Page 1 of 1

OCR: colored images OCR'd to grey

Posted: Mon Oct 10, 2016 5:13 pm
by claude vidal
Using PDFXchange Editor 6.318.1, when performing Documents\OCR pages... on a PDF document containing images defined as Color Space: DeviceRGB, the output produces those images as DeviceGray. (BTW, shouldn't that be 'grey'?)

I've tried several files, all with the same result, including the PDF-XChange Editor/Editor Plus V6 USER MANUAL.

I couldn't find any setup parameter for the OCR plugin. Is this a bug or am I missing?

Re: OCR: colored images OCR'd to grey

Posted: Tue Oct 11, 2016 7:11 am
by Will - Tracker Supp
Hi Claude,

Thanks for the post - I'm not able to reproduce this here, even with the manual you mentioned. Can you upload a small sample from your PC? Also, please export your settings under Edit --> Export Settings and forward them to us here.
(BTW, shouldn't that be 'grey'?)
Grey and gray are both acceptable. 'Gray' is more popular in the U.S., whereas 'grey' is more popular in Europe. We tend to stick to the U.S. spellings, as do most software manufacturers.

Cheers,

Re: OCR: colored images OCR'd to grey

Posted: Tue Oct 11, 2016 1:38 pm
by claude vidal
I've narrowed down the issue I'm having.

If I set 'output type' to 'Preserve original content and Add text layer', then image color is retained (color space: DeviceRGB).

If I set 'output type' to 'Create new searchable PDF', then image color is lost (color space: DeviceGray).

In both tests:
- Accuracy = high
- Output quality = auto
- Auto deskew = off

Exported settings and sample PDF attached.

Re: OCR: colored images OCR'd to grey

Posted: Tue Oct 11, 2016 2:01 pm
by Will - Tracker Supp
Thanks for that Claude - I was able to reproduce it without importing your settings and does appear specific to the Create New Searchable PDF featire. I'll forward this along to the Dev. Team for investigation.

Cheers,

Re: OCR: colored images OCR'd to grey

Posted: Wed Oct 12, 2016 7:26 am
by Will - Tracker Supp
Hi Claude,

Just an update: I've created an internal development ticket for this issue (RT-3682). We'll fix this as soon as we're able.

Apologies for the inconvenience!