OldPhoto vs Photoshop: What Each One Actually Does to a Damaged Print

What a Photoshop restoration actually involves

Restoring a damaged print in Photoshop usually runs through four stages, and none of them are automatic.

• Colour first. Silver mirroring and magenta casts are corrected per channel before anything is sharpened. Doing it in this order is why yellowing correction is listed as an early step rather than a finishing touch.
• Dust and scratches next. The clone stamp and the healing brush borrow texture from elsewhere in the frame, one spot at a time.
• Filling voids. Torn areas and lifted emulsion have to be rebuilt structurally, and this is where the hours go.
• Sharpening and output last. How far sharpening goes depends on the destination. Sharpen for screen and then print, and you get halos.

That is why professional work is quoted by the hour rather than by megabyte. A slightly dusty scan can be finished in twenty minutes. Rebuilding a face where the emulsion has lifted takes hours of the same careful borrowing, and it is priced accordingly. The cost side sits in whether professional work is worth it and free versus paid.

One premise gets overlooked. Photoshop is a subscription, and the learning curve is steep. Learning the channel mixer and the clone stamp for two hundred prints in a shoebox is not a good trade, whatever the tool can do in expert hands.

What OldPhoto does, and what it does not

A restoration model has seen a great many damaged prints paired with undamaged ones, and it infers a plausible clean image. It is strong on statistical problems: dust, light scratches, mild fade, flat contrast, slight softness. Those are the failures that show up most often, and statistics solve them well.

It is weak where only one answer is right. Half a missing face gets filled with something statistically average. A uniform gets a plausible colour that may belong to the wrong regiment. That is the same failure mode written up for colour in how accurate colourisation is, and it is worth being honest about before you start.

The second practical limit is repeat passes. Enhancement run twice rarely improves anything, because skin goes waxy and fine detail turns plastic. One clean pass from a good scan beats three stacked attempts, which is why a single upscale is usually the last step rather than an early one.

Where each one breaks

• Surface scratches and dust: OldPhoto wins outright, seconds against an hour of spot healing.
• Fading and flat contrast: OldPhoto wins. Pulling curves is mechanical work, and paying an hourly rate for it is a poor trade.
• Water damage and emulsion loss: Photoshop wins. Filling a void needs a judgement about what belonged there, which is exactly what inference gets wrong. That damage is covered in water damaged prints.
• A badly damaged face: Photoshop wins, and not narrowly. This is the case that sends people to a professional in the first place.
• Two hundred prints from one album: OldPhoto wins. Nobody hand restores a shoebox, and nobody should pay for it. Triage thinking is in album digitisation.
• Needing to explain every change later: Photoshop wins. Layers and history are the evidence, and a model pass is not.

Time and money

Numbers here vary by region and by whoever answers the phone, so read the ranges as shapes rather than quotes. Photoshop means a subscription plus your own hours, and the hours are the expensive part. An automated pass costs cents in credits and takes seconds. What sits between the two is dealt with in free versus paid, and what people actually end up paying for in whether professional work is worth it.

• Loads of prints, all lightly damaged: automated, no contest.
• One print where the face carries the history: hand work, and the hours are easy to justify.
• Both: the common arrangement. Batch everything cheap, then pay a person for the two or three you intend to frame.

The order that gets the best out of both

1. Scan first, and scan properly. Nothing downstream recovers detail the capture never had. Clean the glass, and scan at higher resolution than you think you need.
2. Run the automated pass on a copy. Never the original file. This clears dust, flatness and light scratches in under a minute.
3. Look at it at 100 percent before doing anything else. Eyes, mouth line and hairline are where invented detail shows up first.
4. Send that file to a person if it still needs work. They start from something cleaned rather than something filthy, which is cheaper for both sides.
5. Keep the versions separate. Original scan, automated pass, hand finished file, clearly named. A scheme that survives contact with a real family is in archive organisation.

How to decide about one specific print

Ask what the image has to do. A photograph going into a family history with dates and names is a record, and invented detail in it becomes somebody else's problem later. A photograph being reprinted for a birthday is a keepsake, and plausible is plenty. The same split, from the tooling side, is in app versus desktop workflows.

FAQ

• Is Photoshop the standard answer for old photo restoration? It is the industry standard, but standard refers to what professional retouchers use, not to what an ordinary person should learn. For everyday dust and fading, the time spent learning it costs more than the result is worth.
• Without Photoshop skills, how far can I get? Most common damage is within reach of automated tools now. What genuinely needs hand work is structural loss such as half a face missing or large areas of lifted emulsion, and that still ends up with a person. This year's comparison is in OldPhoto vs Remini 2026.
• OldPhoto first, or Photoshop first? OldPhoto first. A retoucher starting from a cleaned file costs less than one starting from a dirty file, and reversing it means paying an hourly rate to redo mechanical work a model already finished.
• Should the finished restoration be labelled? Yes, in both directions. Colourised results that get published should be labelled as an interpretation, and anything going into a family history should keep the original scan as the record. The wider trade-off is in AI versus traditional restoration.

← All articles