Support multiple cores when OCRing

I would really, really like it if DTPO could support as many cores as I have when OCRing. I just got a 27" iMac with a quad-core i7 (with hyperthreading) and it’s a shame that most of the CPU is idle when OCRing. I’ve found myself doing a LOT of OCRing and have managed to back the queue up so far it’s taken 36 hrs to clear it.

I seem to recall a past conversation where it was mentioned that the OCR engine was limited to a single thread for license reasons. I would be interested in paying extra to add-in a multicore OCR license. If you could find a way where the base DTPO was single-threaded, but I could purchase an additional license key and enable more threads I’d buy it in a heartbeat.

I recently tried the PDFPen 5 demo, and its OCR engine ran four threads and ripped through a document. I don’t know the quality, but I’d prefer to keep things in DTPO (with its queuing and Scansnap support built-in).

That’s indeed the case but we’ll consider this for future releases/upgrades if possible.

This would be an great feature.
Another variation of this would be for it to complete the queue and wait on me to enter in the details or to auto save with a sequential name that i would return later to rename the documents.

Hello development team,

this would really be a great feature to add. We have powerful machines with quad cores and it would be really good to make them work! Make DevonThink support multiple cores!

Best regards!

I could not agree more. I’d be willing to pay something more to enable multiple cores for the OCR function.

Love to see this feature for the future. :slight_smile:

So it’s almost a year and a half later and this is still pending. Could you perhaps provide an update on this please? This is becoming more of an issue and we’re currently having to rely on external software to do the OCR before importing the PDFs into DevonThink.

Thanks and regards,

Not much to report other than we are working with what we can. We understand the desire for this but short of writing our own multi-core OCR engine (and no, that’s not some trivial task or on our radar) we have to work with the APIs and allowances of the licensor.

Let us know if we need to stage a sit-in with Abbyy.