Scanned documents are a particular kind of frustrating. The text is right there on the screen — you can read it — but you can't select it, search within it, or paste any of it into another document without typing it out by hand. This is the gap OCR fills: it reads the image and hands you back actual, editable text.
Here's how to do it reliably with Portimg's OCR tool, and what to watch for to get clean output on the first pass.
What you're actually doing
When you run a scanned document through OCR, the tool analyses the pixel content of the image and identifies character shapes, spacing, and line structure. The output is plain text — the same as if you'd typed the document yourself. From there you can paste it into Word, a Google Doc, an email, a spreadsheet, or anywhere else text goes. You can search it, edit it, reformat it, and run it through spell check.
Printed text from clean scans converts with high accuracy — typically 95% or better on a well-prepared image. That means a page of 400 words might have fewer than 20 characters that need correcting, which is usually faster than retyping even a short paragraph. Handwriting is harder and more variable; clear printed handwriting often comes through well, while cursive or informal script will need more review.
Running the conversion
Go to portimg.com/image-to-text-ocr and upload your file. JPG, PNG, and WebP are all supported. The tool processes the image and returns the extracted text in a results panel within seconds. Copy it directly or download it as a text file.
No account is needed. Files are deleted from the server immediately after processing, which matters when you're dealing with documents that contain personal or sensitive information — contracts, medical records, financial statements.
Preparing your scan for better results
The most common reason OCR output needs heavy correction isn't the tool — it's the image. A few preparation steps make a significant difference:
Resolution first. 300 DPI is the minimum for reliable results on standard-sized text. Below that, fine details in letterforms start to disappear, and visually similar characters — 'l' and '1', 'O' and '0', 'B' and '8' — become hard to distinguish. If you're scanning with a dedicated scanner, set it to at least 300 DPI. If you're using a phone, shoot from close enough that the text fills most of the frame, and make sure the image is in focus before you capture it.
Even lighting, no shadows. A shadow crossing part of the page drops the contrast in that area and increases errors there. Flat light from two sides, or diffuse natural light from a window, keeps the whole page evenly lit. Direct flash creates a bright spot in the centre and shadows at the edges — not ideal.
Straight alignment. Text running at an angle is harder to segment into lines. A few degrees of skew is usually fine, but a page photographed at a sharp angle may produce garbled output. If your scan is visibly tilted, rotate it to horizontal before uploading — most phone photo apps and basic image editors have a straighten tool.
Crop to the text. If the image includes a wide border, the surface the document is resting on, or anything else that isn't the text you want, crop it before uploading. It reduces noise in the input and occasionally improves accuracy near the edges of the page.
Documents that come as PDFs
Scanned PDFs are images embedded in a PDF container — they look like text but aren't selectable, for the same reason a photograph of a document isn't. To run OCR on them, convert the pages to images first using Portimg's PDF to image tool, then upload the resulting images to the OCR tool. For multi-page documents, process each page and combine the text outputs in order.
If you need to do any image preparation — adjusting contrast, fixing exposure on a dark scan, rotating a tilted page — Portimg's image editor handles that in the browser before you run the OCR step.
When it's worth doing
OCR makes most sense when the volume of text you'd otherwise retype is significant, or when you need the output to be searchable rather than just readable. A full page of a printed report, a multi-page contract, an archive of old letters — these are clear cases. For a two-line caption or a short label, typing is faster and not worth the round-trip.
The best way to calibrate expectations is to run one of your actual documents through and see what comes back. A good scan of printed text will usually surprise you with how little correction it needs.