k2pdfopt: optimizes PDFs for viewing on e-readers

[geekmaster]

Originally Posted by RefUser

You may call me silly (or at least my question here), but I still see this thread as a possibility to ask willus, since he answered some questions and announced the last releases here and I don't need an answer tomorrow. If he goes by in a month or two, it's perfectly fine.

In addition, interest of a lot of people could be a reason for someone interested in F/OSS (see his website) to release the source code.
Using the dual-licensed mupdf (see changelog) the code can be quite useful for some pdf readers, but even studying it and writing an export option for e.g. evince or okular could be easier than a fresh start.

How shall I know, whether I'm the only one interested in such a port (and/or a source code release) without asking in public?

I did not notice the replies by willus. If he is active here, then your request is valid -- sorry. I was just trying to point out that willus WOULD be more likely to respond to requests at his own website. I misread your post when I called it silly. I *thought* it said willus did NOT post in this thread. Oops, sorry. It was not "silly". It only *seemed* silly because I did not read it correctly on my tiny Nexus 7 screen.

[Paveuta]

Willus: I went crazy yesterday to find a way to convert some documents I need to read for work for my new Sony PRS-T1 in a PDF readable way... I am still useless at Calibre and want them in there fast. No solution would work, your little gem... a charm!

So a big THANKS!! from me for solving my problem.

I know the answer is somewhere in the forum, but do you know why the highlighting function on my device does not work, only the handwriting on the page?

I will keep on looking but again, thanks for your little masterpiece.

[thatworkshop]

Originally Posted by RefUser

What about the first post? It was not posted by willus himself and he did answer a lot of questions in this thread.
Finally, I asked here (and not via mail to the author), because I also wanted to see, whether other users are also interested in such a port.

I asked William through email for the same requests you mentioned here, like 6-7 months ago... His concern was if Kindle can handle the time/resource consuming that k2pdfopt undertakes and I replied yes... I think he might be working on something, but I'm not sure... About source code, it's unlikely, but I'd really appreciate too.

[thomass]

Originally Posted by Paveuta

... do you know why the highlighting function on my device does not work, only the handwriting on the page?...

The app needs to convert the pages into images in order to do this process. Therefore you need to do an OCR (with ABBYY, etc) on the output file to be able to highlight, etc.

[willus]

Jens Wallauer has contributed a Windows GUI for k2pdfopt.

[willus]

Originally Posted by thomass

The app needs to convert the pages into images in order to do this process. Therefore you need to do an OCR (with ABBYY, etc) on the output file to be able to highlight, etc.

Thomass is correct that k2pdfopt converts to images, so no highlighting can be done. I am hoping to add OCR capability to k2pdfopt soon (via Google's Tesseract engine). I have a test version working on English text. I haven't been able to find as much programming time as I've wanted to this summer, but eventually I'll get it released.

[willus]

Originally Posted by cscat

I asked William through email for the same requests you mentioned here, like 6-7 months ago... His concern was if Kindle can handle the time/resource consuming that k2pdfopt undertakes and I replied yes... I think he might be working on something, but I'm not sure... About source code, it's unlikely, but I'd really appreciate too.

I am not working on anything for ARM. An ARM port from me is not likely to happen any time soon. My main concern is that I have no experience with ARM development and would have to port nearly a dozen open-source libraries (~1 million lines of code) to ARM--no small task, and I have other priorities at the moment (like adding OCR). Finally, as cscat said, the memory use and CPU use of k2pdfopt is considerable. I think the results would typically be frustratingly slow.

[willus]

Originally Posted by colateral

thank you willus

for exemple this document

[MOD: Link to copyright material removed]

after using the app , / lastest version / with default setting
i get the result
[MOD: Link to copyright material removed]

kobo read arabic pdf directly like kindle .

Colateral--sorry I didn't see this until now. You need to find a way to post or get me a sample of your source file. Maybe if it's only one or two pages that should be enough and hopefully won't offend the copyright police(?). Sorry for your trouble.

[Paveuta]

Originally Posted by thomass

The app needs to convert the pages into images in order to do this process. Therefore you need to do an OCR (with ABBYY, etc) on the output file to be able to highlight, etc.

Ehm... still not that expert but I will look up what you said and update.

Thanks for replying!

Giuditta.

