Illustration of a scanned page with a transparent text layer floating just above its surface

PDF guides

Why an OCR'd PDF Still Looks Like a Scan

An OCR’d PDF looks identical to the scan it came from, yet you can search and select its text. This is the actual mechanism behind that — not a metaphor, the real technique this tool uses.

Ready to try the tool this guide describes?

What "invisible text layer" actually means

The PDF format has a built-in text rendering mode that tells a viewer to process text (for selection, search, and copy) without painting any pixels for it — the standard calls this the "invisible" rendering mode. OCR PDF writes each recognized word using that mode, directly on top of the original scanned page image, which is left completely untouched underneath.

This is different from simply pasting extracted text into a text box or a separate page: the word is positioned at the same coordinates as the visible word in the scan, so selecting a sentence on the page selects the text that actually appears there.

How a scanned word's location becomes a location on the page

The OCR engine reports where it found each word as a pixel-based bounding box on the rasterized image it was given. That box is converted from image pixels into the PDF’s own coordinate system, accounting for the scale the page was rendered at and the page’s own rotation, so the invisible text lands in the right place even on a sideways scan.

Why low-confidence words are dropped

Not every recognized "word" is embedded. Words the OCR engine is not reasonably confident about are filtered out before they ever reach the PDF. Embedding low-confidence guesses would make the file worse to search, not better — a search that returns garbled noise is less useful than a search that simply doesn’t find that one word.

What this does and doesn't change about the file

The visible content stream of a processed page is never rebuilt or replaced — only new, invisible text objects are added. A page that already had real text is skipped entirely for exactly this reason: adding a second invisible layer on top of existing real text would be redundant, not helpful.

How OCR Text Layers Work FAQ

Is the invisible text a special Looty Tools trick?
No — it uses a standard PDF text-rendering mode that every compliant PDF viewer already supports for exactly this purpose.
Why doesn't the page look different after OCR?
Because the original page image is never modified — only new, invisible text is added on top of it, at the position of each recognized word.
Can I make the OCR text visible?
Not through this tool. The invisible rendering mode is intentional so the page keeps its original scanned appearance.
Why is the text position only approximate, not pixel-perfect?
Each word’s invisible text box tracks its visible counterpart closely enough for reliable selection and search; exact character-by-character width matching is not required since the text itself is never rendered.
Does this technique work the same way on a rotated scan?
Yes — the page’s own rotation is accounted for when mapping each recognized word’s position, so the invisible text still lines up correctly.

Related guides

Open the tool

Jump into OCR PDF when you are ready to process your files.

← Back to all guides

More from Looty

Explore Looty’s Ecosystem

Discover more ways Looty can help you learn, organize, create, and make an impact.