k2pdfopt: optimizes PDFs for viewing on e-readers

[danilo93]

Originally Posted by willus

Thanks for posting the source document. First off, the page size is 16.6 x 25.4 inches, which I'm guessing should actually be centimeters, so you need to use the document scaling factor to get it to the right size. Secondly, some of the pages are skewed, so you need to turn on the auto-straightening feature. I think with these two adjustments, the result will be satisfactory:

-ds 0.4 (set document scaling factor to 0.4--or select "ds" in the menu and then enter 0.4 for the value).

-as (Turn on auto-straightening--or select "a" in the menu).

Aaaaaaaaaaaaaaaaaaaaa! Yeeeeeaaah! These options are great! I needed this! Thank you sooo much! I understand now what this first "ds" option is...I just don't know how did you know that value should be 0.4! Can you tell me how did you calculate it? Or you just tried it untill it was ok...

[willus]

Originally Posted by danilo93

Aaaaaaaaaaaaaaaaaaaaa! Yeeeeeaaah! These options are great! I needed this! Thank you sooo much! I understand now what this first "ds" option is...I just don't know how did you know that value should be 0.4! Can you tell me how did you calculate it? Or you just tried it untill it was ok...

I'm glad that worked. I use Sumatra PDF to view PDFs on my PC. If you press Ctrl-D while viewing a PDF with Sumatra, it will display the size of the page being viewed. I'm sure other PDF viewers have similar functionality. I just guessed that the size was supposed to be in centimeters rather than inches, since it seemed about right for a letter-sized page, so I used 1/2.54 = roughly 0.4.

[Fourl29]

Looks like a great piece of software, but how is it any better than what say an inbuilt re-flow-er or the send to kindle pdf software amazon use?

[willus]

Originally Posted by Fourl29

Looks like a great piece of software, but how is it any better than what say an inbuilt re-flow-er or the send to kindle pdf software amazon use?

There are a number of examples on the web page that include both the original and k2pdfopt-converted PDFs. You can convert these originals using any other method you like and compare the results. The FAQ page might help you as well.

[danilo93]

Originally Posted by willus

I'm glad that worked. I use Sumatra PDF to view PDFs on my PC. If you press Ctrl-D while viewing a PDF with Sumatra, it will display the size of the page being viewed. I'm sure other PDF viewers have similar functionality. I just guessed that the size was supposed to be in centimeters rather than inches, since it seemed about right for a letter-sized page, so I used 1/2.54 = roughly 0.4.

Thank you so much for this! It really helped! You rock!

[markom]

Originally Posted by Fourl29

Looks like a great piece of software, but how is it any better than what say an inbuilt re-flow-er or the send to kindle pdf software amazon use?

Do you know any other reflower there, inbuilt or standalone application, that reflows pdf images ?

If for example our pdf is ocr-ed image we can use inbuilt reflower in our e-ink reader but for ocr layer only, not original image, so if our scanned pdf contains a lot of formulas, scientific expressions, odd scripts etc. there will regrettably be plenty of ocr errors, but not with k2pdfopt which produces exact copy of sentences, reflowed.

[Fourl29]

Originally Posted by markom

Do you know any other reflower there, inbuilt or standalone application, that reflows pdf images ?

If for example our pdf is ocr-ed image we can use some reflower but for ocr layer only, not original image, so if our pdf contains a lot of formulas, scientific expressions, odd scripts etc. there will be a lot of ocr errors, but not with k2pdfopt which produces exact copy of pdf imager..

Save me the defensive anger and just tell me that it reflows non ocr images. I am not supposed to know everything the software does that's why I posed a question. And if you are aware of any other program that does it be my guest and mention it.

It can reflow an image too, ok, I get this, but it doesn't use ocr so it just chops up the image and reflows it thus. That's something that sounds pretty cool. I think amazon's pdf to kindle does something similar, although it's not always effective, and the ocr might always kick in.

My main qualm with similar converters and implementations is for the software to be able to figure out what it can effectively do and what not. So for example if it can ocr a formula correctly, that is without seriously messing it up, that it should go ahead and ocr it, but if it can tell that the end result is a mess, it should just crop/resize/split the image instead and leave anything it can't deal with as image. That's the problem I get with most of my technical books. Formulas and tables get screwed over.

[markom]

Abbyy Finereader 11 has training mode, where we can teach program how to interpret special characters.

http://www.youtube.com/watch?v=LAJm3J36tLQ

I don't know though, if there is possibility to force Abbyy to automatically show formulas as images or at least whole line (if there is some mathematical expresssion inside), instead of doing it manually page by page.

I still use Abbyy or Acrobat for detailed or quick ocr-ing of k2pdfopt files, even though there is ocr capability within app now.

[kalehrl]

Hi willus
Whenever there is some text in the middle of the page between 2 columns, the output is wrong. Please find attached the examples. Is there any way to rectify it?


[This Creative Commons licensed magazine is allowed to be attached to a message on MobileRead.]

[willus]

Originally Posted by kalehrl

Hi willus
Whenever there is some text in the middle of the page between 2 columns, the output is wrong. Please find attached the examples. Is there any way to rectify it?

Interesting magazine. I had read some about Raspberry Pis--they look like a fun hobby device.

I tried some different options and this is the best I could come up with. It's not perfect, but it did keep the basic flow in the right order.

k2pdfopt -cgr 0.6 -ch 1 -crgh 0.01 magpi.pdf

-cgr 0.6 gets k2pdfopt to look over a wider range than normal for column splits--the middle text moves the column splits outside of its normal range (0.33). A value of 1.0 scans the full page width for column splits.

-ch 1 reduces the minimum region height for column breaks from 1.5 inches to 1 inch since the middle text is less than 1.5 inches high.

-crgh 0.01 allows a slightly smaller-than-default (default = 1/72 inch) gap to separate regions of differing column widths since the vertical gap between columns of different widths is small in this case.

You might also try -col 4, but that may result in "over-columnizing" on other pages. I prefer to avoid it unless the bulk of the document has 3 or more columns. See also this post with a similar issue.

[kalehrl]

Yes, raspberry pi is a wonderful, small, low-cost linux tinkering device
I converted the magazine with your command line options and now it looks much better.
The middle text is no more wreaking havoc with the conversion.
Many thanks willus.

[willus]

Originally Posted by kalehrl

Yes, raspberry pi is a wonderful, small, low-cost linux tinkering device
I converted the magazine with your command line options and now it looks much better.
The middle text is no more wreaking havoc with the conversion.
Many thanks willus.

Glad to help. Hopefully at some point I can make k2pdfopt smart enough to find things like that automatically (without having to resort to precise command-line adjustments).

[hedere]

hi willus, thank you for you efforts.
i'm resizing pdf articles which includes tables with numerics, but some tables are oversized, some are not. i couldn't find a solution with various options..i attached th pdfs.. what do you think?
thanks

[Copyright PDFs deleted. Please do not post copyright material to MobileRead without the explicit permission of the copyright holder.]

[pspvhp]

thank. very cool

[willus]

Originally Posted by hedere

hi willus, thank you for you efforts.
i'm resizing pdf articles which includes tables with numerics, but some tables are oversized, some are not. i couldn't find a solution with various options..i attached th pdfs.. what do you think?
thanks

[Copyright PDFs deleted. Please do not post copyright material to MobileRead without the explicit permission of the copyright holder.]

@hedere -- I sent you a private message. Please check.