Optimal Configuration for Best OCR Output Using Scribe.js

[zakariamehbi]

Hello,

I noticed that the current documentation lacks details in this area, so I’d like to directly ask the team for those seeking the best OCR results.

What would be the ideal configuration for Scribe.js to achieve the highest quality OCR output on a PDF, without considering performance or execution time constraints?

Thank you in advance for your help!

[Balearica]

The default settings already optimize for quality.

1. The default recognition mode is already quality rather than speed.
2. For .pdf files, the extractText function extracts existing text content rather than running OCR for text-native files, however ignores existing text content (and runs OCR instead) for image-native files with an existing OCR layer.
   1. This results in the highest quality result for files where Scribe.js is more accurate than whatever program created the original layer, which is generally true.