The Pre-OCR Checklist: Five Minutes of Preparation Beats an Hour of Proofreading

Quick answer: OCR quality is decided before you press the button. Five checks: resolution 300 DPI+, page straight, edges cropped, language confirmed, worst page piloted first. This routine cuts post-OCR proofreading dramatically. Open Image OCR — everything runs locally in your browser, and files never leave your device.

Steps

1. Resolution: 300 DPI for scans; phone photos must show sharp glyph edges when zoomed.
2. Straighten pages beyond 2° of tilt — skew is the top cause of dropped lines.
3. Crop black borders, shadows, and desk background.
4. Re-confirm the language option matches the document (including simplified/traditional mixing).
5. Pilot the worst-looking page first — if it passes, the batch will.

Important notes

• Native PDFs skip OCR entirely — the text is already there.
• Flag stamp and handwriting regions in advance; they need manual handling.
• Table pages: plain-text output flattens structure — process separately and plan the rebuild.

Common questions

Can I just run the whole batch? You can — but a wrong language setting discovered late voids the entire run; the single-page pilot costs almost nothing.

Is preparation really worth it? For archives, absolutely: 3 minutes of proofreading saved per page is 5 hours per 100 pages.