Scan a Recipe Card, Get a Typed Recipe

Almost nobody abandons a family cookbook because they stopped caring about it. They abandon it on card eleven, at half past ten at night, having typed out ten recipes and counted the forty still in the tin.

The typing was always the bottleneck. Gathering the cards is the lovely part, and printing the book is somebody else's problem. The bit in the middle - transcribing forty recipes in someone else's handwriting - is where the project quietly dies. This is the tool that removes it.

What it actually does

A photograph of a recipe card is not a recipe. It's a picture. You can't search it, you can't scale it from six servings to twelve, and it can't be laid out on a page next to forty others in a consistent style. Printed straight into a book, it's a scrapbook.

The scanner turns the picture into structured text. It works out which words are the title, which lines are ingredients, which paragraphs are the method, and which numbers are the oven temperature, the timing and the number of servings - and it puts each one in its own field.

That structure is the whole point. It's what lets the same recipe sit in a contents page, appear in an index, and be laid out to the same template as every other recipe in the book. The difference between a folder of photos and an actual cookbook is almost entirely this step.

Why this is the step that mattered

Fifteen minutes a recipe is the honest figure for typing up someone else's handwriting - longer when the card is faded, or when you keep stopping to squint at whether that's a 5 or an 8. Forty cards is ten hours. Ten hours is not an evening; it's a project that needs a free weekend, and free weekends are exactly what people don't have.

Bringing that down to the time it takes to take a photo and read the result back is the difference between a cookbook you finish and one you keep meaning to start.

Getting good results from it

The scanner is only ever as good as the photo you give it. Everything here takes seconds and makes a real difference:

• Lay the card flat. On a table, not in your hand. A curled or tilted card distorts the text.
• Shoot square-on. Directly above the card, not at an angle.
• Fill the frame. Get close enough that the card takes up most of the photo, but keep all four edges in shot.
• Use window light. Daylight from the side, overhead light off.
• No flash. It blows out the middle of the card and throws a hard shadow off the edge, which is the worst of both.
• Don't crop into the margins. Notes squeezed up the side of a card are often the best part of it - "double this, he always wants more".

The lighting principles are the same ones that apply to photographing food on a phone: window light, no flash, and never mix daylight with a kitchen bulb.

Front, back, and recipes that run long

Plenty of recipes don't fit on one side of a card. The method carries on around the back, or spills onto a second card, or there's a note added years later in a different pen.

Add each photo to the same recipe and the scanner reads them together as one. Photograph them in reading order and you'll get a single complete recipe rather than three fragments.

Worth being clear about the limit: it's one recipe at a time. You can give one recipe as many photos as it needs, but photographing a stack of twenty different cards won't produce twenty recipes - it'll try to make sense of them as one. Work through the tin card by card.

What to always check

Read every scanned recipe back against the card before you approve it. Not because the scanner is unreliable, but because of the kind of mistake it makes when it does slip. It won't produce obvious nonsense you'd catch at a glance - it'll produce something entirely plausible that happens to be wrong.

The things to look at, in order of how much damage they do:

• Quantities. A misread 5 for a 6, or a decimal point in the wrong place. Ingredient lists are where this hurts.
• Oven temperature. Especially whether an old card meant gas mark, °C or °F. A card written in the 1970s may well mean gas mark 4.
• Tin and tray sizes. Often abbreviated to something only the writer understood.
• Faded pencil. If you struggle to read a line, assume the scanner did too, and check that one properly.
• Abbreviations. "tsp" and "tbsp" are worth a glance - they're one letter apart and three times the quantity.

Nothing goes to print without you approving it, so there's a proper opportunity to catch all of this. Use it. If a recipe matters enough to go in the book, it matters enough to read once.

Keep the handwriting

Don't delete the photo once the recipe is typed up. Upload it alongside the finished recipe and it can be printed with it.

This is consistently the page families react to most: the original card in the handwriting of the person who wrote it, next to a version you can actually cook from. The typed recipe is the useful half. The photograph is the half people keep the book for.

If you're working through a whole tin, it's worth reading how to digitise old handwritten recipe cards first - the scanner handles the transcribing, but the gathering, file-naming and backing-up around it still repay doing properly.

Common questions

Does it actually read handwriting?

Yes - that's the point of it. Printed text is the easy case; the reason this exists is the tin of index cards written in biro forty years ago. Very ornate cursive and badly faded pencil are still the hardest things to read, for the scanner and for you, so those are the ones to check most carefully.

Can I photograph a whole stack of cards at once?

No. It's one recipe at a time. You can give a single recipe several photos - front and back, or a method that runs over two cards - and it reads them as one recipe. But a photo containing twenty different cards won't come back as twenty recipes.

How accurate is it, and do I really need to check?

It's good, and on a clear photo of a legible card it's very good. Check anyway. The errors worth catching are the plausible ones - a wrong quantity, a temperature in the wrong scale - and those are invisible unless you compare against the card. It takes under a minute per recipe and you're reading the book through before printing regardless.

Can I scan a recipe out of a cookbook or a magazine?

We'd steer you away from it. A published recipe is someone else's work, and copying it into a book you print and hand out is a different thing from keeping it in your own kitchen. If a bought cookbook's dish has become a family fixture, the better move is to write up how you make it now - the changes you've made, the timings that actually work in your oven, why it ended up in the rotation. That's a more honest page, and a more interesting one.

What if the card is so faded I can't read it myself?

Photograph it anyway and see what comes back - a scanner reading contrast rather than meaning sometimes gets further than a person does. If what it produces looks like guesswork, treat it as guesswork. Ask whoever still cooks the dish, and keep the photo of the card in the book either way.

Does the original photo get kept?

Yes. The photo stays attached to the recipe and you can print it alongside the typed version. Scanning a card doesn't replace it with text - it gives you both.