8 Tips for Accurate Text Extraction from Photos (OCR)

Updated: 2026-11-10

OCR (optical character recognition) turns pictures of text into editable text. Good results depend far more on the photo than on the software. Follow these eight tips with Image to Text.

1. Light it evenly

Daylight from a window or a bright room is best. Avoid shadows from your hand or phone and glare from glossy paper.

2. Shoot straight on

Hold the phone parallel to the page. Tilted photos make lines curve, and OCR reads curved lines poorly.

3. Fill the frame with text

Crop away the table, fingers and background. The text should cover most of the image.

4. Keep it sharp

Tap to focus, hold still, and check the photo by zooming in. If you cannot read small letters when zoomed, OCR cannot either.

5. Choose the correct language

Each language has its own model. Arabic needs the Arabic model; French accents read best with French. For mixed documents choose a combination such as English + Arabic.

6. Use screenshots when you can

A screenshot of a web page or PDF gives perfectly clean letters and nearly perfect results.

7. Split complex layouts

Tables, columns and pages full of pictures can mix up the reading order. Crop and process one column or section at a time.

8. Always proofread numbers

OCR can confuse 0 and O, 1 and l, 5 and S. Check amounts, dates, names and ID numbers before using them.

Tips for Arabic text

• Printed Arabic in common fonts (Naskh, simple Kufi) reads well; decorative calligraphy does not.
• Heavy diacritics (tashkeel) can reduce accuracy; plain text works best.
• The result box switches to right-to-left automatically when you choose Arabic.

From a PDF?

Convert the pages to images first with PDF to JPG at 150 or 300 dpi, then extract the text.

Guides

Tools are provided as-is. Keep a copy of important originals. See our disclaimer.