The Frustration Everyone Has Had
You receive a PDF — maybe a contract, a textbook chapter, or a printed form someone scanned and emailed. You try to highlight a sentence. Nothing happens. You try to select text to copy a paragraph. The cursor selects a giant rectangle instead of words. You hit Ctrl+F to search for a specific term. "No matches."
The PDF looks like text. It has letters, words, sentences arranged in paragraphs. So why can't you interact with any of it?
The answer is simple: what you're looking at isn't actually text. It's a picture of text.
Two Very Different Kinds of PDFs
PDF is a flexible format that can contain almost anything: text, images, vector graphics, form fields, embedded fonts, JavaScript, even video. But for the purposes of this question, every PDF you'll ever encounter falls into one of two categories:
Text-based PDFs. These are PDFs created digitally — exported from Microsoft Word, generated by a report tool, saved from a web browser, or produced by any software that knows about text. The PDF file actually stores the characters of the text, the font used to display them, and the positions where they should appear on the page. When you select a word, the PDF reader knows exactly which characters you're selecting.
Image-based PDFs. These are PDFs created from physical paper — typically through a scanner, a photocopier with "scan to PDF" features, a phone camera app like Adobe Scan or CamScanner, or sometimes from screenshots. The PDF file contains a high-resolution photograph of each page. From the PDF's perspective, there is no "text" on the page — just pixels arranged in patterns that happen to look like letters to human eyes.
This is why selection, highlighting, and searching don't work on scanned PDFs. The PDF reader has no idea what those pixels mean. To the software, the page is functionally identical to a photo of a sunset or a picture of a cat — just a colorful image.
Here's how the two types compare at a glance:
| Behavior | Text-based PDF | Image-based PDF |
|---|---|---|
| Select & copy text | Yes | No |
| Search (Ctrl+F) | Yes | No |
| Highlight words | Yes | No |
| Stays sharp when zoomed | Yes | No (pixelates) |
| Typical source | Word, browsers, report tools | Scanners, phone cameras |
| File size | Small | Large (often much larger) |
How to Tell the Difference
You can check which type of PDF you have in two ways:
The cursor test. Try to select a single word. If your cursor turns into a text-selection cursor and highlights only that word with a colored selection, it's a text-based PDF. If you instead get a large rectangular selection that covers a chunk of the page (or part of an image), it's image-based.
The zoom test. Zoom in to 400% or more. Text in a text-based PDF stays perfectly crisp at any zoom level because it's rendered from font data on the fly. Text in an image-based PDF gets visibly pixelated or blurry, because you're just scaling up a fixed-resolution image.
What OCR Actually Does
OCR stands for Optical Character Recognition. It's the technology that bridges the gap between image-based and text-based PDFs.
When you run OCR on a scanned PDF, software analyzes the image of each page, identifies regions that look like letters, recognizes which character each shape represents, and produces a text-based representation of what it sees. The output is then attached to the PDF as an invisible text layer behind the image.
The image stays exactly as it was — the page still looks identical when you open it. But now, behind the scenes, there's a hidden layer of actual text that lines up with the image. When you click and drag to select a sentence, you're really selecting the invisible text underneath the picture, which happens to be positioned to match what you see.
This is how a scanned book can become a searchable, copyable document without any visible change.
The Limitations Worth Knowing
OCR is impressive but it isn't magic. The quality of the recognition depends heavily on the input:
Resolution matters. A scan at 72 DPI (typical for low-quality phone photos) will produce many recognition errors. 300 DPI is the standard minimum for reliable OCR. Below 150 DPI, expect significant mistakes.
Image quality matters. Blurry, skewed, or shadowed scans confuse OCR engines. Modern engines correct for some rotation and lighting, but a clean, flat scan always produces better results than a phone photo taken at an angle.
Fonts matter. OCR works best with common, clean fonts like Arial, Times New Roman, or standard book typography. Handwriting, decorative fonts, and very old printed text (like 19th-century books with worn type) produce more errors.
Layout matters. Multi-column layouts, tables, and complex formatting can confuse OCR. The software might read across columns instead of down, or it might miss table structure entirely.
Even high-quality OCR typically has an accuracy rate of 95-99%, which sounds excellent — but that means 1-5 errors per 100 characters, or roughly one error every line or two of text. For a 200-page book, that's potentially thousands of mistakes. Some are obvious ("0" instead of "O"), but many are subtle and require human review.
When Image-Based PDFs Are Actually Useful
It's worth noting that image-based PDFs aren't always a mistake. There are legitimate reasons to keep a PDF as an image:
- Legal documents where signatures, stamps, and exact visual appearance matter
- Historical preservation where you want to record exactly what the original looked like
- Forms with handwriting where the handwritten responses are the content
- Documents you don't want easily edited — modifying an image-based PDF is much harder than modifying text
If you've ever received a "PDF" of a signed contract, it was probably image-based on purpose. The signature is an integral part of the document, and converting it to text would lose the signature itself.
What You Can Do About It
If you have a scanned PDF and need to select, copy, or search the text inside it, you have a few options:
Run OCR on it. This is the standard solution. Many tools can OCR a PDF and produce a searchable version. The original layout stays visible, but the text layer is added underneath.
Retype the parts you need. If you only need a small portion of the document — a single quote, a phone number, a paragraph — it might be faster to just type what you see than to OCR the entire document.
Use AI to extract the text. Modern language models can read images and extract text content. For one-off needs, uploading a single page image to a chatbot and asking "what does this say?" is often quicker than running formal OCR. Just remember that this sends your document to a remote server — something worth considering for anything sensitive, as we cover in the hidden risks of uploading PDFs to free online tools.
Accept that some PDFs are just pictures. For documents you don't need to interact with — a contract you're just reading, a flyer you're filing — it's fine to leave the PDF as-is.
Frequently Asked Questions
Can I convert a scanned PDF to editable text?
Yes — running OCR on a scanned PDF produces a text layer you can select, copy, and search. To get fully editable text (for example, to paste into a Word document), you extract the OCR output. Keep in mind that OCR accuracy depends on scan quality, so you'll usually need to proofread the result.
What DPI is best for OCR?
300 DPI is the widely accepted standard for reliable OCR. Scanning at 150 DPI can work for clean, large text but produces more errors. Below 150 DPI, accuracy drops sharply. Going above 600 DPI rarely improves results and just creates much larger files.
Is OCR 100% accurate?
No. Even high-quality OCR typically achieves 95-99% character accuracy. That sounds high, but it means roughly one error every line or two. Clean scans of common fonts produce the best results; handwriting, decorative fonts, and low-resolution scans produce many more errors.
The Bottom Line
If your PDF won't let you select or search text, it's not broken and you're not doing anything wrong. The file simply doesn't contain text — it contains a picture of text. OCR can extract the text into a searchable layer, but the quality of the result depends on the quality of the original scan. Understanding the difference between text-based and image-based PDFs is the difference between fighting your software and choosing the right tool for the job.