What Happens If You Download Without Drawing Any Redaction Boxes?
Every other page on this site assumes a box got drawn and asks what survives in the finished file. This one starts a step earlier, at the toolbar itself, for the moment the box count is still zero and you press Download anyway.
The rest of this site treats redaction as something that has already happened. It asks what a finished, boxed-in file still leaks, whether the text under a black rectangle can be recovered, whether the hash or the page count or the search index changed as a result. Every one of those questions assumes at least one rectangle exists somewhere in the document.
This page is about the case where none does. You load a file, look at it, and press the big blue Download Redacted PDF button without ever clicking and dragging on the page — maybe to test the tool, maybe out of habit, maybe because you cleared a box you didn't like and got pulled away before redrawing it. The question is narrower and more mechanical than anything else here: what does that button actually require before it runs, and what happens to your file when the requirement turns out to be nothing?
What the Download button actually checks
Checked directly against the script this page loads, the answer is almost nothing. The function that controls every toolbar button's enabled state sets downloadBtn.disabled = !state.pdfDoc — the only condition is that a PDF document object exists in memory. There is no companion line anywhere that reads a box count, totals the rectangles across all pages, or refuses to proceed when that total is zero. Compare that to the two buttons sitting immediately to its left: Undo and Clear page both grey themselves out the instant the current page's box array is empty, because their enabled state is tied to boxes.length. That pairing quietly trains you to expect that a greyed-out control means "nothing here to act on" and, by extension, that an active control means there is something to act on. Download breaks that pattern. It is active the moment a file loads and stays active no matter how many — or how few — rectangles you have drawn anywhere in the document.
What happens to the file anyway
Pressing Download runs the same export routine regardless of box count, and that routine does real, irreversible work even with nothing marked. It creates a brand-new PDF document from scratch, then walks every page of your original one at a time: each page is rendered onto an offscreen canvas at twice its natural size, a white background is painted in first, and the result is encoded as a JPEG at 0.92 quality. Only after that does the code loop over whatever rectangles exist for that page and fill them in black — and when the array is empty, that loop simply runs zero times. Nothing about the surrounding pipeline notices or changes. The freshly rasterized, JPEG-compressed page is embedded into the new document exactly as it would be if it carried ten redactions, and the finished file is finalized with its Producer and Creator metadata fields both overwritten to "HidePDF."
The consequences are the same ones a real redaction produces, minus any actual coverage. Selectable text is gone, because the output pages are images, not text streams — there is nothing left to highlight or search, the same loss described for genuinely redacted exports elsewhere on this site. Embedded fonts and vector line work are gone too, replaced by whatever a JPEG encoder does to them at 92% quality. The file's hash is unrecognizable from the original's, for the reasons laid out in does redacting a PDF change its file hash — a rebuilt document with a fresh creation timestamp shares no bytes with its source no matter what was or wasn't boxed. File size usually moves as well, sometimes up and sometimes down depending on how compressible the original content was versus a photographic JPEG of it; the mechanics are the same ones covered in does redacting a PDF change its file size or page count. None of this is announced. The status line simply says "Done. Redacted file downloaded," which is accurate about the action taken and silent about the fact that the action redacted nothing.
When this actually bites you
The scenario that causes real harm is testing the tool on a document you actually care about. It is a reasonable instinct — drop in a real file to see how the viewer handles it, how pages render, how the toolbar behaves — and then, satisfied it works, press the same button you would press when finished. Nothing stops you, because nothing is checking. What leaves the browser's download folder a second later is a degraded, re-encoded, non-searchable copy of a sensitive document that was never actually covered anywhere, and it now sits on disk looking enough like "the redacted version" that it can get attached to an email by mistake.
A second version happens mid-job. You draw boxes across most of a document, reach the last page, decide the box you just drew on it is in the wrong spot, hit Clear page meaning to redraw it immediately — then a call comes in, or a tab switch happens, and by the time you return the page in front of you has a visible zero in the box counter. If that is the page you stop on and you press Download without noticing, every page still gets flattened through the same pipeline; only that one page ships with no coverage at all, and the file as a whole looks finished because every other page does carry its boxes.
