Date stamps, pen notes and show-through on old photo scans

It usually starts on a quiet weekend. You pull out the shoebox of prints from the drugstore photo counter — birthdays, beach trips, holidays from three decades — and start scanning them before the next family reunion. Then the scans come up on a big screen and you notice what you never saw on paper: an orange date glowing in the corner of half the prints and someone's ballpoint notes across a few. On others there's a faint, backwards ghost of writing from the back of the print.

Two questions follow. Can any of that come off? And can it come off without hurting pictures that are already fragile? Both have answers, and the answers are different for each kind of writing.

The short answer: what comes off cleanly and what doesn't

• A date stamp on a plain patch of sky, wall or floor is usually the easy case. What sits underneath is simple to rebuild.
• Writing across someone's face can come off, but whatever fills the gap is the AI's guess at what was there, so check it closely.
• Writing showing through from the back isn't on the picture at all. You fix it at the scanner, not in an editor.
• A date printed in the white border sits outside the picture. Crop the border when you scan and no AI is needed.

The rest of this article explains why. It also covers what to do first when a print is both stamped and damaged — and why that date is worth saving before it disappears.

Four kinds of writing on old prints

One thing about the tool first, because it explains everything below. When Remove text finds the text for you, you don't outline anything: it reads the whole photo for anything that looks like writing, turns every piece it finds into one area and fills that area in from what surrounds it. So the hard part is rarely the writing itself. It's what the writing is sitting on.

The orange date the camera printed

Date printing was a feature of everyday family cameras. Canon's camera museum sums up its Datematic, marketed in November 1974, in one line: "A camera for the masses, it featured date imprinting and a body and exterior made of reinforced plastic." The same page mentions "a convenient date on/off switch" and adds: "The date could also be imprinted on the lower right corner of the photograph." Later compacts moved the feature to the camera back; the museum lists the date-imprinting version of one of them, sold in the Americas, as the (New) Sure Shot QD.

However the digits got there, they were exposed onto the film when the picture was taken, the same way as everything else in the frame. They are in the picture, not on it. The camera's switch only mattered before the shutter fired. On the print in your hand there's no layer to turn off and no clean original underneath, and the orange date is as real as the sky behind it. The good news is where it lands: the same corner of every frame, which often holds pavement, a floor, a wall or a patch of sky — the easiest things to rebuild.

A date printed in the white border

Some prints carry a date in the white margin instead. The Image Permanence Institute at RIT, which documents how photographs were made, explains: "In 1960 it became possible to optically print the processing date in the border of the print." It adds that in the 1970s most prints were made without borders. Two things follow. That date is the day the film was processed, not the day the picture was taken. And because it sits in the margin, outside the picture, the clean fix is to crop the border when you scan.

Pen notes on the front

"Aunt Carol, 1968," a name under each face, a note in marker across the sky. You can see the stroke, the ink color and the slant of someone's hand. These are the unpredictable ones. In a corner of plain sky they come off easily; across a sleeve or a face, the area underneath is detail that nothing nearby can suggest, so the fill is a guess.

Writing or a paper brand showing through from the back

The last kind was never meant to be seen. The same institute notes: "Beginning in 1958 Kodak begin to include a backprint on their products." Later on, "beginning in the late 1970s some photofinisher companies could print text, including processing date and exposure information, on the back of prints." Add whatever someone wrote on the back by hand, and a scanner can pick all of it up through the paper: faint, reversed and visible mostly in the lighter parts of the picture.

Kind of writing	How to recognize it	What usually sits under it	How hard it is to remove
Camera date stamp	Orange or yellow digits in the same corner of every frame	Sky, a wall, pavement or a floor near the corner	Usually the easiest: the area under it tends to be one even patch
Pen notes on the front	Visible strokes, ink color, a handwriting slant	The picture itself, sometimes a face	Moderate to hard, depending on what the ink crosses
Show-through from the back	Faint, reversed, scattered across the light areas	Anything	Hard in an editor, because it's a scanning problem rather than writing on the picture
Date in the white border	Evenly printed, outside the image area	The white border	Easy: crop the border when you scan

The last column is words, not scores, on purpose: the result depends on the print. One boundary is worth drawing here and then leaving alone. A logo, a semi-transparent watermark or a copyright mark is a different job for Remove watermark, and which remover fits a logo, a line of text or a stray object sorts out the mixed cases.

Fix show-through at the scanner, not in an editor

Show-through deserves its own answer, even though the answer means you won't need a tool for it. The writing isn't on your picture; the scanner is reading it through the paper. Epson's troubleshooting page has an entry titled "Back of Original Image Appears in Scanned Image" and gives the fix in one line: "Place a piece of black paper on the back of the original and scan it again." The same page suggests adjusting the brightness setting in its scanning software when that's available. Rescanning the batch this way is faster than chasing faint marks across every print one at a time.

While the scanner is open, set the resolution once for the whole box. The Library of Congress guide to scanning personal collections recommends 300 dpi in general for 4×6 and 5×7 photos, and 400 or 600 dpi if you plan to enlarge them to 8×10 or bigger. So 300 is the starting point for a snapshot you'll keep at its own size, not a ceiling. If you're making an archival master of a print you care about, the National Archives standard is covered in our guide to scanning a print before you restore it.

What Remove text does, and what to check afterward

When Remove text finds the text for you, it works on the whole photo at once: it reads the picture for anything it recognizes as text, combines every piece it finds into a single area and fills that area in from the surroundings. Three things follow from that.

