How to Convert an Image to Text With Free OCR
By Ali Hamza · Published · 4 min read
On this page
To extract text from an image, open an OCR tool such as ToolMany's Image to Text (OCR), upload a clear photo or screenshot, select the language of the text, and run recognition. Then copy the result and proofread it. Accuracy depends mostly on the image: sharp, evenly lit, straight-on photos of printed text come out nearly perfect, while blurry or angled shots don't.
This guide explains how to give OCR the best possible input and what to check afterwards.
What OCR is (and when you don't need it)
OCR, short for optical character recognition, looks at the shapes in an image and works out which letters they are. ToolMany runs the open-source Tesseract engine inside your browser, so the image isn't uploaded.
You only need OCR when the text is part of a picture: a photo of a page, a screenshot, a scanned document or an image of a receipt. If you have a digital PDF made from Word or a website, the text is already there. Use PDF to Text to pull it out exactly, with no recognition errors. Quick test: if you can select a word in the PDF with your cursor, it's a digital PDF.
For a scanned PDF, convert the page to an image first with PDF to JPG & PNG (PNG at 2× or 3×), then run OCR on that image.
Step-by-step: image to text with ToolMany
- Open Image to Text (OCR).
- Choose your image: a photo, screenshot or scan in JPG, PNG or WebP.
- Pick the language of the text. The tool supports English, Spanish, French, German and Japanese. The first time you use a language, its recognition data is downloaded to your browser.
- Start recognition and wait for it to finish. Larger images take longer, especially on phones.
- Copy the extracted text and proofread it against the original.
Seven ways to improve OCR accuracy
1. Get the lighting even
Shadows across a page are the most common cause of errors. Photograph documents near a window in daylight, and avoid leaning over the page so your head and phone don't cast a shadow. Turn the flash off; it creates a bright hotspot and glare on glossy paper.
2. Shoot straight on
Hold the phone parallel to the page, directly above it. Photographing at an angle makes letters at the far edge smaller and skewed. Many phone camera apps show a level indicator; use it.
3. Fill the frame with text
Crop away the table, your fingers and the background. Recognition works better when the text is large in the image. As a rule of thumb, normal body text should be at least 20–30 pixels tall in the image. Crop and straighten the photo with the edit button in your phone's Photos app, or with Windows Photos or macOS Preview, before running OCR.
4. Use enough resolution, but not too much
For scans, 300 DPI is the standard sweet spot for OCR. For phone photos, the camera's normal resolution is plenty. Don't upscale a small, blurry image, because enlarging it doesn't add detail.
5. Increase contrast
Faded receipts and grey photocopies are hard to read. Use the Photo Filter Editor to raise contrast and slightly lower brightness so text becomes dark and the background light.
6. Rotate the image upright
OCR expects horizontal lines of text. Rotate the image so text reads left to right before recognising it.
7. Choose the right language
The language setting tells the engine which letters and words to expect. Recognising a German document as English, for example, garbles accented characters like ä, ö, ü and ß.
What OCR struggles with
| Content | What to expect |
|---|---|
| Clean printed text (books, letters, screenshots) | Very accurate |
| Receipts and thermal paper | Usually good on fresh receipts; faded ones need contrast fixes |
| Tables and multi-column layouts | Text comes out, but columns may be merged or reordered |
| Handwriting | Generally poor; neat block capitals work best |
| Stylised fonts, logos, text on photos | Hit and miss |
| Very small print, such as legal footnotes | Needs a high-resolution, close-up image |
Proofreading: what to look for
Even 98% accuracy means a couple of mistakes per 100 characters, so proofread anything important. Common mix-ups:
- 0 and O, 1, l and I, 5 and S, 8 and B. Check every number in invoices, IDs and phone numbers.
- "rn" read as "m" and vice versa
- Missing spaces, or line breaks in the middle of sentences
- Punctuation dropped or doubled
Paste the text into the Word Counter if you need to check length or fix capitalisation quickly.
Frequently asked questions
Is the OCR free and private?
Yes. It's free with no sign-up, and recognition runs inside your browser using Tesseract. Your image isn't uploaded; only the language data is downloaded the first time you use a language.
Can OCR read handwriting?
Tesseract is designed for printed text, so results on handwriting are usually poor. Neat block capitals have the best chance.
Why is my OCR result full of strange characters?
The image is probably blurry, low-contrast, angled or set to the wrong language. Retake the photo straight on in good light, crop to the text, and check the language setting.
Can I convert a scanned PDF to text?
Yes, in two steps: turn each page into a PNG with PDF to JPG & PNG at 2× or 3×, then run each image through the OCR tool.
Does OCR keep the formatting?
No. OCR returns plain text. Bold, fonts, columns and tables aren't preserved, so you'll need to reformat the text after pasting it into a document.
