Restoring a boxful, and deciding what not to restore
Almost everything written about this describes one photograph. Scan it, repair it, admire it. Nobody arrives with one photograph. They arrive with a box, or a drawer, or four albums with the glue gone brittle, and the decisions that determine how that turns out all happen before any repair does.
The work splits into two jobs that get confused with each other constantly. Copying is fast, mechanical and applies to everything. Restoring is slow, careful and applies to almost nothing. Treating them as one job is why boxes get opened, stalled and closed again.
Copy everything before you repair anything
Do one pass over the whole box and copy every print in it before improving a single one. Not the good ones, not the interesting ones. All of them, in whatever order they come out.
The reason is that the prints are the fragile part and the files are not. Every day the box stays a box is another day of the same slow fading that got you here, and the copying pass is the only step that stops the clock. A mediocre copy of everything is worth far more than a beautiful copy of the few photographs you got through before you lost momentum.
- Keep the settings the same for the whole pass. Changing them mid box turns one collection into several that no longer match.
- Copy the back of any print with writing on it. Names, places and occasions are the part that cannot be reconstructed later by anyone.
- Do not stop to fix anything. Every repair during the copying pass costs you the pass.
Sort by what a photo is for, not by how damaged it is
Damage is the obvious sorting criterion and the wrong one. A badly faded photograph of a wedding is worth more attention than a pristine one of a hedge, and the box is full of both.
Sort instead by what each print is doing. Some are for looking at. Some carry information rather than sentiment: a house, a uniform, a street sign, handwriting on the back. Some are of people nobody living can name, which is a category with its own deadline attached.
- For looking at: the small number that will be printed, framed or sent to somebody. These are what restoration is for.
- For the record: identifiable but not decorative. Copy them well, label them, and leave them alone.
- Unidentified: worth showing to the oldest person who might recognise a face, and worth doing before anything else on this list.
Restore the smallest number of prints you can
Restoration does not scale. It is per print, it needs judgement on each one, and the effort is unrelated to how much you care about the result. A boxful does not need restoring. A handful does, and the rest need to be findable.
Choosing that handful is the actual skill here. The test is not how damaged a print is, it is whether anybody will ever look at it closely enough for the damage to matter. Most photographs in most boxes will be looked at once, in a scroll, at the size of a thumbnail.
The ones not worth restoring, and the ones you should not touch
Some prints are better left as copies, and saying so is not defeatism. Duplicates, near identical frames from the same afternoon, and photographs whose subject nobody can identify all cost the same effort as the good ones and return almost nothing.
Then there is damage that software genuinely does not fix. A tear through a face is reconstructed by inference rather than recovered, so the result is a plausible face rather than that face. Heavy silvering across the subject is a change to the object rather than to the image. Clara does not pretend otherwise on either, and the damage guide covers which kinds of damage fall where.
A separate category is prints you should not handle at all yet. Anything damp, stuck to glass, or stuck to another print is a conservation problem before it is a scanning one, and separating it yourself is how the emulsion comes away. Copy it exactly as it lies and deal with it later.
Naming is what makes the copy worth having
A folder of files named by whatever the camera called them is a second shoebox with worse lighting. It is searchable by nothing, sortable by nothing useful, and it will be inherited by somebody who has even less idea who these people are than you do.
The pass worth doing is naming, and it has the same deadline as the unidentified pile: it depends on somebody who still recognises faces. Year, place, and who is in it, even approximately. Approximate is not a failure here. An album marked as probably a christening in the sixties is infinitely more useful than an unlabelled file, and nobody can add that information later.
More guides
Questions
Where do I start with a whole box of old photos?
Copy all of them first, in one pass, without stopping to fix anything. The prints are what is deteriorating and the files are not, so getting everything copied is the step that stops the loss. Sorting and restoring are both easier afterwards and neither is urgent.
How many photos should I actually restore?
Far fewer than you expect. Restoration is worth the effort on prints somebody will look at closely, which is usually a small number per box. Everything else needs to be copied and labelled rather than repaired.
Can every old photo be restored?
No, and it is worth knowing which before you spend time. Damage around the subject is recoverable, damage through a face is reconstruction rather than recovery, and a print that is damp or stuck to something needs a conservator before it needs any software.