wrongly. For genealogy archival, be conservative about accepting invented content. What looks pretty today looks like falsification to your great-grandchildren.
• It struggles with collodion wet-plate artifacts (1850s–1880s). The irregular coating pattern gets read as damage and smoothed out. Scan and leave alone; the artifacts are part of the artifact.

For a head-to-head of how different tools handle family work, my best AI old photo restoration tools 2026 roundup and the deeper photo restoration software comparison break out the tradeoffs by use case. If you are cross-shopping a specific competitor, the ArtImageHub vs Adobe Photoshop, ArtImageHub vs Hotpot AI, and ArtImageHub vs ImageLarger writeups compare workflows directly. The honest answer on whether a general-purpose chat model will do this job is in can Gemini restore old photos.

A Worked Example: Restoring a Great-Grandparent's Wedding Photo

Let me walk through a real example from last year, anonymized. A client brought me a 1917 studio wedding portrait — cabinet card mount, albumen-on-gelatin print, moderate foxing, one visible crease running through the groom's collar, a sepia tone shifted slightly brown, and — this is the part that mattered most — handwritten names on the reverse that had faded to the point of being barely legible.

My workflow:

1. Scan the front at 1200 DPI, 48-bit color, TIFF. Scan the reverse at 800 DPI with strong side-lighting to bring up the handwritten indentations. The side-lit scan of the reverse made the names readable where a flat scan had not.
2. Run the front through old photo restoration on default. Review at 100%.
3. The auto pass handled the foxing and the overall fade. It over-corrected the sepia toward a cooler neutral, which I did not want. I kept the fade correction but manually warmed the result back toward the original albumen tone.
4. The crease through the collar was the trickiest bit. The auto pass had filled it plausibly but introduced a faint vertical artifact. I opened the output in Photoshop and did a 4-minute clone-stamp pass with the healing brush on the residual line.
5. Saved three versions. Archival TIFF with full metadata embedded (date, people, provenance). Web-quality JPEG for sharing. A 5×7 reprint file at the original sepia tone for the family's display. See my how to print restored photos guide for the print-fidelity specifics.

Total elapsed time: 22 minutes. The genealogical payoff — legible names, sharp faces, distributable copies — came from the full workflow, not just the AI step. This is representative of the middle of the yellow pile; green pile averages about 4 minutes, red pile can run an hour or more. For wedding work specifically I have a fuller writeup in wedding photo restoration.

Tool Comparison for Genealogy Collections

Genealogy has different tool requirements than single-image restoration. Batch throughput matters, metadata handling matters, and the ability to round-trip through a spreadsheet matters. Here is how the options sort out in my working experience.

| Tool | Strengths for Family History | Limitations | Best For | |------|------------------------------|-------------|----------| | ArtImageHub (this site) | Fast batch on typical yellow-pile damage; good tone preservation on pre-1920 sepia material; straightforward export naming | Still iterating on very severe emulsion loss | Most household genealogy collections | | MyHeritage Photo Tools | Integrated into an ancestry platform; face-aware on portraits | Locks into their ecosystem; weaker on non-face damage; subscription gates features | Users already in MyHeritage's family tree workflow | | Photoshop (manual) | Unlimited control, ideal for red-pile hand work | 30–90 minutes per photo; steep skill requirement; not realistic for 500-image batches | 5–10% of the collection that AI cannot handle | | Adobe Photo Restoration Neural Filters | Good for single photos; integrates with the Photoshop workflow | Still requires Photoshop licensing; per-image, not batch | Existing Photoshop users, one photo at a time | | Remini / FaceApp-style mobile | Fast on phone; useful for quick shares | Over-smooths skin; not archival quality; watermarks or subscription gates on full-res | Family group chats, not masters | | Scan-only services (ScanMyPhotos, Legacybox) | Handle the volume-digitization pain | Do not restore; do not preserve reverse-side handwriting by default | Families with thousands of prints and no time to scan |

For what it is worth, most of my genealogy clients end up on a two-tool stack: ArtImageHub for the 90% automated pass, and Photoshop for the 10% hand work. The right combination is the one that lets you finish — the tool that gets 800 restorations done is better than the tool that would get 40 done perfectly.

Before You Call a Professional

A professional conservator is the right answer for a small subset of genealogy work. Hire one for:

• Cased images (daguerreotypes, ambrotypes) with visible degradation of the silver or collodion.
• Prints bonded to glass or to album pages where separation requires solvent work.
• Any photo identified as having auction-level monetary value (civil war portraits, early celebrity studio work, identified historical figures).
• Family bibles or scrapbooks where the substrate itself is deteriorating and the photos are incidental to a larger conservation problem.

Expected cost: $150 for a consultation, $200–600 per image for moderate work, $500–2,000+ for archival treatment on the most valuable material. The American Institute for Conservation maintains a regional directory, and most major university libraries can refer local conservators.

For the other 90% of a family archive — snapshots, portraits, event photos — DIY scanning plus AI restoration is appropriate, and arguably better than professional treatment because the alternative (waiting a decade to afford a conservator for 500 photos) is how collections get lost.

The Distribution Problem: Getting the Archive Into Family Hands

A restored archive that sits on one hard drive is not finished. The entire point of genealogy work is that the archive outlives you. Budget time for distribution.

My default distribution bundle:

• A cloud folder (Google Drive, Dropbox, or FamilySearch Memories) with the restored JPEGs and a CSV of the catalog.
• A physical USB drive mailed to two or three committed family members — not because USB is archival, but because it is tangible and passes down.
• A small printed book of the 30–50 best shots, captioned. Print-on-demand services handle this for $40–80 and the book is frequently the most meaningful deliverable of the whole project.
• A brief written guide explaining how the archive is organized, so the next person who inherits it can continue it. This document is the difference between a living archive and a box of files nobody dares touch.

