# Scanned Images to PDF: One Ordered, Searchable File

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2026-09-15T08:36:57Z
- Updated: 2026-10-11T19:22:20Z
- Category: Converting scans
- Canonical: https://www.schemalaw.com/blog/scanned-images-to-pdf

> Turn scanned images into one PDF: put the pages in order, combine them, fix size and rotation, then run OCR so the finished file can be searched.

To turn a folder of scanned images into a PDF you can search, do it in two stages. First sort the page images into the document's true order and combine them into one PDF with your computer's print or PDF function. Then run that PDF through OCR, since combining alone gives you pictures in a new container.

The folder usually exists by accident. A scanner was left on a setting that saves each page separately, a client photographed a document one sheet at a time, or a records office sent `page_1.jpg` through `page_23.jpg` in a zip file. Twenty-three files are not a document yet. Making them one is a small job with several easy ways to go wrong.

## Why combine scanned images into one PDF before OCR?

Combine scanned images into one PDF before OCR so that one document is one file with one page count. The recognized text then comes back as a single continuous export in page order, the file can be named, cited and sent as a unit, and nobody downstream has to guess whether a page is missing.

Consider what the alternative costs. Recognize twenty-three images separately and you have twenty-three text or Word files to join, in an order you must get right a second time. A sentence that crosses from one page to the next sits in two files. A search for a name returns a list of file names instead of page numbers. Every later step repeats the sorting problem you could have solved once.

A single PDF also changes how the document behaves in the office:

- **Page references mean something.** "Page 14" is a fixed place in a PDF. `IMG_4471.jpg` is not a page number anyone else can use.
- **Completeness can be checked.** A 23-page PDF either has 23 pages or it does not. A folder can lose a file without any sign.
- **It travels whole.** One attachment, one upload, one entry in the document management system.
- **Limits are counted sensibly.** Tools that count documents treat each loose image as a document. Schema Law's free plan, for instance, allows one document a day of up to three pages, so three page images are three days of uploads when left separate and one upload when combined.

There is one case for leaving an image alone. If the picture is a single page and all you want is the wording, upload it as it is. The post on [getting text from a scanned image](https://www.schemalaw.com/blog/scanned-image-to-text) covers that route and the image types involved.

## How do you put loose page images in the right order?

Put loose page images in order by renaming them with zero-padded page numbers taken from the document itself, then sorting by name. Trust what is printed on each page over the file name or the file date. Before you combine anything, confirm that the count is complete and that no page appears twice.

### Why the existing order is unreliable

Three things commonly scramble a folder of pages.

**Names that sort as text.** A computer sorting names alphabetically puts `page_10.jpg` before `page_2.jpg`, because "1" comes before "2". The result is 1, 10, 11, 12 and so on up to 19, then 2, 20, 21. Some file browsers apply a smarter "natural" sort and some combining tools do not, so a folder that looks right on screen can still be assembled wrong.

**Dates that reflect copying.** Sorting by date modified works only if the files still carry their capture times. Downloading, unzipping or copying can stamp every file with the same minute.

**Capture order that was never page order.** A duplex job scanned as fronts and then backs, a page that was retaken after a blur, a sheet that was fed upside down and redone at the end. The camera roll records what happened, not what the document says.

### A method that holds

Work through a fictional example: the Estate of Marguerite Ellison, where a relative has sent 23 photographs of a trust instrument and its schedules.

1. **Copy the folder.** Work on the copy. The files as received stay untouched in a subfolder named "as received".
2. **Open every image and read its page number.** Use the printed folio, "Page 7 of 21", a schedule letter or the paragraph numbering. Write the true position down against the current file name.
3. **Rename with padded numbers.** `Ellison trust p01.jpg`, `p02`, up to `p23`. Two digits for up to 99 pages, three beyond that. Padded names sort the same way in every program.
4. **Handle pages with no number.** A cover sheet, a signature page or an exhibit tab may carry none. Place it by context: the last words of the page before and the first words of the page after.
5. **Look for gaps and doubles.** If the instrument says 21 pages and its schedules add two more, you need 23. A jump from "Page 11" to "Page 13" is a missing page. Two files both showing "Page 9" are a retake, so keep the sharper one and move the other to a "not used" subfolder instead of deleting it.
6. **Ask for what is missing now.** A request made the same day is easier for everyone than one made after the PDF has been circulated.

