Home › PDF tools › Scan to PDF

Scan to PDF

Turn paper into one PDF without sending it anywhere: photograph each page with your camera, or pick photos you already have, then tidy the order, choose a scanner filter and a page size, and download a single file named scanned-<date>.pdf. Everything happens in this tab — no upload, no account. And no OCR: the text in a scan stays a picture, as the FAQ explains.

Page source

The camera needs your permission and a secure (https) page. If it is unavailable or you refuse, the reason is printed below and the photo picker keeps working — nothing is uploaded on either path.

Choose images

Select one or many files at once (hold ⌘ or Shift), or add them in batches — on a phone this opens the camera roll. Each picture becomes one page. Picking the same file twice is ignored.

Scanner filter

Original stores the picture as it is: a JPEG is copied into the PDF unchanged and a PNG is stored losslessly with its alpha. Grayscale and B&W are redrawn: B&W picks a threshold from each page's own brightness histogram and is stored as PNG, which keeps flat paper clean.

Page size

No page is ever cropped: each picture is scaled to fit inside the printable area and centred.

Orientation

Auto turns a page sideways when the picture is wider than it is tall, so a landscape page does not sit in the middle of a portrait sheet with empty bands.

Margin

The margin shrinks the picture; it never crops it. On A4 with 10 mm the largest picture that fits is 190 × 277 mm.

Add at least one page to begin. Nothing is uploaded — the PDF is written in this tab.

Pages–

Picture data in–

Filter–

Page setup–

No scan built yet.

The page count and the file size shown after a run are read back from the PDF that was just written, not estimated in advance. This tool adds no watermark, no stamp and no extra page, and your pictures on disk are never modified. What it cannot do is make the text of a scan selectable: there is no OCR here, so the words in the output are pixels. If you need searchable text, start from a PDF that already contains it or run an OCR program on the result.

What this scan to PDF converter does

A scan to PDF tool takes pictures of paper and writes them into a PDF, one picture per page. That is exactly what this page does, with two ways in. The camera path asks the browser for the device camera over the standard getUserMedia API, shows the live picture in a <video> element, and copies the current frame into a canvas each time you press Capture page. The photos path reads pictures from your disk or phone camera roll through the browser's file picker, so it also covers pages you photographed earlier, images someone sent you, or screenshots of a form. Either way the pages land in one list that you can reorder, and the PDF is assembled in this tab by the open-source pdf-lib library and handed back to you as a download. There is no upload step, no account, no queue and no watermark, and the page keeps working with the network switched off once it has loaded.

This is a page builder, not an OCR engine, and that distinction matters more here than anywhere else in the PDF family. A photograph of a page is a picture of paper. This tool does not read characters out of that picture, so the PDF it produces has no text layer at all: the words cannot be selected, copied, searched, or read aloud by a screen reader, and you cannot search inside the file for a name or an invoice number. Something that markets itself as "scanned PDF to text" is doing optical character recognition, which is a different job with different software — this page does not do it and does not claim to. If the text has to stay searchable, start from a PDF that already contains real text (for example a document exported from a word processor), or run an OCR program on the file afterwards, before or instead of using this tool.

What this page does do is make each page clean and printable. It can keep the colour picture, flatten it to grayscale, or push it to high-contrast black and white using a brightness threshold computed from that page's own histogram, which is what makes faint print, pencil and pen strokes stand out. It then places the picture on a page of the size you choose, scaled to fit and centred, never cropped. It does not straighten a crooked page, remove the shadow of your hand, or find the paper's edges and cut away the desk around it — those need a vision pipeline this page does not have, and nothing in the copy here pretends otherwise.

How to use this scan to PDF converter

Choose the page source. Camera captures pages live in the tab; Photos opens the file picker. Photos is the default, because it works on every device, including a desktop with no camera.
If you chose the camera, press Turn on camera and accept the browser's permission prompt. Frame one page, keep the camera square to the paper, and press Capture page. The camera keeps running between shots, so you can work through a stack; press Stop camera when you are done, and it also stops by itself if you switch source, hide the tab or leave the page.
If you chose photos, press Choose images and select one or many files at once — on a phone this is your camera roll. Every picture becomes one page, and picking the same file twice is ignored so you cannot create an accidental duplicate.
Put the pages in order. Drag a row by its ≡ handle, or use the ↑ and ↓ buttons, which work with a keyboard. Remove anything you do not want with ✕, or start over with Clear all.
Choose the scanner filter, the page size, the orientation and the margin. Each row rewrites itself immediately and shows the exact page it will produce, in points and in millimetres, before anything is built.
Press Create PDF. The pages are processed one at a time; the status line counts them off and the readout shows the page count and the final file size, both read back from the file that was just written.
Press Download to save the file — it is named scanned-YYYY-MM-DD.pdf, and your browser puts it wherever downloads normally go. Your original pictures are untouched.

