How the scanner works
A photograph of a page is not a scan. It is taken at an angle, so the page is a trapezium rather than a rectangle; it is lit from one side, so one edge is brighter than the other; and the paper itself is rarely flat. This tool turns that photograph into something closer to what a flatbed scanner would have produced, in five steps.
- Find the page. The image is reduced to a working copy, split into paper and background, and the outline of the largest sheet-shaped region is measured. If nothing convincing is found, the scanner says so rather than guessing.
- Confirm the corners. You can drag any of the four corners. Nothing is applied until the corners are where you want them.
- Correct the perspective. A projective transform maps the quadrilateral you marked onto a rectangle. This is not a rectangular crop: the four corners can be in four completely different positions and the result is still square.
- Even out the lighting. The brightness of the paper is estimated across the page and divided out, which removes most of a shadow, a side light or the shading either side of a crease.
- Finish. Depending on the mode you choose, the page is left in colour, converted to greyscale, or reduced to black and white with a local threshold that keeps faint pencil and pale stamp ink.
Using the camera
Press Start camera . Your browser will ask for permission — that prompt belongs to the browser, not to this page, and this page cannot answer it, dismiss it or change the setting afterwards. If you allow it, the rear camera is requested first, because that is the one with the better sensor on almost every phone.
While the camera is running, the scanner looks for the outline of a page about twice a second and draws it over the preview. If no outline appears, that is real information: move the page onto a surface that contrasts with it, or fill more of the frame.
Torch appears only when the browser reports that your device's camera actually supports it — on most desktop webcams and in some mobile browsers it will not appear at all, and showing a button that does nothing would be worse than showing none.
If permission is refused
The scanner will tell you plainly and offer Import files instead. It will not claim to be able to open your browser's private settings pages, because no website can do that. In most desktop browsers the control is the small icon at the left of the address bar; on a phone it is usually under Settings, then Site settings, then Camera.
Importing photos and PDFs
Press Import files , or drag files onto the drop area. JPG/JPEG, PNG, WebP, GIF, BMP, SVG and AVIF are read when the browser can decode them. TIFF/TIF and HEIC/HEIF can use compatibility decoders. PDFs are rendered page by page, and Word DOCX files are rendered into document pages so the actual page content is visible.
HEIC/HEIF photographs can use a browser-side compatibility decoder when the browser does not natively decode them. If a HEIC file still cannot be read, the scanner names the file and reason. Setting Camera → Formats → Most Compatible on the phone, or sharing the photo as a JPEG, solves it.
An imported PDF page is already flat, so no perspective correction and no automatic crop is applied to it. Cropping something that is already a clean document page is the quickest way to lose part of it.
Import accepts PDF files, Word DOCX documents and browser-decodable image files. DOCX content is rendered into page images so it can use the same crop, perspective, clean-up, reorder and A4 export pipeline as a photographed document.
Corners and perspective
Press Edit on any page to open the corner editor. The four red handles sit on the corners of the paper. Drag them with a finger or a mouse; on a keyboard, tab to a handle and use the arrow keys, holding Shift to move in larger steps. The preview beside the image updates as you move them.
Detect edges runs the detection again. If it cannot find a page it tells you and leaves your corners exactly as they were. Whole image puts the corners back at the four corners of the photograph, which turns the perspective correction off entirely.
| Situation | Result |
|---|---|
| Page photographed from one side, so it looks like a trapezium | Corrected completely — the output is rectangular and the text lines run straight across |
| Page rotated in the frame | Corrected — the page comes out upright |
| Page photographed from very close, so it bulges (lens distortion) | Partly corrected. A projective transform cannot undo barrel distortion; step back and zoom instead |
| Page curved, for example near the spine of a thick book | Not corrected. The curve is not a flat plane, and a homography only maps flat planes |
Clean-up modes compared
| Mode | What it does | Best for |
|---|---|---|
| Original | Nothing at all. The pixels are passed through untouched. | Photographs, artwork, anything where accuracy of tone matters |
| Automatic | Evens out the lighting, then sets the black and white points from the image itself | Most documents |
| Colour document | Evens the lighting on each colour channel separately, so the paper goes white while a red stamp stays red | Certificates, forms with coloured ink, signed and stamped documents |
| Greyscale | As Automatic, but with the colour removed | Printed text where colour adds nothing and only enlarges the file |
| Black and white | A local threshold, so a faint mark in a dark corner survives while blank paper stays white | Fax-style documents, receipts, the smallest possible file |
| Strong clean-up | Full lighting correction, a gentle speckle filter and firmer levels | Badly lit photographs and creased paper |
Shadows, creases and folds
This is worth being precise about, because it is the thing most often over-promised.
