Hidden Text in PDF
Detects invisible text hidden in a PDF, telling legitimate OCR layers from suspicious content.
The file never leaves your computer — reading the PDF's internal structure (text, color, opacity, position, images) happens entirely locally, with no network call.
What hidden text in a PDF is
A PDF stores text and how it's drawn as separate instructions: the content (the letters) and the appearance (color, opacity, whether something is drawn in front of or behind something else). It's entirely possible — and common — for those instructions to result in text that exists in the file, can be selected, copied and read by any program, yet never appears to the naked eye on screen or in print. This technique has an old legitimate use (a search layer over a scanned document) and a more recent malicious one (hiding an instruction for an automated system inside a document that looks harmless).
Why most cases are NOT suspicious
Practically every 'searchable' PDF generated from a scanned document (Adobe Acrobat, a Google Docs export with OCR, Tesseract) has an entire layer of invisible text: the OCR software reads the page image, recognizes the words, and overlays that text — invisibly — right on top of the image, purely to enable search and copy. That's the overwhelming majority of real-world invisible-text cases in PDFs, and it poses no risk at all. That's why this tool never gives a single verdict: invisible text (Tr 3) covering most of a page and aligned with a background image is treated as informational, not as an alert.
The hiding techniques checked
Five ways to make text technically present but visually absent: invisible rendering mode (the PDF operator Tr 3 tells the reader to never draw the text); a fill color identical (or near-identical) to the background color; opacity close to zero; a position outside the page's visible area (outside the MediaBox); and text drawn normally but covered afterwards by an opaque image — visible only if someone removes the image on top.
How the tool classifies each excerpt
Every hidden text excerpt found falls into one of two categories, never a generic alarm. 'OCR layer' (informational): invisible text (Tr 3) covering at least 75% of the page area, aligned with an image occupying that same area — the pattern of any legitimate searchable PDF. 'Suspicious': everything else — white text on a white background outside a scanned-image context, text outside the page, an isolated excerpt of invisible text with no matching background image, or normal-looking text covered afterwards by an opaque image. Even invisible text over a full-page image stays suspicious when the technique used is color or opacity instead of Tr 3 — no real OCR tool uses either of those two, only Tr 3.
Known limitations
Each excerpt's bounding box is an approximation (sum of character advances, always axis-aligned), not the glyph's exact geometry — heavily rotated text may have its position estimated less precisely. Images repeated in a tiled pattern may not be recognized individually. Rendering mode 7 (clip-only, paints nothing) isn't checked in this version — in practice, virtually all real hiding uses mode 3. The tool reads up to 50 pages of the document.
Frequently asked questions
No. That's the normal, expected behavior of any 'searchable' PDF made from a scanned document — the invisible text is exactly what the OCR engine read from the image, overlaid to enable search and copy.
Because in that case the invisible text is usually a transcript of the entire page (sometimes thousands of characters) — dumping it all on screen wouldn't help the analysis. The tool points out which pages have it, without dumping the whole content.
It means there's text a person reading the page would never see, but an automated system (copy-paste, indexing, an assistant that reads the document and acts on what it finds in it) would read normally. That can be harmless (a forgotten draft, an invisible watermark) or deliberate — it's up to whoever is analyzing it to judge the context.
No — that's a different problem, solved by the Image Steganography tool. This tool only analyzes real text, drawn as text in the PDF; it never decodes patterns hidden inside an image's pixels.
One of the eight modes the PDF standard defines for drawing text (ISO 32000-1) — most of them paint the text (filled, outlined, or both); mode 3 tells the reader to process the text normally (allowing selection, copy, search) without drawing any of its pixels.
Real scans rarely cover 100% of the page (there's usually a small margin left over), but they do cover the vast majority of it. 75% is generous enough to recognize real scans with a margin, without confusing a small, deliberately placed image with a whole digitized page.
No. All analysis happens in your browser — the PDF is never sent to any server.
Nexinon Principles
Privacy
Your data never leaves your browser.No account needed
Use it now, no account or password.Free
No usage limits, no paid plan.Trustworthy content
Full explanation behind every tool, not just the result.