For the print-quality specifics — paper choice, color management, sizing — the how to print restored photos guide covers what works.

Special Cases: Less Common But Worth Knowing

Some family archives have particular quirks that do not fit the standard workflow.

Military service photos

Often the most treasured and the most technically challenging. Dog-eared wallet carries, cracked emulsion from pocket storage, unit patches blurred by wear. Scan at 1200 DPI, accept that cropping in tight on the face is often the best presentation, and cross-reference unit insignia against military history databases for dating. The vintage photo repair techniques guide handles the condition issues.

Immigrant-era studio portraits (1890s–1920s)

Often commissioned right after arrival, these represent an enormous percentage of surviving US family collections. They scan cleanly — studio conditions were controlled, lighting was good, and the prints were made by professionals on quality stock. The photographer's mark is usually readable, which gives you both a date window and often a neighborhood.

Color snapshots from 1972–1985

The problem era. Early color processing produced prints that fade and shift to pink within 30 years. AI handles this particular failure mode well because it is so consistent. Our fix color shift in old photos guide covers the chemistry and the correction approach.

iPhone-era digital (2008+)

Not all "old" photos are on paper. Digital photos from 2008–2015 often survive only as low-resolution Facebook downloads, and the originals are gone from phones that were traded in. These respond well to AI upscaling rather than traditional restoration. See restore old photos iPhone for the mobile-originating workflow.

Photos inherited on 35mm slides

Underrated. A box of 500 Kodachrome slides from the 1960s can be the single best-preserved source in the entire collection — Kodachrome is famously stable, and slides were stored in the dark. Do not treat these as an afterthought; they can be the crown jewels.

Frequently Asked Questions

How long does a full genealogy photo restoration project take?

For a typical 500-image family collection, budget 40–80 hours end to end. That breaks out roughly as 15 hours cataloging and organizing the physical box, 20 hours scanning, 10 hours running the AI restoration pass and reviewing output, and 5–20 hours on the red-pile hand work depending on how many photos are in bad shape. Most people complete a project in 3–6 months of weekend work. Trying to do it in a single push burns out; setting a pace of 20–30 photos per weekend is more sustainable.

How much does genealogy photo restoration cost if I do it myself?

If you already own a flatbed scanner, the tooling cost is $0–10 per month for an AI restoration subscription. Without a scanner, budget $200–400 for a decent entry-level model (Epson V600 is the common choice) or $15–30 per session for a shared scanner at a FamilySearch Center or public library. For a 500-image project, total out-of-pocket cost is usually under $100 plus your time. Professional scanning and restoration services for the same volume would run $3,000–10,000.

Should I restore the originals or keep them as-is?

Keep the originals. Never alter a physical heirloom. All restoration work should happen on the scanned digital copy, with the original stored in acid-free archival sleeves at stable room temperature. The original carries physical evidence that the scan cannot — paper stock, photographer's imprint, handwritten annotations, embossed mounts — and a generation from now somebody will want to examine it for a detail you did not anticipate.

What do I do with the originals after I have scanned everything?

Distribute them deliberately. A set to the family member most likely to care (and most likely to survive you). A set to the youngest cousin interested in genealogy. Duplicate prints and negatives can go to a local historical society if they document a regional community. Do not throw out originals you have scanned — the scan is a reproduction, not a replacement. Store what you keep in acid-free boxes, out of basements and attics, at stable humidity.

How do I identify unknown people in old family photos?

Cross-reference. Compare faces in the unidentified photo against known portraits of family members at similar ages. Look at clothing, hairstyles, and backgrounds for date clues. Check the reverse for handwriting — even partially legible names help. Post the image to family social media with an explicit "who is this?" request; older relatives often know on sight what the archive has lost. Genealogy forums (especially RootsWeb and the FamilySearch community) will often help identify uniform insignia, studio photographer marks, and regional clothing styles.

Is it worth colorizing old family photos?

It is a choice, not a necessity. Colorization is a presentation layer, not a restoration — the original was monochrome, and adding color is interpretation. Done well and tastefully (faces in natural tones, clothing in plausible colors for the era), a colorized version is engaging for younger family members who struggle to connect with black-and-white. Done poorly (oversaturated, anachronistic colors), it cheapens the original. My complete guide to colorize old photos covers the process and the editorial decisions involved.

Can AI restoration introduce errors into genealogical records?

Yes, and this is worth taking seriously. AI face restoration on a blurry photo produces a plausible face, not necessarily the actual face. AI inpainting on a torn edge invents content. For genealogical purposes — where the archive is a historical record — I keep both the unrestored scan and the restored version in the archive, clearly labeled. The unrestored scan is the primary source; the restoration is a presentation. This matters most for photos that are the only surviving image of a person; for one of twenty shots at a reunion, the distinction is academic.

What scanning resolution do professional archivists use?

The Library of Congress standard is 600 DPI for standard reflective prints, 1200 DPI for small formats and anything with fine detail worth preserving, and 2400+ DPI optical for negatives and slides. These are minimums, not recommendations — archival projects today commonly scan at 800 or 1200 DPI for paper prints to allow for future display at higher pixel densities than current monitors. 48-bit color depth is standard for archival work even on black-and-white material.

---

A genealogy archive is one of the few projects in your life where the payoff compounds for generations. Start with the worst-condition photos first, because their clock is running fastest. When you are ready to digitize and restore, bring your scans to old photo restoration and let the batch process handle the green pile while you focus your attention on the photos that actually need hand work. The family collections that survive are the ones somebody decided to save before it was too late.