A fold in paper is visible for two separate reasons. The first is shading : the crease catches the light differently, so one side of it is darker. That is a slow change in brightness across the page, and it can be measured and divided out. Doing so removes most of the visible line while leaving the text alone, because text is a fast, fine-grained change that plays almost no part in the brightness estimate.
The second reason is physical damage : the fibres of the paper have been broken, or the print itself has cracked along the fold. Nothing in a browser — and nothing in any other scanner app — can restore ink that is no longer on the paper. What is left after the shading is removed is what is genuinely there.
So the honest claim is this: the scanner substantially reduces the visual effect of wrinkles, folds and uneven lighting. It does not remove creases, and it does not repair damaged paper. Any tool that claims otherwise is either painting over your document or guessing at what used to be underneath.
Managing pages
Every captured or imported page appears in the Pages list with its own controls: Edit , rotate left, rotate right, move earlier, move later, and delete. Rotation is in ninety-degree steps and applies to the finished page, not to the original, so it can be undone from the editor at any time.
Scanning a long document works page by page: capture, check, capture the next. There is no fixed limit on the number of pages, but each page is held in your device's memory, so a phone with little free memory will slow down after a few dozen high-resolution pages. If that happens, export what you have and start a new batch.
Exporting
Download PDF builds a single PDF containing every page in the order shown. Each page is stored as a JPEG inside the PDF. You can choose the page size — fitted to the scan at 150 dpi, A4, or A4 — the output resolution, and the JPEG quality. Higher settings mean a larger file; 2200 pixels on the long edge at 82% is a sensible default for text that is going to be read on a screen or printed at ordinary size.
Download JPG images saves one file per page. Browsers treat several downloads in a row with suspicion, so the first time you do this your browser may ask whether to allow multiple files.
The PDF is assembled in the page itself. It is not sent to a server and then sent back, which is why it works with no network connection once the page has loaded.
Getting a better scan
- Contrast with the background. A white page on a white table is the hardest case for any edge detection. A dark desk, a dark folder or even a dark jumper underneath the page makes a large difference.
- Fill the frame. The more of the frame the page occupies, the more pixels each letter gets. Getting closer beats increasing the export resolution.
- Light from the side, not from behind you. A lamp directly behind you throws your own shadow across the page. Light coming across the page is easier to correct.
- Flatten the paper first. Thirty seconds under a heavy book does more for a creased page than any software.
- Hold still. Motion blur cannot be undone. If the preview looks soft, it will be soft in the scan.
- Check the preview before you apply. The result panel in the editor shows exactly what will be exported.
Privacy
Photographs taken or imported here stay on your device. They are held in your browser's memory for as long as the page is open and are gone when you close or reload it. Nothing is uploaded, and there is no account, no sign-in and no storage of your documents on any Docelora server.
The core PDF reader is served from Docelora itself, under
/scanner/vendor/
, and carries its own licence file. Optional compatibility libraries for Word, TIFF or HEIC input are loaded only when that format is selected; the selected file content is processed in the browser.
The browser's camera permission is between you and your browser. Revoking it stops this page using the camera immediately.
What this cannot do
- It cannot read the text on the page. There is no optical character recognition here, so the PDF contains pictures of your pages and is not searchable.
- It cannot flatten a curved page, such as the inner margin of a thick bound book.
- It cannot repair physical damage, restore faded ink or remove a crease from the paper itself.
- It cannot correct heavy lens distortion from a very close, very wide-angle shot.
- It cannot undo motion blur or recover detail from an out-of-focus photograph.
- It cannot open a password-protected PDF.
- HEIC/HEIF, TIFF/TIF and Word DOCX files use browser-side compatibility converters when a browser does not natively provide the needed decoder.
- It cannot use the camera on a page served without https, because browsers do not permit it.
Frequently asked questions
Is the Docelora document scanner free?
Yes. There is no charge, no account and no page limit. The tool runs entirely in your browser, so there is no per-scan cost to pass on.
Are my documents uploaded to a server?
No. Capture, correction, clean-up and PDF assembly all happen in your browser on your own device. The image data never leaves it, which you can confirm by disconnecting from the network after the page has loaded — the scanner keeps working.
Why will my camera not start?
There are four common reasons: permission was refused, the page is not on an https address, another program already has the camera open, or the browser does not offer camera access at all. The scanner names which one it hit rather than showing a general failure message.
Can this page change my camera permission for me?
No, and any site that claims it can is misleading you. Camera permission lives in the browser's own settings, which a web page cannot open or alter. The scanner tells you where the setting usually is and offers file import as an alternative.