[RefUser]

Originally Posted by willus

I am not working on anything for ARM. An ARM port from me is not likely to happen any time soon. My main concern is that I have no experience with ARM development and would have to port nearly a dozen open-source libraries (~1 million lines of code) to ARM--no small task, and I have other priorities at the moment (like adding OCR). Finally, as cscat said, the memory use and CPU use of k2pdfopt is considerable. I think the results would typically be frustratingly slow.

I made some tests a few days ago and got about 2-3 minutes per page on a 600Mhz ARM with qemu-x86 emulation. No idea how much the emulator slows things down, but a native binary should be at least 2 times as fast (not to mention what's possible on a 1,2Ghz dual core ARM). Therefore, converting a research paper during lunch seams plausible


Did you consider releasing your program as open source software? Mupdf and it's thirdparty libraries easily compile for ARM (android, and GCC/X11 based). So if you have no time (or reason) to port it, please let others try to port/use it.
Btw: You know that Mupdf is not BSD-licensed but GPL (and you link statically)?

[JJJM]

Hello,

I am new to this software which I just discovered. I have been playing around with it but I do not get proper results. Thus I wondered I should deal with options and its parameters but I do not know how.

My goal is very simple, I just want to read novels I have scanned using my kindle 2. So the structure of the document is very simple and similar along the pages. Could you please tell me the options I should use to get a good pdf for my kindle?

I give you a pdf sample of first 10 pages of a book I wanted to convert.

http://www.mediafire.com/?c5psm6s2arkn476

Thanks in advance for your advice.

[willus]

Originally Posted by JJJM

Hello,

I am new to this software which I just discovered. I have been playing around with it but I do not get proper results. Thus I wondered I should deal with options and its parameters but I do not know how.

My goal is very simple, I just want to read novels I have scanned using my kindle 2. So the structure of the document is very simple and similar along the pages. Could you please tell me the options I should use to get a good pdf for my kindle?

I give you a pdf sample of first 10 pages of a book I wanted to convert.

http://www.mediafire.com/?c5psm6s2arkn476

Thanks in advance for your advice.

Thanks for the example--that helps. I'd use these options:

-m .05 -as -ws .2 -col 1

-m .05 to clip the margins less, -as to autostraighten each page, -ws .2 will separate words more consistently for your file, and -col 1 since the source is not multicolumn. K2pdfopt still has a little trouble because the lines of text are slightly distorted (curved), but mostly the output looks very readable to me. If you want the output text larger or smaller, add -odpi XXX where XXX is larger or smaller than the default of 167, respectively.

[JJJM]

Originally Posted by willus

Thanks for the example--that helps. I'd these use options:

-m .05 -as -ws .2 -col 1

-m .05 to clip the margins less, -as to autostraighten each page, -ws .2 will separate words more consistently for your file, and -col 1 since the source is not multicolumn. K2pdfopt still has a little trouble because the lines of text are slightly distorted (curved), but mostly the output looks very readable to me. If you want the output text larger or smaller, add -odpi XXX where XXX is larger or smaller than the default of 167, respectively.

Thanks for your quick and precise reply.

I have exactly typed: k2pdfopt *.djvu -m 0.05 -as- -ws 0.2 -col 1

And it says invalid entry. I have checked examples and I think I did it right but there must be sth wrong. Could you help?

Thanks again.

[knc1]

Originally Posted by JJJM

Thanks for your quick and precise reply.

I have exactly typed: k2pdfopt *.djvu -m 0.05 -as- -ws 0.2 -col 1

And it says invalid entry. I have checked examples and I think I did it right but there must be sth wrong. Could you help?

Thanks again.

Try: "Options before Arguments"
k2pdfopt -m 0.05 -as -ws 0.2 -col 1 *.djvu
And if that doesn't work, try it on a single file first.
Note that whitespace and number/position of "-" are significant.

[JJJM]

Originally Posted by knc1

Try: "Options before Arguments"
k2pdfopt -m 0.05 -as -ws 0.2 -col 1 *.djvu
And if that doesn't work, try it on a single file first.
Note that whitespace and number/position of "-" are significant.

Hi again.

I have tried this way and does not work.

Even I have used less options. Also with a single file. I have also changed 0.05 by 0,05, but nothing.

But if I use no options, it works.