k2pdfopt: optimizes PDFs for viewing on e-readers

[pepijndevos]

Hi, I read most but not all of this thread, and I have this error:

Error rendering page 10 of djvu file blah.djvu.

 ** ERROR reading page 10 from blah.djvu.

Ay ideas?

[willus]

Originally Posted by pepijndevos

Hi, I read most but not all of this thread, and I have this error:

Error rendering page 10 of djvu file blah.djvu.

 ** ERROR reading page 10 from blah.djvu.

Any ideas?

It's hard to tell what the problem is without more information. Can you post the DjVu file that's causing the problem? Does k2pdfopt continue to convert subsequent pages in the source document or does it stop converting at page 10? What platform are you working on and what version of k2pdfopt are you using? Are you able to view page 10 of the DjVu file with any other DjVu reader (e.g. Sumatra PDF Reader or DjView)?

[pepijndevos]

Right. Well, it does continue. I'm on Mac, using the latest from your website. 1.51 or something, if I'm not mistaken. I have no other viewer installed. Probably just a corrupt file.

This is where I got the djvu file from, same old Lisp in Small Pieces I had trouble with last time

http://thepiratebay.se/torrent/6405192

[willus]

Originally Posted by pepijndevos

Right. Well, it does continue. I'm on Mac, using the latest from your website. 1.51 or something, if I'm not mistaken. I have no other viewer installed. Probably just a corrupt file.

I'm going to assume it's a corrupt file unless you can demonstrate otherwise by maybe checking it with an alternate reader.

[kalehrl]

Hello willus and other members of this forum.
I've recently bought my Kindle 4 black and I'm having a great time with it.
I found your great tool and converted several pdf with good results.
However, I have one pdf document which is driving me crazy.
Often page numbers end up in the middle of pages and I can't find a way to erase them.
I tried using -m option followed by 0.25,0.25,0.25,0.50 to cut more at the bottom but it didn't have a desired effect. I also tried to find an option to force pages to remain where they are in original document but it doesn't exist.
Could you have a look at it please.
Thank you.

[willus]

Originally Posted by kalehrl

Hello willus and other members of this forum.
I've recently bought my Kindle 4 black and I'm having a great time with it.
I found your great tool and converted several pdf with good results.
However, I have one pdf document which is driving me crazy.
Often page numbers end up in the middle of pages and I can't find a way to erase them.
I tried using -m option followed by 0.25,0.25,0.25,0.50 to cut more at the bottom but it didn't have a desired effect. I also tried to find an option to force pages to remain where they are in original document but it doesn't exist.
Could you have a look at it please.
Thank you.

You almost got it. If you use the command line, you do -mb 0.7 so that you cut off 0.7 inches at the bottom of each page, which is just enough to get rid of that pesky page number. If you use the interactive menu, you use option "m", and then type 0.25,0.25,0.25,0.7 (I should make -m on the command line work with a comma-separated list, too). Or, if you want the page numbers, use -bp (break pages) to break the destination page at the end of each source page, but k2pdfopt is still behaving strangely with your document--it leaves all of the vertical space between the page content and the page number, even when it should be compressing it. I'll look into that.

[kalehrl]

Thank you for your reply willus.
I actually tried higher bottom margin numbers but I thought it cropped some text besides getting rid page numbers because I didn't look carefully enough. When I examined the output pdf file again after your reply, I saw it didn't crop any text and it is OK now.
Thanks once again.

[pepijndevos]

This app makes my pdf book very readable, but there are a few small glitches.

The original book used hyphenation, but after con- version, it results in long words being split up. Sometimes this is very am- using.

At the start of a chapter, the first character spans multiple lines. This cause all the characters to the right to be copied as one block.

Similar to the above page number problem, this book lists the current chapter at the top of every page.

[kalehrl]

Originally Posted by willus

You almost got it. If you use the command line, you do -mb 0.7 so that you cut off 0.7 inches at the bottom of each page, which is just enough to get rid of that pesky page number. If you use the interactive menu, you use option "m", and then type 0.25,0.25,0.25,0.7 (I should make -m on the command line work with a comma-separated list, too). Or, if you want the page numbers, use -bp (break pages) to break the destination page at the end of each source page, but k2pdfopt is still behaving strangely with your document--it leaves all of the vertical space between the page content and the page number, even when it should be compressing it. I'll look into that.

