A phone camera is not a scanner. A flatbed scanner presses the paper flat, lights it evenly from underneath, and photographs it from a fixed distance at a right angle. A phone does none of that. It photographs a curved sheet, from an angle, under whatever light happens to be in the room, with your own shadow falling across it.
That gap is the reason most phone "scans" look like photographs of paper rather than documents. The good news is that almost all of the difference comes down to four things you control — light, angle, framing, and the correction you apply afterwards. Get those right and a phone photo becomes indistinguishable from a scanner output for everyday use: sending a signed letter, filing a bill, attaching a certificate to an application form.
This guide covers how to take the photo, and then how to correct it in the browser using the scanner built into Pennary. Nothing is uploaded — the camera feed and the correction both run on your own device.
Bad lighting cannot be fixed later. If part of the page is in shadow and part is blown out white, no amount of processing will recover even, readable text — the information simply is not in the photograph.
Where you put the paper matters more than people expect, and for a reason that is not obvious: automatic edge detection has to find the boundary between the page and whatever is behind it. If the paper is white and the desk is also white, there is no boundary to find. The software is not being stupid — the edge genuinely is not visible, and often you cannot see it clearly yourself either.
Put the page on something darker than the page. A wooden desk, a dark folder, a closed laptop lid, a book cover. Any of these gives a clean, high-contrast edge, and detection that failed on a white table will usually succeed immediately.
Two related traps:
Hold the phone parallel to the page, directly above its centre. Every degree of tilt stretches one end of the document relative to the other. This can be corrected afterwards, but correction always costs some sharpness, so it is better not to introduce the distortion in the first place.
Frame the page so that all four corners are visible with a small margin of background around them. This margin is not cosmetic. Perspective correction works by finding the four corners of the sheet and mapping them to the four corners of a rectangle. If a corner is outside the photograph, it cannot be found, and no software — Pennary's or anyone else's — can invent it. A cut-off corner is a permanent defect in that capture.
Pennary shows a dashed guide rectangle in the live camera view for exactly this reason. Fit the whole page inside the dashes and leave the margin.
Two more habits worth building:
Once the photo is taken, Pennary shows four draggable corner points on the detected page boundary. If detection found the page, the points will already be roughly in place and you only need to nudge them. If it did not — white-on-white, a torn edge, a heavily stamped old document — drag each corner onto the actual corner of the paper yourself. Manual placement always works, and it takes about five seconds.
Confirm, and the page is warped flat: the skewed quadrilateral becomes a proper rectangle, and the background is discarded. You then get a second, optional trim step for cropping margins.
| Mode | What it does | Use it for |
|---|---|---|
| Off | No processing. The photo as captured. | Photographs, colour artwork, anything where accuracy of tone matters more than legibility. |
| Auto | Corrects uneven lighting, lifts shadows, keeps colour. The default. | Almost everything — letters, forms, bills, documents with coloured seals, stamps or pen signatures. |
| Grey | As Auto, converted to greyscale. | Text documents where colour adds nothing but file size. |
| B&W | Pure black and white, no intermediate tones. | Clean typed text on white paper. Produces the smallest files. Avoid it for anything with a photo, a light pencil mark or a faint stamp — these disappear entirely. |
Auto is the right answer far more often than people assume. The reflex to reach for B&W usually comes from wanting a smaller file, but B&W discards information permanently. If the document has any signature, seal or handwriting on it, stay on Auto or Grey.
For a document of several pages, capture them in one session rather than one at a time. Tap ＋Page after each capture; up to 25 pages can be queued. Each page appears as a thumbnail you can tap to re-edit or delete before you finish.
There is a practical reason to do it this way beyond convenience. A stack of pages shot in one sitting is shot on the same desk, in the same light, at roughly the same distance. Pennary uses the corrections you made on the first page to seed the corner positions and enhance mode for the pages that follow, so page 12 needs far less adjustment than page 1 did. Closing the scanner and starting again discards that, because the next session might be a different desk entirely.
When you are done: Save downloads a multi-page PDF, Share passes it straight to your phone's share sheet for WhatsApp, email or Drive, and Done hands the pages into the main editor if you want to merge them with an existing file.
| Symptom | Cause | Fix |
|---|---|---|
| Edges not detected, no page found | Paper and background are the same brightness | Move the page onto a darker surface, or place the four corners manually |
| Text readable but grey and washed out | Enhance left Off | Switch to Auto |
| Signature or stamp vanished | B&W mode discarded the mid-tones | Use Auto or Grey for anything hand-marked |
| Page corrected but the printing on it is still slanted | The original was photocopied crooked | Nothing at capture stage can fix this — the paper is straight, the ink on it is not |
| Blurry despite good light | Focus locked on the desk, or hand movement | Tap the text to focus; brace elbows on the table |
| Corner missing after correction | That corner was outside the frame | Retake with a margin around all four corners |
| File far too large to email | Colour scans at high resolution | Use Grey, or compress afterwards |
Everything described here happens inside your browser. The camera stream, the perspective correction and the PDF assembly all run on your device using standard web technology. No image is transmitted to Pennary or to anyone else, which is worth knowing if the documents you scan are identity papers, medical records, financial statements or official correspondence.