ABBYY FineReader doesn't seem to retain page layout

[u238110]

Screen shot speaks for itself. Also, anything else a beginner should know about ABBYY FineReader?




Does epub even support centered text and thing likes that? The methodology of these e-reader formats seems to be that of conforming the text to the display rather than maintaining the original look of the document. In any case, I still want to know why ABBYY FineReader is doing this.

[DaleDe]

Certainly ePub supports centered text and all sorts of things like that. Read about ePub in our Wiki and check out CSS.

Dale

[Tex2002ans]

EPUB export is only available as "Formatted Text" or "Plain Text". If you take a look at the explanation in the User's Guide, page 24:

http://www.abbyy.com/fr11guide_en.pdf

[...]
c. Formatted Text
Retains fonts, font sizes, and paragraphs, but does not retain the exact spacing or locations of the objects on the page. The text produced will be left–aligned. Texts in right–to–left scripts will be right–aligned.

Note: Vertical texts will be changed to horizontal in this mode.

d. Plain Text
This mode does not retain text formatting.

Although that is a slight lie, the EPUB export strips out even the fonts/font sizes to make even cleaner code.

If you wanted more complex formatting to be carried over, you would have to export to one of the formats supported in "Exact Copy", "Editable Copy", or you can take a shot at trying to clean up the HTML Exported by using "Flexible Copy" (although let me tell you, the HTML is UGLY).

[u238110]

That explains it! Thanks! Now, going by DaleDe's post, it seems like ABBYY FineReader uses a stripped down implementation of epub. Meaning they don't allows users to use the full extent of the format. Any idea why this is?

[Tex2002ans]

Originally Posted by AIDM2

That explains it! Thanks! Now, going by DaleDe's post, it seems like ABBYY FineReader uses a stripped down implementation of epub. Meaning they don't allows users to use the full extent of the format. Any idea why this is?

Well the EPUB export was just introduced in Finereader 11.... so they probably have tweaks they can do in making the export better in future versions (I see Finereader 12 was just recently released earlier this month, although no mention of better EPUB support).

Since the document is stored in a proprietary format, the Finereader engine itself probably transforms their data into some intermediary, and then transform THAT into the desired output format:

• When choosing "Formatted Text"
  • Step 1: Full Representation
  • Step 2: Formatted Text intermediary (stripping out most formatting, making it left aligned, only carrying over basic formatting)
  • Step 3: Export to chosen format
• When choosing "Plain Text"
  • Step 1: Full Representation
  • Step 2: Plain Text intermediary (stripping everything, only carrying over basic text)
  • Step 3: Export to chosen format.

So in order to support a new format, they probably only have to create a new "template" at Step 3, to take everything in that intermediary, and convert it into its equivalent in the output format.

So it probably isn't as simple as just "let's add center/right/justified justification when exporting to EPUB". An update to transfer over alignment in Step 2 would mean having to update all of the Step 3 transformations (most likely very complex) for all the other supported formats.

If you want to carry over more formatting, you would have to choose one of the many formats supported in "Exact Copy", "Editable Copy" or "Flexible" (HTML).

I personally think that the EPUB output is the best/most minimal output, with the cleanest code. Then you can easily just add your little tweaks (alignment, headings, split chapters, etc. etc.) using Sigil or Calibre's Editor. If you need any help with that, I would be glad to help... I use Finereader every day converting books.

[u238110]

Thanks a lot. Where can I post an ad asking if someone can digitize books for me (destructive method)?

[DaleDe]

Originally Posted by AIDM2

Thanks a lot. Where can I post an ad asking if someone can digitize books for me (destructive method)?

Check our wiki. There are plenty of people that will do this. E-book_conversion#Conversion_Services (mostly conversion services) and also Digitizing Paper Books to Ebooks

Dale

[Hitch]

Originally Posted by AIDM2

Thanks a lot. Where can I post an ad asking if someone can digitize books for me (destructive method)?

AIDM2:

I can highly recommend someone; PM me if you need references for someone to whom we refer our own clients.

Hitch

[u238110]

So, there's a book that has sanskrit text on the bottom of almost every page (see image). Would it be possible to program ABBYY FineReader to ignore every instance of that text?

Source: The Materia Medica of the Hindus (1877), Udoy Chand Dutt, Sir George King

[Tex2002ans]

Originally Posted by AIDM2

So, there's a book that has sanskrit text on the bottom of almost every page (see image).

Also, here is a link to the Archive.org version:

https://archive.org/details/materiamedicahi00kinggoog

Originally Posted by AIDM2

Would it be possible to program ABBYY FineReader to ignore every instance of that text?

You will have to manually intervene in those cases.

You can either (ranked from worst to best):

• Ignore those sections and clean it in when you output
• You can manually delete the garbage text from the Text Window (right side)
  
• You can resize the "Text Recognition Box" to not include the footnote
  • Don't forget to "Read" the page again.
  
• You can resize the text box, and then mark the footnote as an Image instead.
  • I would choose this method.
  • Don't forget to "Read" the page again.
  

Making the Sanskrit footnotes images will allow you to export them all, and perhaps you can then feed those images into an OCR tool that can read Sanskrit, you can more easily manually transcribe the text, or you can as a last resort, embed the sanskrit in the book as images (although I HIGHLY recommend against embedding text as images).

[u238110]

Thanks for taking the time to type up that quality description.

[mrmikel]

If you are going to save anything as an image, if the resolution is settable, make it as high as possible if there is a chance you might try to recognize it. You can always reduce resolution, but you can't go up.

[Tex2002ans]

Originally Posted by mrmikel

If you are going to save anything as an image, if the resolution is settable, make it as high as possible if there is a chance you might try to recognize it. You can always reduce resolution, but you can't go up.

Yes, always a good idea in ANY program to export as source/highest resolution.

I always assume this (although I guess if someone is not familiar with Finereader, or just getting into working with OCR/images, they wouldn't set this the first time around).

In Finereader, you can do this by going into Tools - Options - (press on the little "Save" tab) - Go through each format and make sure "Picture Settings" is set to "Best quality (source image resolution"):