Worked example

Three test pictures were built for this page to mirror a three-page document: page-1.jpg (2400 × 3200 px, 556,266 bytes), a portrait text page; page-2.jpg (3200 × 2400 px, 556,615 bytes), the same kind of page the other way up; and page-3.png (1600 × 1200 px, 8,207 bytes), a flat grey scan. Together they are 1,121,088 bytes (1.07 MB). They were loaded through the photo picker with the settings this page starts with — A4, orientation auto, margin small (10 mm), filter Original — and the downloaded PDF was then opened again with a PDF library to read its real geometry out of the file. All of it was done in Chrome 153 on macOS; the page sizes and placements below are arithmetic and will be the same anywhere, while the byte counts come from that browser's JPEG encoder.

Page 1 — 595.28 × 841.89 pt (A4 portrait, because 2400 is narrower than 3200). The picture is drawn at 538.58 × 718.11 pt, centred at x = 28.35, y = 61.89.
Page 2 — 841.89 × 595.28 pt (A4 landscape). The picture is drawn at 718.11 × 538.58 pt, centred at x = 61.89, y = 28.35.
Page 3 — 841.89 × 595.28 pt (A4 landscape: 1600 is wider than 1200, and Auto follows the picture). The picture is drawn at 718.11 × 538.58 pt.

Nothing was cropped. With a 10 mm margin the usable area of an A4 portrait page is 538.58 × 785.19 pt (190 × 277 mm), and of a landscape page 785.19 × 538.58 pt; each picture was scaled down until it fitted inside that box and then centred, which is why page 1 is limited by its width while pages 2 and 3 fill the width of their sheet. The file came out at 1,126,470 bytes — 5,382 bytes more than the 1,121,088 bytes of pictures that went in, because the two JPEGs are kept in the PDF exactly as they were (556,266 and 556,615 bytes appear in the output unchanged) and only the page structure and the file's own metadata are added on top. Two builds of the same three pages differ in exactly one thing — the creation timestamp pdf-lib writes into the document information dictionary — so a repeat build can land a byte or two either side of that total, which is why the readout in this tool reads the size of the file it just wrote instead of predicting it.

The scanner filters, on the same three pages with the same page setup, measured on the files that came out and then checked by decoding what is embedded in them:

Filter	Output file	What the stored pages actually contain
Original	1,126,470 bytes (1.07 MB)	The two JPEGs appear in the PDF byte for byte; the PNG is re-stored losslessly at its original 1600 × 1200 pixels.
Grayscale	626,831 bytes (612.1 KB)	Colour is gone: 0 of the 11,040 sampled pixels across the three embedded JPEGs had unequal red, green and blue values.
High-contrast B&W	131,687 bytes (128.6 KB)	Only two values exist anywhere in the three pages — 0 and 255 — and the white share was 79.2%, 79.9% and 85.0% per page. An independent implementation of the same threshold rule on the same three pictures gives 79.2%, 79.9% and 85.1%, which is the check that the page is really thresholding its own histogram and not guessing.

The size differences between those rows are large, and they are described here exactly as measured on these three pictures — not as a promise about yours. A detailed photograph keeps far more pixels than this test page's flat ink, and the file size follows the pixel count, so the same settings on your own scan will land somewhere else. That is why the readout after every run is measured from the file that was just written.

Two more things were checked in the same session, because both are visible in the row list. Moving page-1.jpg down one step with the ↓ button and building again produced a file whose page 1 is the 3200 × 2400 landscape picture at 841.89 × 595.28 pt: the order in the list is the order in the PDF. And the camera path was exercised with a browser's synthetic video device at 1920 × 1080 px, where three Capture page presses gave three A4 landscape pages, each drawing its frame at 785.19 × 441.67 pt, in a 74,676-byte file. A real phone camera delivers its own resolution, and a larger frame makes a larger PDF — that is the one number in this section that will certainly differ on your device.

Finally the page settings, again taken from real runs of this tool on the same three pictures:

Settings	Page 1 (pt)	Page 2 (pt)	Page 3 (pt)
A4 · auto · 10 mm	595.28 × 841.89	841.89 × 595.28	841.89 × 595.28
A4 · auto · none	595.28 × 841.89	841.89 × 595.28	841.89 × 595.28
Letter · landscape · 20 mm	792 × 612	792 × 612	792 × 612
Fit to image · no margin	1800 × 2400	2400 × 1800	1200 × 900

Read the table left to right: the margin never changes an A4 or Letter sheet, only how much of it the picture fills, and forcing landscape puts every page the same way up — useful when the sheets will be stapled, awkward when one of them holds a tall portrait page, because that picture simply becomes smaller. Fit to image makes the page follow the picture instead: those pages are 1800 × 2400 pt, 2400 × 1800 pt and 1200 × 900 pt, and the first of them is 25 × 33.3 inches. That is right for a poster and wrong for a printer tray, which is why A4 is the default.