I transferred the file to my Kindle but the text is so small that I can hardly read it. I tried to magnify it using interactive 'd' option and setting 'Output/Destination pixels per inch [167]:' to 300 according to the manual, but it didn't increase the font at all. Any ways to magnify it?

[willus]

Originally Posted by pepijndevos

The original book used hyphenation, but after con- version, it results in long words being split up...

I can't tell because you didn't post the source PDF, but if the source is low quality and the hyphens are touching the letters next to them, then the hyphen-detection algorithm in k2pdfopt will not consider it a hyphen and will not remove it. I've also tweaked the algorithm for the next release. If you want to post your source, I'll test it before the next release.

Originally Posted by pepijndevos

At the start of a chapter, the first character spans multiple lines. This cause all the characters to the right to be copied as one block.

I haven't tried to implement something to detect this yet. It's not trivial to reliably detect such a situation, but it could certainly be attempted. I'll put it on my list of possible future features. Theoretically there are adjustments you could make that would get k2pdfopt to detect the letter as a separate column (you'd have to set -cgr higher, close to 1, and -ch lower, probably to 0.5 or a little lower), but there may be adverse side effects (it may break the source contents into multiple columns where you don't want it to).

Originally Posted by pepijndevos

Similar to the above page number problem, this book lists the current chapter at the top of every page.

You can use the -mt option (top margin ignore). E.g. -mt 0.5 will ignore the top 0.5 inches of each source page.

[willus]

Originally Posted by kalehrl

I transferred the file to my Kindle but the text is so small that I can hardly read it. I tried to magnify it using interactive 'd' option and setting 'Output/Destination pixels per inch [167]:' to 300 according to the manual, but it didn't increase the font at all. Any ways to magnify it?

Magnification isn't working because there is a box around the text of each page and k2pdfopt is fitting that box to the display width. This is a tricky doc because it really needs to be full width to be typeset correctly. If you try and erase the vertical lines of the boxes (-evl 2 or menu option 's') so that k2pdfopt can wrap the text, a lot of the formatting is lost. For now, landscape mode is probably your best option (-ls or menu option 'l'). It will turn the doc sideways on your ebook so that the text is larger.

[davr]

This looks like a great tool, thanks for putting so much effort into it!

I just got a Kindle Paperwhite, which has a new higher res screen, and am trying to determine the correct width/height settings to prevent jagged text. The actual display is 1024x768 I believe, but I can't figure out the exact settings I need for -w and -h. Can you explain in more detail the method you used here: http://www.willus.com/k2pdfopt/help/screensize.shtml to determine the correct values? In particular the "special custom PDF images", wondering if I could re-use them. Thanks!

[willus]

K2pdfopt v1.60 is released. The major new feature is an option for native PDF output, i.e. using the exact rendering from the source file (with translation/scaling/cropping) rather than converting to bitmaps, much like soPdf and Cut2Col (imitation is the sincerest form of flattery, no?). There are several other new features as well as bug fixes. See the web site for details.

[willus]

Originally Posted by davr

Can you explain in more detail the method you used here: http://www.willus.com/k2pdfopt/help/screensize.shtml to determine the correct values? In particular the "special custom PDF images", wondering if I could re-use them.

I created a test image of exactly a certain size (not critical what is in it--maybe a grid or something) using the Windows Paint software, embedded that image into a PDF file, sent the PDF file to my kindle, and then did a screen grab on the kindle-displayed PDF and sent that screen-grabbed bitmap back to my PC to analyze it. Once the original bitmap that I created with Paint exactly matched the screen-grabbed bitmap, I had a 1-to-1 mapping. It took some doing, as described under v1.16 of the version history. I had to put little 1-pixel marks in the corners with some padding on the right and bottom.

[willus]

New help page for k2pdfopt v1.60 native PDF output.