PDF guides
What Affects OCR Accuracy in a Scanned PDF
OCR accuracy is not random — a handful of concrete factors determine whether a scan recognizes cleanly or produces garbled text. Here is what actually matters, and the real limits of this tool’s engine.
Ready to try the tool this guide describes?
Resolution
Text recognition accuracy drops sharply below roughly 200 DPI. This tool rasterizes each page at a fixed scale chosen specifically to stay above that floor, regardless of how the source PDF itself was scanned — but a source scan that was already captured at very low resolution can’t be recovered by rasterizing it larger.
Skew and page orientation
A page that is rotated (as recorded in the PDF itself) is handled automatically — the tool reads the page in its displayed orientation before recognizing it. A page that is skewed at a small, non-90-degree angle (a crooked scan, not a rotated one) is harder for any OCR engine to read cleanly and is not corrected by this tool.
Contrast and scan quality
Faint text, heavy background texture, or low contrast between text and page all reduce recognition confidence. Low-confidence words are filtered out before being embedded, so a poor scan typically results in fewer words recognized rather than incorrect ones — see the companion guide on how the text layer itself works for why that filtering exists.
Language
This version recognizes English only. Text in another language will not be recognized correctly, and there is no language selector to change that.
Page-count and file-size limits, and why
OCR PDF accepts up to 25 pages and 50 MB per file. Recognizing a page is the most computationally expensive step of any PDF tool on this site — rasterizing at OCR-quality resolution and running full-page recognition — so the limit exists to keep processing time and browser memory use reasonable on typical hardware.
What Affects OCR Accuracy FAQ
- Why did some words get left out of the searchable text?
- Words the OCR engine wasn’t confident about are filtered out before being embedded, rather than adding likely-wrong text to the file.
- Does a bigger file mean better accuracy?
- Not necessarily — what matters is the effective resolution and clarity of the scanned content, not the PDF’s file size.
- Can I improve accuracy by rotating a skewed page first?
- Rotating a page (90-degree steps) is handled automatically. A small-angle skew from a crooked scan is a different problem this tool does not straighten.
- Will non-English text be recognized at all?
- No — this version only recognizes English text.
- Why is there a 25-page limit instead of a higher one like other PDF tools here?
- Because OCR is the most resource-intensive per-page operation this site performs — rasterizing and recognizing every page in a browser tab — so the limit is deliberately lower than tools that only rearrange or rasterize pages for preview.
Related guides
- How to OCR a Scanned PDFTurn a scanned or image-based PDF into a searchable, selectable document using Looty Tools' browser-based OCR, with the exact page and file-size limits and what happens to pages that already have text.
- How OCR Text Layers WorkThe technical mechanism behind an OCR’d PDF: why the page still looks like a photograph while the text underneath becomes searchable and selectable, and what an invisible text layer actually means.