It doesn't pick and choose. Everything it reads as text gets the same treatment. If the photo also shows a store sign, a slogan on a T-shirt or "Happy Birthday" piped on a cake, those go along with the date. If one of them matters to you, keep an untouched copy of the scan.

It can only remove what it finds. Faint pencil, a very stylized date or a note that blends into the picture may not register as text, and then it stays. If nothing in the photo registers as text, the tool tells you no text was found and leaves the photo alone. Credits are used only when text is actually removed, and the editor shows the cost before you run it.

Film digits have soft edges. A date exposed onto film doesn't have the crisp outline of digital text stamped onto a phone photo. It fades into a soft glow, and that glow is fainter than the digits themselves, so it's the part most likely to be left behind. After the fill, zoom in on exactly where the digits were. On a plain background the patch should be hard to find; over busy detail it may leave a faint trace.

One more expectation to set: removing text isn't sharpening. The tool fills an area back in; it isn't built to sharpen the rest of the photo, so a soft scan stays soft. If blur is the real problem with a photo rather than the date in the corner, work out which kind of blur it is first, since each kind goes to a different tool.

Damaged and stamped: remove the text first or restore first?

Most old prints aren't only stamped. They're scratched, creased, stained or faded as well, which raises a question that's easy to get backwards: which job goes first? What follows is reasoning from how the two tools work, not the result of a test.

Two facts carry the argument. Restore old photos runs three fixes over the whole picture in a single pass — sharpening with a resolution boost, rebalancing faded color and removing scratches — and none of the three is text removal. Remove text, for its part, fills each area it clears by borrowing from what surrounds it. Put the two together and the order mostly decides itself.

Condition of the print	Do first	Why
Stamped, otherwise undamaged	Remove the text, and you're done	There's nothing to restore
Stamped on an undamaged area, damaged elsewhere	Remove the text first	The restoration pass then covers the whole print, the filled patch included
The text sits right on a crease, stain or scratch	Restore first	The fill borrows from its surroundings, so damaged surroundings get copied into the patch
Faint reversed writing all over	Go back to the scanner	It's show-through, and no editing tool is the right fix

The third row is the one people get wrong, because the instinct is to clear the writing first. Picture a date stamp that landed on a corner with a crease and a brown stain. Remove it now and what fills the gap is copied from the crease and the stain right next to it. Run Restore old photos on that print first and remove the date from the repaired copy afterward: now the fill has a clean background to borrow from. For prints that are torn, missing pieces or water-damaged, what a badly damaged print can come back from goes through that triage.

For the first two rows, removing the text first has one more advantage. A freshly filled patch is clean and smooth, which can make it stand out on a print that has yellowed and softened over the years. Because the restoration pass works on the entire picture, running it afterward should bring the patch and the rest of the print closer together. That, too, is reasoning rather than a measured result.

Before you erase it: the date is data

Before you clear a whole box, slow down for four cases.

The orange date is a record, not just a blemish. It shows the date the camera was set to when the shutter fired — usually the day the picture was taken, if someone set the clock. With family photos it's often the only record left: the people in the picture may not be around to ask, and nobody kept a log. Keep one untouched scan, and copy the date into the file name before you remove it. The Library of Congress suggests descriptive file names built on a "yyyymmdd" date structure, so a picnic photo might become 19840704-picnic.jpg and keep its date long after the digits are gone from the picture.

A date in the border is a different date. As the Image Permanence Institute describes it, that's the processing date, the day the lab developed and printed the film. It belongs in a caption, but don't label it as the day of the photo.

Handwriting can be the keepsake. A grandmother's note on the back, a father's signature in a corner, a few hurried words from someone who's gone. To a scanner it's text to clean up; to a family it can matter more than the picture. Scan the back as its own file before you touch the front.

Leave record photos alone. A photo kept as a record, as evidence or for any official purpose is worth more untouched than tidied up, even when the only change would be a line of text in a margin.

Family albums are the photos people least like sending anywhere, so here's how uploads are handled. Your scans travel to us over an encrypted connection. The outside AI service that processes each one erases it within 24 hours, and the copies held by Vipix are removed on the timetable our Privacy Policy sets out. Vipix does not use your photos to train AI. One last check before a cleaned-up scan joins the family archive: open the downloaded file, because some downloads come with a small vipix.vn mark, and whether yours does depends on the tool and on whether you were signed in when you ran it.

If you've just scanned a box and every print has a date in the corner, try one print first. Run Remove text on a single scan and check that the patch matches what's around it. One test tells you how the rest of the box will go, and whether that date is worth removing or worth keeping.

Frequently asked questions

Can you remove the date stamp from a printed photo?

Not from the paper itself. Scan the print, then remove the date from the digital copy. The print stays as it is, which is also a good reason to keep it.

Why does writing from the back show up in my scan?

The scanner picks it up through the paper, whether it's a note someone wrote on the back or text a photo lab printed there. Epson's fix is to place a piece of black paper on the back of the print and scan it again.

Is the date in the white border the day the photo was taken?

No. The Image Permanence Institute describes it as the processing date, the day the film was developed and printed, not the day the picture was taken.

What resolution should I scan old snapshots at?

The Library of Congress recommends 300 dpi for 4×6 and 5×7 photos in general, and 400 or 600 dpi if you plan to enlarge them to 8×10 or bigger. For an archival master, the National Archives standard is covered in our guide to AI photo restoration.

Can the same tool remove a watermark or a logo?

That's a different job. Logos, semi-transparent watermarks and copyright marks belong to Remove watermark, for photos you hold the rights to. Remove text is built for writing.