Who is this scan to PDF converter for?

Anyone with paperwork and no scanner: ten pages of a rental agreement, a passport copy, an insurance form — photograph them with the phone you already have and send one PDF instead of ten loose pictures.
Office and admin work: turn signed delivery notes, invoices or a handwritten approval into a single file for a records system, on a work laptop, with no installation and no third-party upload to justify to IT.
Students and teachers: scan notes, worksheets or a marked assignment page by page and hand in one ordered PDF, in the right page order, without a flatbed scanner in the room.
People handling confidential papers: payslips, medical letters, bank statements and identity documents can be scanned here because the frames never leave the device — there is no upload step to trust, no server copy and no account.
Field and site work: meter readings, stock sheets, a damage report written on paper — capture and assemble the PDF on the spot, with no signal needed once the page has loaded.
Sellers and claimants: build the evidence pack a marketplace dispute, a warranty claim or an expense report asks for, with the pages in the order the claim form expects.

Filters, page sizes and what the output really is

The three filters are not three levels of the same thing — they produce different kinds of page, and the table below is the honest version of what each one does to the pixels and how the page is stored inside the PDF.

Filter	What happens to the picture	How the page is stored
Original	Nothing. Colour, contrast and resolution stay as they were.	A JPEG is stored in the PDF byte-for-byte with no re-compression; a PNG is decoded and re-stored losslessly, alpha included. WebP, GIF, HEIC and anything else a PDF cannot hold are redrawn on a canvas and stored as JPEG at quality 0.85.
Grayscale	Every pixel is replaced by its brightness (299/587/114 weighting of red, green and blue), so colour information is gone for good.	Redrawn on a canvas and stored as JPEG at quality 0.85.
High-contrast B&W	Grayscale, then every pixel becomes pure black or pure white at a single threshold derived from that page's own brightness histogram, so faint print and pen strokes survive while paper texture and light shadows are pushed to white.	Stored as PNG, so large flat areas cost almost nothing and no JPEG speckle appears around small text.

Page size. A4 (595.28 × 841.89 pt) and Letter (612 × 792 pt) fix the sheet, and the picture is fitted inside it with your margin subtracted from all four sides, then centred. Fit to image does the opposite: the page is the picture measured at 96 pixels per inch (0.75 pt per pixel) plus the margin, which means a phone photo of 3000 × 4000 pixels produces a page about 31 × 42 inches. Auto orientation turns a sheet sideways when the picture is wider than it is tall, which is what stops a landscape photo sitting in the middle of a portrait page with two white bands. A picture smaller than the printable area is stretched up to fill it — no pixels are thrown away, but a small picture blown up to A4 looks soft on paper.

Resolution is the one thing that decides the size of your file. Every page in the output is an image, and image bytes are governed by pixel count: a 4000 × 3000 frame from a phone camera holds twelve million pixels, and JPEG at quality 0.85 spends a fraction of a byte on each. Ten such pages are a large PDF, and no setting on this page changes that arithmetic — there is no "compress scan" slider here and no claim of a size reduction. If the result is too heavy for an email gateway, choose Grayscale or High-contrast B&W, which cost far less than colour, capture at a lower resolution, or send the file through Compress PDF afterwards and read the measured sizes it reports.

Limits to be plain about. There is no OCR, so no text layer, no search inside the file and no way for this or any later step to select a word you can see. There is no deskew, no shadow removal and no automatic edge detection — the geometry of your photo is the geometry of the page. The camera path needs permission and a secure connection, and it is deliberately optional: if it fails, the reason is shown and the photo picker carries on. Everything is processed in your browser's memory, so the practical ceiling is your device's RAM rather than a page counter; on a phone, keep the tab in the foreground while a long scan is being built.

Common use cases

Scanning a multi-page contract or tenancy agreement with a phone and sending one PDF instead of twelve photographs.
Photographing a stack of receipts or delivery notes into one claim file, in date order, before the paper is thrown away.
Making a scan of an ID page, a signature page and a completed form into the single attachment an application portal demands.
Turning a whiteboard or lecture notes into a high-contrast black-and-white PDF that reads clearly when printed.
Copying a chapter out of a book for personal study, keeping the pages in order and the file on your own machine.
Building a scanned proof of condition — a rented flat, a returned parcel — where each page is a dated picture of the item.