A third is a handoff. Someone opens a long scan purely to confirm it loaded correctly and has the right number of pages before passing the actual redaction work to a colleague, and closes out the way every other page here trains you to close out — Download, because that is the button that ends the session. The file that leaves is not a neutral "I just looked at it" artifact. It is a fully rebuilt copy, with nothing marked, now in circulation under a name that says "-redacted" at the end of it.
Common mistakes and misconceptions
Assuming the button would stop you. It has no mechanism to. The disabled state is tied only to whether a file is loaded, never to how many rectangles exist anywhere in the document.
Assuming a zero-box download is harmless because "nothing was redacted." The file is still fully rebuilt — raster pages, a new hash, usually a different size, and HidePDF stamped into the Producer and Creator fields. It is not the file you opened, even though it was never marked.
Trusting a glance to confirm nothing changed. At normal zoom a flattened page can look identical to the source. Try to select a line of text, search the document, or zoom into fine print, and the raster nature becomes obvious — but by then the file has usually already been sent.
Treating this as a bug to report. It is a consequence of one code path handling every box count the same way, with no special case carved out for zero. There is no planned fix to watch for, because there is nothing broken in the conventional sense — the fix is procedural: read the box counter before you press the button, every time.
Confusing an intentional flatten with an accidental one. Rasterizing a PDF with no redaction marks is a legitimate thing to want occasionally — a file with no live text layer or form fields, for compatibility reasons. The failure mode here is not that the capability exists; it is sending that output to someone under the belief that it represents a redaction when the box count behind it was zero.
Related guides
See also, or use the redaction tool above:
Frequently asked questions
Does HidePDF block the download if I haven't drawn any redaction boxes?
No. Checked directly against the script this page loads, the Download Redacted PDF button is disabled only while no PDF is loaded at all — the exact line is downloadBtn.disabled = !state.pdfDoc, with no reference anywhere to how many boxes exist on any page. That is a real contrast with the two buttons next to it: Undo and Clear page both grey themselves out the instant a page's box count reaches zero, which trains you to read a greyed-out button as 'nothing to act on here.' Download never behaves that way. It stays bright blue and clickable from the moment a file loads, whether you have drawn forty boxes or none, because the code path it triggers treats 'rebuild every page' as the job regardless of what, if anything, you marked.
If I download with zero boxes drawn, is the file I get identical to the one I uploaded?
No, even though nothing was redacted. The export function walks every page, renders it onto an offscreen canvas at twice its natural size, fills in a white background, and encodes the result as a JPEG at 0.92 quality — then embeds that image into a brand-new PDF document created from scratch. That loop runs exactly the same way with an empty box array; the only difference an empty array makes is that the fillRect call that burns in black rectangles has zero rectangles to draw. Everything else proceeds identically: the original's selectable text, embedded fonts and vector edges are gone, replaced by a raster image. The output gets a fresh hash (see does a redacted PDF change its file hash, which covers why even identical runs never match byte-for-byte), typically a different file size, and new Producer and Creator metadata fields both set to 'HidePDF.' It looks like your document. It is not your document's bytes.
Why would anyone download a PDF here without drawing any boxes in the first place?
Usually not on purpose. The common path is loading a real, sensitive file just to see how the tool behaves before committing to drawing anything, then reaching for the obvious big blue button out of habit — it sits in the same toolbar position on every page, next to New PDF, and pressing it is the natural way to end a session. A second path is clearing boxes you decided were wrong, meaning to redraw them, getting interrupted, and downloading anyway because the file is 'ready' in your head even though the box counter reads zero. A third is a reviewer who opens a document purely to check page count or orientation ahead of the person who will actually redact it, and closes out the obvious way. None of these trigger a warning, because the button does not look at the count.
Is there a safe way to just flatten a PDF here without accidentally shipping a non-redacted copy?
Yes, as long as it is intentional and labeled honestly. Rasterizing a PDF into flattened image pages is a legitimate, if incidental, use of this same pipeline — some workflows want a file with no live text layer or editable form fields, and a zero-box export delivers exactly that. The risk is purely one of mislabeling: a flattened file is not a redacted file, and nothing in the download's name, metadata or appearance will tell a later reader which one it is. Before you send anything out, glance at the 'Boxes on this page' counter in the status bar for every page you care about, or simply remember whether you dragged a single rectangle. If the count is zero and you needed coverage, start over with New PDF rather than trusting the file already in your Downloads folder.