A small table kept during step two saves confusion on anything longer than a few pages.

| File as received | Printed page | New name | Note |
| --- | --- | --- | --- |
| IMG_4460.jpg | Cover | Ellison trust p01.jpg | No folio |
| IMG_4461.jpg | 1 of 21 | Ellison trust p02.jpg | |
| IMG_4470.jpg | 9 of 21 | not used | Blurred, retaken |
| IMG_4471.jpg | 9 of 21 | Ellison trust p10.jpg | |
| IMG_4483.jpg | Schedule B | Ellison trust p23.jpg | Last page |

## Combining scanned image files into a single PDF

To combine scanned image files into a single PDF, select the ordered images, send them to your system's print-to-PDF or image-to-PDF function with scaling set so nothing is cropped, and save one file. Then page through the result against your list. No special software is needed for a few dozen pages.

Most computers and phones can do this with what is already installed: a print dialog with a PDF destination, a "create PDF" action in the file browser, or the export function of an image viewer. PDF editing programs offer a "combine files" command that does the same with more control. The menu names differ, and the steps are alike.

1. **Select the renamed images in order.** Click the first, shift-click the last. Check that the selection runs from p01 to p23 and includes nothing else from the folder.
2. **Choose the PDF route.** Print and pick the PDF printer, or use the combine or create-PDF command.
3. **Set one image per page.** Print dialogs for pictures often default to contact sheets or several photos per sheet.
4. **Set the fit.** Pick "fit to page" or "shrink to fit". Avoid "fill page", which enlarges the image until it covers the sheet and cuts off whatever overhangs, often a line of text at the top or bottom.
5. **Leave quality at its highest.** If the dialog offers image quality or a "reduce file size" option, do not reduce. Recognition needs the detail.
6. **Save with the document's name.** `Ellison trust instrument and schedules (combined from 23 photos, 2026-07-14).pdf`.
7. **Verify.** Open the PDF. Confirm 23 pages. Step through from first to last reading only the page numbers. Zoom in on one page of small print and compare its sharpness with the source image.

### What to watch in the result

- **Soft text.** If the PDF page is visibly fuzzier than the image it came from, the route you used resampled the pictures. Try a different one, such as a direct image-to-PDF export in place of printing.
- **Big white margins.** Printing a photo onto a virtual sheet adds margins. That is harmless for reading. It does shrink the image slightly on the page, which is one more reason not to reduce quality.
- **File size.** Twenty-three full-resolution photos can make a PDF of tens of megabytes. Schema Law takes files up to 20 MB, and other tools have their own ceilings. If the combined file is over the limit, split it into two parts at a natural break in the document. The post on [converting a large scanned PDF to Word online](https://www.schemalaw.com/blog/convert-large-scanned-pdf-to-word-online) deals with oversize files in detail.

Keep the renamed images after you combine them. If a page turns out to be wrong, you can rebuild the PDF in a minute from the ordered set.

## Page size and orientation when image files differ

When image files differ in size and orientation, aim for a PDF in which every page is upright and nothing is cropped. Uniform page size is a secondary goal. Rotate sideways images before combining, and let each page keep its own proportions if forcing one size would cut content.

Mixed sets are normal. A bundle may hold letter-size pages, one legal-size page, a small receipt and a spreadsheet printed across the long edge of the sheet. Photographs add their own variation, since each was taken from a slightly different distance.

| What differs | What goes wrong in the PDF | What to do before combining |
| --- | --- | --- |
| One image is sideways | The page reads at ninety degrees; recognition may fail on it | Rotate the image in a viewer and save it |
| One image is upside down | The page looks plausible as a thumbnail and is unreadable | Open every image at full size, not as a thumbnail |
| A longer page among standard ones | "Fill page" crops its top and bottom | Use "fit", or a route that sizes each PDF page to its image |
| A small item, such as a receipt | It is blown up to fill a sheet and looks coarse | Accept a small image on a large page, or keep its own page size |
| A table printed across the long edge | It is shrunk onto an upright sheet with wide blank bands | Let that page be wider than it is tall; do not turn the text sideways to fit |
| Photos taken at different distances | Text size jumps from page to page | Leave it; cropping each image to the paper's edge evens it out if it matters |

### Rotation that sticks

There are two kinds of rotation, and they behave differently. A phone photo carries a tag that says which way up it was taken, and software reads the tag to display the picture upright. Some combining routes ignore the tag and place the raw pixels, so a photo that looked upright in the viewer arrives sideways in the PDF. If that happens, open the image, rotate it and save, which turns the pixels themselves.

Rotating a page inside a PDF program after combining also works, and it is the easier fix for one or two pages. Check the result in a second viewer, because some programs change only how the page is displayed.

### How much tidying is worth it

For a working copy, upright and complete is enough. For a set that will be produced, filed or shown to a client, spend a little longer: crop each photo to the paper's edge so the desk is gone, and keep page sizes consistent where the originals were consistent. Courts that accept electronic filings set their own technical requirements for PDFs, and they differ, so read the rules of the court concerned before you prepare a filing copy.

