It runs on your PC, not ours. A separate cloud product is planned

Editing

Double-Clicking the Text Does Nothing

Type a word you can see on the page into the Find in document box and press Enter. If it answers “not found” while the word is plainly there in front of you, the page holds no text: what you are looking at is a picture of a document, and no editor on any platform lets you retype that. Everything below follows from the one test.

Support articleUpdated

Checked against the current Windows build. Where an article names a menu, a button or a message, that is the wording in the application, not a paraphrase.

The find box settles it in one keystroke

The Find in document box sits at the top right of the window. A page with text answers with a hit count and highlights what it found. A page without text answers “not found”, and if the word is visibly on the page, that is your answer rather than a search that went wrong.

A second test is the Edit text tool in the Edit tab: click directly on a line and the panel either fills with the words you clicked or says “No text there. Click directly on a line.” From a script, creasepoint find scan.pdf --query Bengaluru does the first test, and creasepoint objects scan.pdf --page 1 lists what the page actually contains — a scanned page is usually one image object and nothing else.

A picture of words is not words

A PDF page holds objects: text runs placed at coordinates, images, vector paths. A scan holds one image that happens to be a photograph of a page. Double-clicking asks which text run is under the pointer, and on a scan the honest answer is none, so nothing opens. That is not a limitation of this application. No editor can retype a glyph that was never a glyph.

Mixed documents confuse the diagnosis. A filing can be typed for two pages and scanned for the rest, and a scanned annexure is often bound to a digital covering letter. Test the page you actually want to change rather than the first one, and test the part of it you want to change — a heading can be real text on a page whose body is an image.

OCR makes the words findable, not editable

The OCR tool sits in the Convert tab and its panel is headed Make searchable. Tick the scripts the document uses under LANGUAGES, leave “Skip pages that already have text” ticked when the document is part typed and part scanned, and press Recognise. Each page is rendered at 300 dpi, recognised, and given an invisible text layer underneath the image — one text object per recognised word, stretched to the box that word occupies on the page.

Search, copying and screen readers start working, and the page looks precisely as it did, because the image was never touched. The recognised words do not become editable: they sit under a picture that still says what it said. Double-clicking will still do nothing afterwards, and that is correct rather than a failure. The result is held in the edit session, so Save or Save a copy once you have checked it.

Double-click works, and opens more than you meant

Where there is text, double-clicking opens the whole block, re-wrapped to its own width. A block is a guess made from geometry — the same font, the same size, even baselines, a shared left edge — so a table with tidy rows can read as one paragraph, and a paragraph that changes font halfway reads as two. Untick “Whole paragraph” in the Edit on page panel and you get the single line instead.

Some lines belong to no recoverable block at all: a heading on its own, a cell in a table, a caption under a figure. Those still edit as lines. If the box that opens covers more of the page than you expected, press Escape rather than typing. Nothing has been staged and nothing has reached the file.

When the words themselves have to change

If the page is a scan and the text has to be genuinely different, recognition is not the tool for it. Find the document the PDF was printed from. Failing that, Export to Word recovers text and paragraphs from a PDF that has text — including one you have recognised — and loses tables, images and the page design on the way, which suits a document you were going to rewrite anyway. The Word export guide sets out what survives the trip.

Recognised text carries its mistakes into whatever you export. Check names, dates and figures against the image before you build anything on them: a machine reading 1,08,600 where the page says 1,08,000 looks entirely calm about it.

IF THIS DID NOT HELP

The two places worth looking next.

Read the matching guide if you want the reasoning rather than the fix, and check limitations before spending any longer on it — some of what people look for here is absent on purpose.