How to Extract Text From Images and Screenshots With OCR
Sooner or later you need the words inside a picture: a quote in a screenshot, a paragraph on a printed page, a receipt, a slide photographed from the back of a room. Retyping is slow and error-prone. Optical character recognition (OCR) reads the picture and gives you editable text in seconds.
How OCR works
OCR analyses an image for shapes that look like letters, groups them into words and lines, and outputs them as text. Modern engines compare each shape with trained models of characters and use language knowledge to choose the most likely word, which is why a smudged letter inside a common word is often still read correctly. The tool runs the recognition engine in your browser, so your image is not uploaded to a server.
What makes OCR accurate
| Factor | Best practice |
|---|---|
| Resolution | Sharp text with letters roughly 20–30 pixels tall or more |
| Contrast | Dark text on a plain light background |
| Angle | Straight and level; skewed or curved pages lower accuracy |
| Lighting | Even light with no shadows or glare |
| Fonts | Plain printed fonts read best; handwriting and decorative fonts are less reliable |
How to extract text step by step
- Choose a photo, screenshot or scan, or drag it into the upload area of Image to Text.
- Wait while the engine reads the image. The first run downloads the engine, so it takes longer; later scans in the same browser start faster.
- Review the result and fix mistakes.
- Copy the text or download it as a .txt file.
Preparing a better image
- Crop away borders and background so the engine focuses on the text.
- Take photos straight on, from directly above the page, rather than at an angle.
- Use screenshots instead of photos of a screen whenever you can.
- Avoid shrinking an image too far. If it is huge, reduce it moderately with the image compressor, but never below the sharpness you need to read it.
- For low-contrast scans, raise contrast in any image editor before running OCR.
Typical errors and how to fix them
| Mistake | Why it happens |
|---|---|
| 0 and O, 1 and l and I mixed up | Similar shapes; check codes, serial numbers and passwords by eye |
| Broken words at line ends | Hyphenation in the original; rejoin them by hand |
| Columns merged into one stream | The engine reads straight across; copy each column separately |
| Missing accents or symbols | Small marks are lost in low-resolution images |
Always proofread OCR output. Numbers deserve special care: a wrong digit in an amount or a phone number is easy to miss.
Privacy and sensitive documents
Photos of IDs, contracts and bank statements contain private data. Because recognition runs locally in your browser, the image stays on your device, which is the safer approach for sensitive material. Still, avoid shared computers, and delete extracted text you no longer need.
Key terms: OCR, optical character recognition, image to text, screenshot to text, scan to text, text extraction, recognition accuracy, resolution, contrast.
Frequently asked questions
Is my image uploaded to a server?
No. Text recognition runs inside your browser, so your image stays on your device.
Why is the first scan slower?
The first use downloads the recognition engine, which your browser then caches. Later scans start faster.
What images work best?
Clear, well-lit, high-contrast printed text at a straight angle. Handwriting and stylised fonts are less reliable.
Is there a limit on how many images I can process?
There is no daily limit; images are processed one at a time.
Can OCR read handwriting?
Only poorly. Neat block capitals have a better chance, but expect to correct a lot.
Have a photo or screenshot full of text?
Try Image to Text (OCR) →Put this guide into practice
Image to Text (OCR) — Extract text from a photo or document, free.
Image Compressor & Resizer — Compress, resize and convert images — free, no upload.