Latest
Billions for AI, Pennies for the Plumbing: AI Companies Should Pay to Keep PHP AlivePHP2Shell: AI Found a 22-Year-Old Crack Under Your Website in Two DaysPetrov Day: The Officer Who Didn’t Believe the ComputerKindle Unlimited Is Now Kindle+. AI Audiobooks Just Joined the BuffetMo Gawdat Says the AI Dystopia Starts in 2027. Read the List, Skip the DatesThe Society of Authors Signed an Ethical AI Charter, Then Ran From It in Eight DaysChina Is Using Bots to Fan the Data Center Backlash. The Backlash Is Still Real.When to Quit Writing a Book: I’ve Abandoned Dozens, and I’d Do It AgainGetty Images Filed for Bankruptcy. Can You Prove You Licensed Your Book’s Images?Fluff Words: Cut Them From Your Prose, Keep Them in Your DialogueWhere Can Writers Learn Storytelling From Movies? Richard Lowe Publishes 1,226 Film and TV Reviews With Writing Craft NotesWhere Are the Independent Bookstores and Public Libraries Worth Visiting in Tampa Bay? Author Richard Lowe Profiles 44 of ThemWhat Should You Know Before Hiring a Ghostwriter? The Writing King Publishes 57 New Answers, Including What to Do When It Goes WrongWhere Can You Hear Founders and Executives Tell Their Own Stories? Richard Lowe’s Leaders and Their Stories Publishes Every Unscripted Interview With a Full TranscriptAuthor Richard Lowe Publishes Eight Volumes of Enemies of You, a Researched Series on the Forces Working Against Ordinary AmericansThe Jackal (1997) ReviewThe Day of the Jackal (2024) ReviewYour First Novel Is an Apprentice Piece: What Beta Readers and Editors Can’t FixWhen the Cloud Gets Bombed: What the AWS Data Center Strikes Teach About BackupsHuman-Written Book Labels Are Gatekeeping With a FeeThe Enshittification of Microsoft Office: Why I Switched to a Free CloneGlobal Encryption Day: The Padlock Every Writer Depends OnWriting Your Worst Moments in a Memoir: Show the Evidence, Skip the VerdictHow to Register a Copyright for Your Book (and What Doesn’t Count)Your AI Chats Aren’t Private: Deleted ChatGPT Logs in CourtThe Board Seat Scam: A 419 Letter in a Business SuitI Joined the Science Fiction Book Club 30 TimesCitations in a Memoir: Endnotes, Footnotes and Fair UseThe Stores That Still Care: Publix, Walmart, Aldi, Best Buy and the RestThe Grocery Store Plan: Why Every Store Starts With FlowersWinnie-the-Pooh’s 100th Birthday: A Century in the Hundred Acre WoodFree With a Fill-Up: How America Stopped Giving Things AwayAI’s Black Box Is Getting a Window. Somebody Still Has to Look Through It“It Stars C. Thomas Howell. No.” Why AI Gets Stuck in a LoopShould You Query or Self-Publish Your Memoir If You’re Not Famous?Judge the Writing, Not the Tool: What the AI Pile-Ons Get WrongThiago Ferreira: Thinkers, Not DoersYana Semova: Let AI Do the AdminLisa Hamilton-Smith: Unstoppable, AI AssistedRobert E. Howard Day: The Texan Who Built ConanElmore Leonard Day: Leave Out the Part Readers SkipRoald Dahl Day: Six Pencils and a Grudge Against Grown-UpsEncourage a Young Writer Day: The Adult Who Read the PagesStephen King Day: The Daily Two ThousandZora Neale Hurston Day: The Ear She TrainedTell a Fairy Tale Day: The Stories Everybody EditedAlice’s Day: The Story Told on a Rowing BoatSuperman Day: The $130 Contract Behind the CapeNational Podcast Day: The Interview Is the Raw MaterialRaymond Chandler Day: The Novelist Who Started at Fifty
The Writing King Your Ethical Ghostwriter. Your Story, Done Right.

AI Photo Restoration for Your Book: Fix the Photo, Keep the Original

TL;DR: Restoring old photos for a memoir or family history book is fine, and AI makes it cheap. Cleanup that removes cracks, dust and stains is safe. Heavier AI enhancement and colorization guess at what’s missing, so they can change faces and invent colors. Keep the original scan of every photo, label what you changed in the caption or photo credits, and check every face with someone who knew the person. Genealogy records and evidence get the untouched original, ideally side by side with the restored version.

A photo in a book has one job. It lets the reader see the person you’re writing about.

