OCR Character Display

[Spyral]

I just tested the new OCR and I’m a bit confused with the setting for the “Character display”. What is the difference of the three settings? In Live OCR I don’t see a difference.

 

[mikelove]

It makes more of a difference in still images; 'raw' means that we take the raw character rects that come out of the recognizer even if some of the sizes are a little bigger or smaller or otherwise weird (a problem with newer ML-based OCR engines) while 'refined' means we arrange them into a nice neat same-sized evenly-spaced line even if it doesn't quite match up perfectly with the original text.

 

[Spyral]

Is it possible to have an overlay similar to the old system where only the characters had one? Sometimes it is important that I can make sure the character was recognized correctly. This is rather difficult at the moment.