Pro tips

Light the page evenly and keep the camera square. The tool cannot straighten a photo or remove a shadow, so the flat, evenly lit frame taken now saves a re-shoot later. Put the paper on a plain, darker surface so the page is easy to see while framing.
Use High-contrast B&W for print and handwriting. It is the filter that makes pencil, faint print and thin signatures legible, and being PNG it also produces the cleanest flat areas. Keep Original when colour matters — stamps, highlighted text, a red seal.
Decide the page size before you scan a stack, not after. Nothing is lost by changing it afterwards, but a scan destined for a printer tray belongs on A4 or Letter with a 10 mm margin, and one that will only ever be read on a screen is fine as Fit to image.
Capture in reading order. Pages land in the list in the order you shoot them, so working front to back means no reordering at all — and if you do need to fix something, the ↑ and ↓ buttons move a page one step and work from a keyboard.
Long scans: desktop browser, few tabs. Every page is held in the tab until the PDF is written. Thirty pages from a phone camera is ordinary work; a hundred large frames is when memory, not page count, becomes the constraint.
Download before changing a setting. Changing the filter, size or page list clears the built file, because it no longer matches what the screen shows. Save it first, or accept that you will press Create PDF again.

FAQ

Are my scans uploaded to a server?

No. Camera frames are copied from the video element into a canvas on this page, photos are read from your disk with the browser's File API, and the PDF is written in the same tab by pdf-lib. Nothing is uploaded, no copy is stored on a server, and no account or email address is needed.

Does the PDF have OCR? Can I search or select the text?

No. This page has no OCR, so the text in your scan stays a picture of paper. In the downloaded file nothing is selectable, copyable, searchable or readable by a screen reader, and there is no text layer to extract. "Scanned PDF to text" is a different job — run an OCR program on the result if you need that, or start from a PDF that already contains real text.

Can I scan with my phone camera?

Yes, in a browser that supports getUserMedia over a secure https connection. On a phone the tool asks for the rear (environment) camera, and each press of Capture page adds one page while the camera keeps running between shots. Camera permission is required, and it stays under your control in the browser.

What happens if the camera is unavailable or I deny permission?

The reason appears in plain text next to the controls — permission refused, no camera found, the camera is busy in another app, or an insecure connection — and the source switches to the photo picker, which works exactly as before. The camera path never blocks the rest of the tool.

Which scanner filter should I choose?

Original keeps the picture as it is; Grayscale converts it to shades of grey and is kinder to stamps, signatures and coloured ink; High-contrast B&W pushes every pixel to pure black or pure white using a threshold computed from that page's own histogram, which is what makes faint print and pen marks legible. B&W pages are stored as PNG so flat paper stays clean instead of picking up JPEG speckle.

Can I reorder or delete captured pages?

Yes. Each row has ↑ and ↓ buttons that work with a keyboard, a ≡ handle you can drag, and a ✕ button that removes that page. Page numbers follow the list from top to bottom, and the next PDF you build uses that order.

Which page size should I choose?

A4 is the default and suits most printers outside North America; Letter is the US equivalent; Fit to image makes each page follow its own picture, measured at 96 pixels per inch plus your margin. A4 and Letter pages do not change with the picture, so a tall photo is scaled down to fit rather than being cropped.

Does the tool crop or rotate my photos?

It never crops. Each picture is scaled until it fits inside the printable area of the page, keeping its own proportions, and then centred, so nothing at the edges is lost. Orientation Auto turns a sheet sideways when the picture is wider than it is tall, and a portrait or landscape page follows your choice. A JPEG carrying an EXIF rotation flag is redrawn through a canvas, which applies the flag, so a phone photo does not arrive on its side.

Why is the PDF so large?

Because it contains pictures, and camera frames carry a lot of pixels. The size is driven by the resolution of the frames, the number of pages, and how each page was stored: a source JPEG is copied in unchanged, a re-encode for Original or Grayscale is JPEG at quality 0.85, and high-contrast B&W is stored as PNG. Fewer and smaller pages make a smaller file, and this page reports the measured size of every file it builds rather than promising one.

Does it remove shadows, straighten a crooked page or detect the paper edge?

No. There is no deskew, no shadow removal and no automatic edge detection, because those need a vision pipeline this page does not have. The high-contrast filter can make a badly lit page easier to read, but it cannot repair the geometry of the photo — for a clean result, put the page on a plain surface and keep the camera square to it.

Can I add another page after the PDF is built?

The built file is cleared as soon as the page list or a setting changes, because it no longer matches what the screen shows. Capture or choose the extra page, press Create PDF again, and download the new file; any copy you already saved stays where your browser put it.

Is this scan to PDF tool free, and does it add a watermark?

Yes, it is free with no sign-up, no email gate, no page counter and no watermark, stamp or extra page on the output. The only limit is your device's memory, because every page is held in the tab until the PDF is written.