Take a cracked, faded, brown-stained print. It does that job badly, and the reader squints at a blur where Grandpa’s face should be and moves on. Clean that photo up, and the same reader stops, looks at his eyes and the cut of his suit, and the chapter about him hits harder. Most family photos have no value as art. They’re the closest the reader will ever get to meeting the people in the book That’s the reason I care how they look.

My rule for any photo going into a book is simple. Give the reader the best experience you can, and keep the original.

Should you use AI to restore old photos for a book?

Yes, if you do it with your eyes open. The fuss about AI photo restoration is mostly overblown, and it’s the same argument people made about colorizing old movies. Some of the worry has a point, but nothing about cleaning up a damaged photo of your mother is dishonest.

A reader picking up a family history book wants the experience of seeing these people. They don’t want a forensic artifact. A museum or a genealogy archive has a different job.

Watch: Restore Old Photos With AI For FREE Old Photo Restoration Made Easy, AI Controversy

What I did with about 10,000 family photos

I had every photo I own scanned by a service in India. I boxed them up, mailed them over and got them back as digital files, about 10,000 photos in all. Every scan came back with some basic cleanup done: cracks taken out, dust removed, that kind of thing. On the service’s website there was a checkbox for heavier enhancement, for something like 50 cents to a dollar extra per photo. That removes big cracks and big damage.

Most of my photos didn’t need it. For the ones where the damage was bad, I checked it every time, because it made a much better photograph.

I’ve spent a big part of my life behind a camera, with close to a million photographs of my own, and I’d still rather look at a clean print than a damaged one. So would your readers.

What does AI photo restoration change in an old photo?

It depends on how hard you push the tool, and the difference matters.

Basic cleanup works on what’s already there. Dust specks, scratches, small cracks and stains get filled in from the pixels around them. The face in the photo is still the face the camera recorded. Old-school retouchers did this with a brush for a hundred years.

Heavy enhancement and colorization work differently. When a face is a few blurry pixels, there isn’t enough information in the photo to rebuild it, so the AI makes its best guess. MyHeritage said so openly when it released its Photo Enhancer in 2020: the tool infers what faces looked like. Colorization guesses too. The software has no idea your great-grandmother’s dress was green, so it picks a likely color.

People get burned at this step. In one genealogy thread I read this week, people described AI tools that changed the shape of a face, made a relative unrecognizable, and in one case erased a woman’s jewelry. A military history buff complained that colorized uniforms kept coming out the wrong color. Those complaints are real, so check the work.

Give the reader the best experience you can, and keep the original. – Richard Lowe
Share on X

Keep the original, always

The software has no idea your great-grandmother’s dress was green, so it picks a likely color.

The original scan is the master. Every restored, enhanced or colorized version is a copy you made from it, and copies get replaced when the tools get better. Store the originals in their own folder with their original file names, back them up in two places, and never overwrite one with an edited version. The restored file gets a new name. If someone ever asks what Grandma looked like before the software touched her, you can show them in ten seconds.

I’ve told readers before what to do with a box of unlabelled photographs, and the same advice applies: scan first, label second, edit last.

How do you scan old photos so they print well in a book?

Scan at a resolution high enough to survive printing. A book printer wants images at 300 pixels per inch at the size they’ll appear on the page, so a small wallet photo that will run half a page needs a much higher scan resolution than a large print. Scan ordinary prints at 600 dpi and small or important ones at 1200 dpi, and you’ll have room to crop and enlarge later. Save the masters in a lossless format like TIFF, and make JPEG copies for everyday use.

You can do this yourself with a decent flatbed scanner and a free weekend. Or mail the whole collection to a scanning service. That’s what I did. A service is faster when you have thousands of prints, and the good ones include basic cleanup in the price.

Do the restoration on a copy, at full resolution, and only then size the image for the book. If you restore a small, low-resolution file and blow it up for the page, the AI has even less to work with and guesses even more. The cleanest faces come from the biggest, sharpest scans.

One more practical point. Photos printed on cheap paper, especially in a paperback, lose detail and contrast. A restored photo with clean edges and good contrast holds up on the page far better than a faded original, and readers notice. Clean, properly sized photos make a family history book feel cared for. Muddy brown smudges make it feel like an afterthought.

How should a book label a restored or colorized photo?

Tell the reader what you did. It costs you one line.

A short caption works fine. “Restored from a damaged print” will do, or “Colorized; original was black and white.” If labeling every caption clutters the page, put a note in the photo credits or the acknowledgments explaining that the photos were cleaned up and some were colorized. Mention that the family keeps the original scans. MyHeritage marks its enhanced and colorized photos with a small icon and keeps the original untouched. That’s a sensible standard for anyone.

