CamScanner alternatives
Turn a photo of a document into a clean, flat PDF scan.
5 of its 6 core features can be rebuilt with AI. Here is one, working, below.
- They makeRevenue estimate on Sensor Tower
- $8M a month
- You pay
- $70 a year
- AI can rebuild
- 5 of 6 core features
Try it
Visual demo only. Copy the prompt below to get the app.
Build it yourself
Real files on your machine, a real database, and it is yours. This one carries the interface as well — the layout, the colours and the shape of the real app, read off its own screens.
Each opens with the prompt loaded and waiting — nothing is sent until you press enter. It is copied at the same time, in case the app is not installed.
Build me a local document scanner I own, to replace CamScanner. - A CLI plus a small local page: Python, FastAPI and OpenCV, data in a SQLite file next to it. - Find the page edges with OpenCV: greyscale, blur, Canny, largest four-sided contour, with four draggable corners for when it guesses wrong. - Warp that quadrilateral front-on with getPerspectiveTransform, then crop and rotate. - Filters as named pipelines: colour enhance, greyscale, and adaptive thresholding for black and white, so a dim room does not come out solid black. - Model it properly: documents, and a page row holding the original's path, its four corners, its filter and its order. Every scan re-renders from the untouched original. - OCR each page with Tesseract locally, store the text, and index it with FTS5 so every scan is searchable. Good on a flat page, poor on a curled receipt. - Export to PDF with img2pdf, the recognised text as an invisible layer. - One command to import a folder of photos as a document, one to back up the database and originals. - Binds to localhost only. No accounts, no telemetry, no network calls at all. - Skip: cloud sync, share links and team folders. That is storage somebody pays for every month. - Include a README with setup, installing Tesseract, and where the scans live. ## Look and feel CamScanner's own, and the first thing to get right is that it is teal, not green. One teal, and it carries the camera button, the tool icons and anything active. Everything else is black on white with grey for meta. It also does not open on a camera. It opens on a file manager with a launcher above it: - A grey search pill across the top with a small filter control at its right end. - A grid of tools, four across, each a rounded-square icon in a circle of pale ground with its name underneath — PDF Tools, ID Cards, Extract Text, Import Images, Import Files, All. A small red NEW flag sits on the corner of anything recent. - A **Recents** heading with a quiet "View All ›" opposite it, then documents as rows: a thumbnail of the page on the left, the file name, and the date and page count under it in grey. The bottom bar has five slots — Home, Files, the camera, Tools, Me — and the camera is a filled teal circle raised proud of the bar, not an icon sitting inside it. The scan screen is the other half: the page preview takes almost everything, corner handles on the detected edges in teal, and the filter choices as a row of small labelled thumbnails along the bottom. Generous spacing, clear hierarchy, big tap targets, phone-readable type, system font. Keep the text dark enough to read at a glance: the softness of a palette belongs to the shapes and the fills, never to the words. Nothing I need to read — dates, numbers, labels — in pale grey on white. Keep it all on my device: no accounts, no sign-in, nothing sent anywhere. Save my data so it is there next time I open it.
Did this prompt work for you?
Thanks — that tells us whether to rewrite it.
More subscriptions die every week.
New apps and the prompts that replace them, as they land. One email, no spam.
Free forever · unsubscribe in one click · your address is never shared
Feature by feature
Every feature, whether AI can rebuild it, and what you would be signing up to keep running if you did. The paid app does that last column for you, and that is most of what the money buys.
| Feature | Core | AI can rebuild | You'd maintain |
|---|---|---|---|
| Photograph or choose a pageThe camera has been available to any web page for years. The only real work is shrinking the photo, because a modern phone picture is far bigger than a scan needs. | Yes | Yes | Nothing |
| Find the edges of the page automaticallyAutomatic edge detection on a cluttered desk is genuinely fiddly. Four handles you can drag get you the same result in two seconds. | Yes | Partly | You drag the corners yourself |
| Flatten the skew into a proper rectangleThis is the trick that makes a photo look like a scan, and it is one well-known piece of geometry. It runs on the page in front of you. | Yes | Yes | Nothing |
| Scan filters: black and white, greyscale, colour enhance | Yes | Yes | Nothing |
| Multi-page documents you can reorder and delete from | Yes | Yes | Nothing |
| Export as a PDFA browser can put images into a PDF, and printing to PDF is built into every phone as a fallback. | Yes | Yes | Nothing |
| OCR: pull the text out of the scanReading text out of a photograph is exactly what the vision model does. | No | Yes | Needs a paid service to be worth reading |
| Cloud sync, sharing links and team foldersSomeone has to pay for the storage and the link that keeps working next year. | No | No | Files stay on the device that made them |
| Signatures and annotations on a page | No | Yes | Nothing |
What you give up
- text recognition good enough to search and copy from
- documents that appear on your laptop as well
- sharing a scan as a link that still works next year
- somebody else keeping years of paperwork safe
Why people still pay
Flattening a photo of a page is one piece of school geometry, and it runs perfectly well in a browser. The subscription is really buying two things that cost money every month: text recognition accurate enough to trust, and a copy of your paperwork on somebody else's computers.
Or use one somebody already built
Free and open source. No prompt, no building, no subscription.
- Open Note Scanner
- Edge detection, perspective correction and PDF export, free and offline. Rougher than the paid apps, and it never asks for an account. (android)
- Stirling PDF
- Everything you would do to the PDF afterwards: merge, split, rotate, compress and OCR. Runs on your own machine. (web, windows, macos, linux)
- Tesseract OCR
- The free text recognition engine most tools are quietly using. Not an app, but it is the part of CamScanner people think is magic. (windows, macos, linux)
Other apps people are freeing
Questions
- Can AI really replace CamScanner?
- 5 of its 6 core features, yes. The demo above is what one prompt produced. What it cannot do is on this page too.
- What does CamScanner cost?
- About $70 a year on the CamScanner Premium plan. Scanning and exporting works free, with a watermark on the page and limits on OCR and cloud storage.
- Is there a free alternative to CamScanner?
- Yes: Open Note Scanner, Stirling PDF, Tesseract OCR. They are listed below, and they are somebody else's finished work rather than yours.