Does it work on an iPhone or an Android phone?
Yes, in a current mobile browser. On iOS the camera is available in Safari and in browsers built on it. Importing photos from the camera roll works on both platforms; HEIC/HEIF can use the compatibility decoder when needed.
What is the difference between this and just taking a photo?
A photograph keeps the angle, the perspective, the background and the uneven lighting. The scanner removes the background, maps the page onto a true rectangle, evens out the lighting and gives you a document-shaped file. The difference is most obvious when you print the result.
Can I scan more than one page into a single PDF?
Yes. Capture or import as many pages as you need; they appear in the Pages list and can be reordered, rotated or deleted. Download PDF puts them all into one file in the order shown.
How do I change the order of pages?
Use the left and right arrow buttons on each page card. The number shown on the card updates immediately, and the export always follows the order on screen.
Can I remove a page I scanned by mistake?
Yes. Press Delete on that page's card. Remove all pages clears the whole batch. Neither affects the original files on your device.
What does "perspective correction" actually mean?
It is a projective transform: the four corners you marked are mapped onto the four corners of a rectangle, and every pixel in between is resampled to match. It is genuinely different from cropping, which can only cut a rectangle out of a rectangle and cannot straighten a page photographed from an angle.
The edges were detected in the wrong place. What do I do?
Open the page and drag the corners where they belong. Detection is a starting point, never the final word — and when it cannot find a page at all it says so and leaves your corners untouched instead of applying a guess.
Can it remove creases and wrinkles from a folded document?
It substantially reduces how visible they are, by measuring the lighting across the page and dividing it out, which removes most of the shading along a crease. It cannot remove the crease itself or repair print that has cracked along a fold. Nothing can, and claims to the contrary usually mean the tool is painting over your document.
Will the clean-up delete faint handwriting, a signature or a stamp?
No. The clean-up modes change the appearance of the paper and the lighting, not the marks on it. The black-and-white mode in particular uses a local threshold specifically so that pale pencil and light stamp ink in a shadowed corner survive, where a single global threshold would lose them.
Can I get the original image back after processing?
Yes, always. The captured or imported image is kept untouched, and every preview is rebuilt from it. Reset to original in the editor restores it no matter how many adjustments you have made.
Which file formats can I import?
JPG, PNG, WebP and GIF images, and PDF files. PDF pages are rendered so you see the real page, not just a file name. HEIC depends on your browser: Safari reads it, most others do not.
Why is my PDF so large?
Because each page is a photograph, and photographs are large. Reduce the output resolution, lower the JPEG quality, or use the greyscale or black-and-white mode — black and white typically produces the smallest files by a wide margin.
Can I search the text in the exported PDF?
Not from this tool. There is no optical character recognition here, so the PDF contains images of your pages. If you need searchable text you will need an OCR step elsewhere.
Does it work offline?
For the core scanner and locally served PDF reader, yes after the page and local assets have loaded. Optional compatibility converters for DOCX, TIFF or HEIC may need to be loaded the first time they are used.
Is there a limit on the number of pages or the file size?
There is no fixed limit in the tool. The practical limit is your device's memory, since every page is held there at full resolution. On a phone, expect to notice slowing after a few dozen large pages; export the batch and start another.
How do I get the sharpest possible result?
Fill the frame with the page, light it from the side rather than from behind you, hold the device still, flatten the paper first, and export at the largest resolution you need. Getting physically closer to the page improves detail far more than raising the export resolution does.
Can I import a Word document?
Yes. Microsoft Word DOCX files are converted in the browser into page images, then treated like imported document pages for editing and A4 export.
Can I mix Word, PDF and image files in one scan batch?
Yes. DOCX pages, PDF pages and supported images can be added to the same Pages list, then reordered and exported together.
Will a Word document keep its original editing format?
No. The scanner is an image-based document workflow. DOCX content is rendered as page images for scanning, correction and PDF or JPG export; it is not exported as an editable Word file.
Can I export every page as A4?
Yes. The scanner uses A4 as its fixed PDF and JPG page format. Portrait or landscape orientation follows the final page orientation, while the content is fitted proportionally to the A4 page.
Which image formats can I import?
Common browser-decodable formats such as JPG/JPEG, PNG, WebP, GIF, BMP, SVG and AVIF are supported. TIFF/TIF and HEIC/HEIF can use compatibility decoders when required by the browser.
Are DOCX files uploaded to a server for conversion?
No. The Word conversion is designed to happen in the browser. The page loads the required conversion libraries when needed, while the selected document remains part of the local browser workflow.
Related tools
Last reviewed: 20 September 2026.