For anything meant as a record, go further. Genealogy sites, family archives, museum displays and anything that might end up as evidence should show the original next to the restored version, or carry the original alone. A researcher fifty years from now needs to know what the camera saw.

Is a colorized photo a lie?

No more than a retouched portrait was in 1950.

Photographers have been adding color to black and white images almost since photography began, with hand-tinted portraits going back to the 1840s.

For decades, studios colorized portraits as a matter of routine, and some families still have prints with the instructions written on the back: eyes blue, hair brown, sweater pink. A photographer who worked in a darkroom before personal computers pointed out in that same thread that retouching, dodging and burning were everyday parts of making a print. The mole on your great-aunt’s chin may have vanished in 1938, long before any AI touched the photo.

A colorized photo is an interpretation. Labeled as one, it’s honest. Passed off as the original, it isn’t.

When should you leave an old photo alone?

Leave it alone when the photo is the evidence. If a family history book uses a photo to prove something, like who was at a wedding or what a regimental insignia looked like, use the original or put the original right beside the restored version. When the color carries meaning, like a uniform, a flag or a team jersey, and you don’t know the true color, skip the colorization or say outright that the color is a guess.

Skip the restoration when the damage is part of the story, too. A burned corner on the only photo that survived a house fire tells the reader something. Some authors want that, and that’s a fair choice.

And check every face. Show the restored version to someone who knew the person. If your aunt says “that’s not Dad’s nose,” believe her, go back to the original and try a lighter touch.

My take

Restore the photos, and keep every original scan. Light cleanup of cracks and dust is safe. Heavy sharpening and colorization guess, and guesses change faces. Label what you changed in the caption, and show every restored face to someone who knew the person. If your aunt says that’s not Dad’s nose, believe her. Where the photo is evidence, use the original or set it right beside the restored one.

What I tell authors putting photos in a book

Scan everything at high resolution before you do anything else. Clean up dust, scratches and cracks freely, because that work doesn’t invent anything. Use heavy enhancement and colorization where they make the photo better for the reader, check the faces, and label what you did. Keep every original. And when the book is a record as much as a story, show the originals.

I use the same approach when I help clients build a family history book from genealogy, photos and recipes. If you’ve got four hundred photographs and no book yet, the photos are a great place to start.

The publishing and marketing hub covers the rest of book production, and if you want help turning a family’s photos and stories into a memoir, my memoir ghostwriting process explains how I work. Fix the photo so the reader can meet the person. Then put the original somewhere safe.

Frequently Asked Questions

Is it okay to use AI to restore old photos for a memoir?
Yes. Cleanup of cracks, dust and stains makes the photo better for readers without inventing anything. Heavier enhancement and colorization guess at missing detail, so check every face and label what you changed.
Can AI photo enhancement change a person’s face?
It can. When a face is only a few blurry pixels, enhancement tools infer what it looked like, and the result may not match the person. Show the restored photo to someone who knew them before you print it.
Should I keep the original photo after restoring it?
Always. Store the original scan under its original file name, back it up in two places, and save every restored or colorized version as a separate file.
Do readers mind colorized photos in a family history book?
Most readers would rather see a clean, colorized photo than a faded one, in my view. Label it in the caption or the photo credits so nobody mistakes it for the original.
What resolution should I scan old photos at for a book?
Scan ordinary prints at 600 dpi and small or important photos at 1200 dpi. Book printers want images at 300 pixels per inch at the printed size, so small originals need higher scan resolution.
Should AI-restored photos be used on genealogy websites?
Post the original, or the original side by side with the restored version, and label the restored one. Genealogy records serve as evidence, and future researchers need to see what the camera recorded.
When should I not restore or colorize an old photo?
Leave it alone when the photo is evidence of something, when colors carry meaning you can’t verify, like uniforms, or when the damage itself is part of the story.

About the Author
Richard Lowe, professional ghostwriter

Richard Lowe is a professional ghostwriter and author with 113+ books authored and 54+ ghostwritten. Before writing full time he spent 33 years in enterprise technology, including 20 years as Director of Computer Operations and Technical Services at Trader Joe's. He writes nonfiction, fiction and memoir, and works with executives and experts on books that build authority.

More about Richard Lowe →

Disclaimer

The views and opinions expressed in this blog post are solely those of Richard Lowe and are based on personal experience and research. This content is for informational purposes only and should not be construed as professional legal, financial, accounting, or business advice. Always consult with qualified professionals before making important business or legal decisions. Richard Lowe is not a lawyer, accountant, or licensed professional advisor, and this content does not establish any professional relationship.

0 comments

No comments yet. Yours can be the first.

Was this useful?

Leave a comment