# Is Your PDF Scanned or Text-Based? A Before-Upload Check

Use text selection, copying, and page sampling to distinguish image-only PDFs, hidden OCR layers, and mixed documents before choosing conversion or translation.

A PDF that looks sharp on screen may contain only pictures of words. Another may show the same scanned page while carrying an invisible text layer behind it. The difference affects search, copying, conversion, and translation.

You can make a useful diagnosis without inspecting the file's internals. Test several pages and focus on what the text layer actually gives you, not only how the document looks.

## Select and copy a complete sentence

Open the PDF in a viewer and try selecting a sentence in the body. Paste it into a plain-text editor. Check the characters and their order, including punctuation, numbers, and any accented or non-Latin text.

If selection highlights a whole page image instead of words, the page is likely image-only. If a sentence copies cleanly, it has a usable text layer at that location. If the copied result is gibberish or alternates between columns, selectable text exists but may not be reliable for further processing.

Do not classify the whole book from its cover. Title pages often use different fonts, images, or layout from the main body.

## Recognize a scan with a hidden text layer

OCR can add searchable text over or behind a scanned image. The visible page remains a scan, while search and selection use the recognized text. That can be useful, but an invisible layer can contain mistakes that the sharp page image hides.

Search for a distinctive phrase, then compare the found text with the image. Copy a line with a number or specialist term. Check whether the selection aligns sensibly and whether its content matches what you can read.

A wrong hidden layer is not proof that the image itself is poor. It may reflect earlier recognition, encoding, or reading-order problems. Keep both possibilities in mind before deciding whether to rescan or reprocess.

## Sample mixed and difficult pages

Many PDFs combine digital text with scanned inserts, appendices, or illustrations. A document may also have different layouts across sections. Use a small sample set:

| Page to inspect           | Reason                                |
| ------------------------- | ------------------------------------- |
| Ordinary body text        | Establish the main content type       |
| Middle and final sections | Find changes in the source or export  |
| Two-column page           | Check extraction sequence             |
| Table or formula          | Check numbers, symbols, and structure |
| Notes or appendix         | Find small text and scanned inserts   |

Record which pages are image-only, which copy correctly, and which have unreliable text. That is more informative than a single label saying “scanned PDF.”

If the scan itself is blurry, cropped, or shadowed, address that before relying on recognition. The [poor-scan preparation guide](https://pdfcraft.ai/en/blog/prepare-poor-quality-scans/) describes external preparation without assuming the website offers image-editing controls.

## Choose the route by what you want next

For an image-only source, OCR must recover text before it can become editable content or a translated reading edition. Structure recovery also matters: recognizing every word does not automatically place columns, captions, and notes in the right order.

If you want continued reading with adjustable font size, consider [scanned PDF to EPUB](https://pdfcraft.ai/en/pdf-craft/scanned-pdf-to-epub/). If you want editable content for notes or further processing, consider [PDF to Markdown](https://pdfcraft.ai/en/pdf-craft/pdf-to-markdown/). Keep the original PDF when exact page appearance matters.

For another language, choose a translation route explicitly. In PDF Craft, PDF output requires translation. EPUB and Markdown outputs allow conversion without translation, so enable it when wanted. The [scanned PDF translation guide](https://pdfcraft.ai/en/blog/translate-scanned-pdf/) covers that full task.

## Keep diagnosis separate from acceptance

These checks tell you what the source appears to contain. They do not prove that a later conversion is correct. After processing, inspect the same difficult pages and compare the downloaded output with the source.

Use the [PDF workspace](https://pdfcraft.ai/en/pdf-craft/) to upload the file and review current settings and limits. New users receive trial credits; processing is charged by usage, with current rates on [Pricing](https://pdfcraft.ai/en/pricing/). The [OCR and translation guide](https://pdfcraft.ai/en/blog/ocr-and-translate-workflow/) explains why text recovery, structure, and meaning need separate attention after the source diagnosis.