Do not straighten, sharpen or filter the images as part of this step. Changes like those belong to a separate decision about [image cleanup](https://www.schemalaw.com/blog/ocr-preprocessing), made with the originals still on hand.

## Does converting a scanned image to PDF make it searchable?

No. Converting a scanned image to PDF only wraps the picture in a PDF file. The page is still pixels with no characters behind them, so search finds nothing and no text can be selected. It becomes searchable when OCR reads each page and a text layer is added to the file.

The belief that PDF means text is common, because most PDFs people meet were saved from a word processor and do contain text. The file type says nothing either way. A PDF can hold text, pictures or both.

You can prove the point on your combined file in a few seconds. Open it, press Ctrl+F and search for a word that is plainly visible on the first page. Zero results on a page full of that word tells you the file is image-only.

So "convert image to searchable PDF" is really two conversions in a row:

1. **Images to PDF.** Ordering and combining the page images. Result: one file, correct pages, no text.
2. **PDF to searchable PDF.** Recognition. Result: the same pages with text positioned invisibly behind the picture of each word.

The second stage is where an OCR tool comes in. In Schema Law you upload the combined PDF and choose Searchable PDF from the [export formats](https://www.schemalaw.com/#formats), described there as your original file with a hidden text layer. That export is available on the free plan for a document of up to three pages, which is enough to test the result on a short set. The mechanics of the layer, and what it does and does not change about the file, are explained under [what makes a scanned PDF searchable](https://www.schemalaw.com/blog/how-to-make-a-scanned-pdf-searchable#what-makes-a-scanned-pdf-searchable).

Some scanning apps and PDF programs do both stages in one pass and label the button simply "scan to PDF" or "save as searchable PDF". That is convenient. Run the search test on their output anyway, on a page from the middle as well as the first, so you know which kind of PDF you were given.

### After recognition

A searchable PDF made from photographs deserves a short check before anyone relies on its search results:

- Search for a surname that appears on several pages and see that each one is found.
- Search for a figure, such as a sum of money, exactly as printed.
- Select a line on the worst-looking page and paste it into a note to see what the text layer holds.

A search that returns nothing may mean the word is absent, or that the software misread it. On photographed pages, do not treat an empty result as proof.

## One PDF per document or one per scanning session

Make one PDF per document. A file should begin where a document begins and end where it ends, regardless of how many were scanned or photographed in the same sitting. A session-sized PDF is quicker to produce and slower to use every time afterwards.

The scanning session is an accident of the day: whatever was in the tray, or whatever the client photographed before lunch. The document is the unit you name, date, cite, send and convert.

In the Ellison set, the trust instrument and its schedules are one document. Keep the grantor, the trustee and each beneficiary named in the file title or a cover note. A pour-over will that arrived in the same envelope is a second PDF. The trust export feeds the estate file, and the will feeds probate on its own.

| | One PDF per document | One PDF per session |
| --- | --- | --- |
| File name | Describes the content: party, document type, date | Describes the event: "scans 14 July" |
| Finding something later | Search or browse by name | Open the file and scroll |
| Page references | Match the document's own numbering | Offset by everything in front of it |
| Sending one item to someone | Attach the file | Extract pages first, and risk sending a neighbour's page along |
| Conversion limits | Each file is smaller in megabytes and pages | Large files hit upload and page caps sooner |
| Recognition output | One export per document | One export that mixes unrelated documents |

### Where to cut

Some boundaries are clear: a new letterhead, a new caption, a fresh "Page 1". Others take a decision.

- **A letter and its enclosures.** Keep them together if they were sent together and will be cited together. Split them if the enclosure is a document in its own right that the file will need separately, and name the two so the link is visible.
- **A contract and its schedules or exhibits.** One file. The schedules are part of the instrument. In the Ellison example, the trust instrument and Schedules A and B stay together as 23 pages.
- **An email chain printed and scanned.** One file per chain as printed.
- **A bundle that arrived fastened together.** If the way it was fastened may matter, keep one PDF of the bundle exactly as received and make per-document copies from it for working use.

### If you already have a session-sized file

Split it with the page-extraction function of a PDF program, one document at a time, and name each piece as you go. Keep the session file until every extract has been checked for its first and last page. The usual slip is cutting one page early and stranding a signature page at the front of the next document.

Then run recognition on the per-document files, not on the session file. The text you export will be organised the way the matter is. If you have many documents to process each month, the [pricing page](https://www.schemalaw.com/pricing) lists how many pages each plan covers and how long a single document may be.

## Try it on a real document

Upload a scanned PDF and export it as Word, a searchable PDF or plain text on the free plan. Your documents are never used to train an AI model. [Start free](https://www.schemalaw.com/api/auth/google)