OCR Software...

[Scribble]

Can anyone give me a few tips on using OCR software to scan NPCs and articles and such?

I'm trying to convert some Dragon and Dungeon stuff to electronic format for use in my games, but I'd rather have something that isn't just a Jpeg or similar... (IE selectable text...)

I've never used OCR software before.

How well does it work?

Does it depend on your software?

How much "afterwards" work would I need to do to clean it up and all that?

 

[Rodrigo Istalindir]

The closer the source material is to double-spaced, single-column, no images just text, the better the OCR will be. Dragon articles are likely to be a real pain given the floating columns, side-bars, watermarked backgrounds, odd colored pages, etc. Tables can be real problematic, too.

It will be do-able, but it's going to be a lot of work, and a lot of fine-tuning the settings for the OCR for each article. Personally, unless I was going to need a lot of info, I think it would be easier to retype what I needed.

What you will find yourself doing is pre-scanning a page, adjusting the settings to accomodate background noise or odd colors, then using the software to frame different sections into distinct groups (eg click to select a column, click to exclude an image, a couple clicks to segregate a table). The good software will do a lot of this for you, but it's rarely 100%, so you can expect to have to tweak each page. Then, once you have the page mapped out, you do a final scan into your destination format (eg Word, PDF).

 

[Mycanid]

Probably your best bet with multi-columned stuff is to select individual parts of the page - that is, one column at a time to "OCR". We use Omnipage Pro 15 in the office here and it does this quite nicely. Give that a try....

 

[XCorvis]

Do you have a scanner? It probably came with some OCR software. ABBY FinePrint (?) and Omnipage are two OCR programs. Test them out and see how they work. If it looks workable, go ahead and try it. Most OCR programs are at least acceptably good. They'll probably get 95% of the document right. You'll want to run through with a spell checker, of course.

 

[Scribble]

Thank you for your replies and suggestions... I'll let you guys know how it went when I finally get around to doing it.

Anyone have any suggestions on how to get more time in my day?

 

[Mycanid]

Err ... other than quitting your job ... no. Of course, you could find a Ring of Sustenance and only need to sleep two hours a day and never eat. That'd save some time I guess....

 

[Asmor]

Google SimpleOCR. It's the best (ok, only) free product I found with an (admittedly short) search.

It doesn't work very well on Dragon articles, though. I scanned one in at a very high res, used Photoshop's histogram to get it almost entirely black text on a white background (only scattered specks left), and there was just a ton of mistakes. I don't know if it's the software or the source material, but personally I'd reccomend not bothering.