k2pdfopt: optimizes PDFs for viewing on e-readers

[willus]

Originally Posted by VirtualMaestro

Another interesting thing is with default params new version 2.01 generates pdf almost 20% greater then old v1.62. Maybe there are some default params changed.

I'll fix the "always on top" issue for the next release (I've already added a minimize button to the convert dialog box). Can you post the source PDF file that results in converted PDF output from v2.01 which is 20% larger than v1.62 (and any special options you used in the conversion)? I'll check it. Part of it may be that v2.01 now automatically includes the text from non-scanned PDF source files.

[kim1406]

Originally Posted by willus

Part of the problem is that your document dimensions seem very large--13 x 20 inch pages--this was making the default output size very large. I used -ds 0.5 to scale them down to a more normal page size. After that, it was a matter of using a high enough value for -gtr and a low enough value for -ws so that k2pdfopt could correctly split the rows and words, and also using autostraighten. This is what I ended up with:

k2pdfopt -ds 0.5 -as -r -dev kpw -gtr 0.05 -ws 0.25 file.pdf

WOW, it is great now. Thank you.

For this issue, if main problem was that file dimensions are large, then, how can we determine if file dimensions are ok or not?

[willus]

Originally Posted by kim1406

WOW, it is great now. Thank you.

For this issue, if main problem was that file dimensions are large, then, how can we determine if file dimensions are ok or not?

You can see the file dimensions listed as k2pdfopt processes the pages (see example below). If they are considerably larger than a letter-sized page, but they don't hold more text than a normal letter-sized page, then you should size them down with the -ds option.

[VirtualMaestro]

Hi Willus,
Thanks for fixing.
Here is source of pdf [Shortened URL - moderator]
And here is archive with converted results (converted by new and old ver):
[Shortened URL - moderator]

Thanks

[willus]

K2pdfopt v2.02 is released. This fixes most reported issues to date, including the hyphen detection. It also implements a faster bitmap re-sampling function for 32-bit versions. See the web site for details.

[willus]

Originally Posted by VirtualMaestro

Hi Willus,
Thanks for fixing.
Here is source of pdf [Shortened URL - moderator]
And here is archive with converted results (converted by new and old ver):
[Shortened URL - moderator]

Thanks

On source PDFs like these that have native text, the new version, by default, now uses "virtual OCR" to embed the unicode text with the graphics so that you can highlight, search, copy and paste. Notice on the new converted PDF you should be able to select text (in most modern PDF viewers, anyway). This is not true for v1.6x. This accounts for the larger size (the embedded text). You can turn this feature off with the option: -ocr-
...but since there is very little penalty (apart from the slightly larger output file size), and a large upside, it is on by default.

[kim1406]

Good news about the new release.

I have a problem with some native pdf files with Arabic letters. K2pdfopt will output files with none-readable fonts/characters (squares instead of normal Arabic fonts). I tried both options, with /and without native PDF output which gave me the same result.

Attached is a sample file along with the output file, you may please look at the problem.
thanks for your great help.

[VirtualMaestro]

Hello Willus,
Thanks for new version!
In new version I have found one bug with Custom Presets.
All properties saves correctly except 'Native PDF output'. When I restarted programm and tried to apply my preset this checkbox wasn't checked.

Thanks

[willus]

Originally Posted by VirtualMaestro

Hello Willus,
Thanks for new version!
In new version I have found one bug with Custom Presets.
All properties saves correctly except 'Native PDF output'. When I restarted programm and tried to apply my preset this checkbox wasn't checked.

Thanks

Did you check something else incompatible with native PDF output, such as OCR or text re-flow? If you did, then native PDF output gets turned off. If you're not sure, then post a screen shot of the options.

[willus]

Originally Posted by kim1406

Good news about the new release.

I have a problem with some native pdf files with Arabic letters. K2pdfopt will output files with none-readable fonts/characters (squares instead of normal Arabic fonts). I tried both options, with /and without native PDF output which gave me the same result.

Attached is a sample file along with the output file, you may please look at the problem.
thanks for your great help.

This is strange. The native PDF conversion displays fine on my system (Windows 64-bit using Sumatra PDF viewer). But turning native PDF off for me gets me the same result as you. It's also giving me the warning: "non-embedded font using identity encoding: SimplifiedArabic" from MuPDF. This is likely at the heart of the issue. If it displays correctly in Sumatra, though, then k2pdfopt should display it correctly since it uses the same MuPDF library as Sumatra. I'll look into it more. Are you sure you truly got a native PDF conversion? If you turn on text wrapping or OCR, it will override native output and turn it off. You can tell by zooming in heavily on the converted file. If the output pixelates at high zoom, it's not native. Can you post the result from the native PDF conversion?

[VirtualMaestro]

Originally Posted by willus

Did you check something else incompatible with native PDF output, such as OCR or text re-flow? If you did, then native PDF output gets turned off. If you're not sure, then post a screen shot of the options.

Oh, yeah, I checked OCR, didn't know about that, sorry, my bad

[willus]

Originally Posted by VirtualMaestro

Oh, yeah, I checked OCR, didn't know about that, sorry, my bad

Not at all. K2pdfopt should only check the correct boxes and should uncheck the incompatible ones for you. So it's really my bad. I'll fix that up in the next release. The default value of native PDF output should also be off--that is also a bug.

[willus]

Originally Posted by kim1406

Good news about the new release.

I have a problem with some native pdf files with Arabic letters. K2pdfopt will output files with none-readable fonts/characters (squares instead of normal Arabic fonts). I tried both options, with /and without native PDF output which gave me the same result.

Attached is a sample file along with the output file, you may please look at the problem.
thanks for your great help.

Okay--first off, the converted file should display correctly if you convert using native mode. I'm guessing you didn't actually get native mode turned on because you had text re-flow or OCR turned on (these disable native-mode conversions). Secondly, I didn't understand why k2pdfopt would not render your font (Simplified Arabic) and Sumatra would when they are both based on the MuPDF library, so I did some investigating. It turns out that there are modifications to the MuPDF library specifically for Sumatra that enable it to correctly render MS Windows system fonts when they are not already embedded in the PDF file (the MuPDF library, as downloaded from the MuPDF site, does not do this). So I have incorporated those Sumatra mods into my version of the MuPDF library, and the next release of k2pdfopt will correctly render this font (or any other MS Windows system font which is not directly embedded in the PDF file).

[kim1406]

Originally Posted by willus

Okay--first off, the converted file should display correctly if you convert using native mode. I'm guessing you didn't actually get native mode turned on because you had text re-flow or OCR turned on (these disable native-mode conversions). Secondly, I didn't understand why k2pdfopt would not render your font (Simplified Arabic) and Sumatra would when they are both based on the MuPDF library, so I did some investigating. It turns out that there are modifications to the MuPDF library specifically for Sumatra that enable it to correctly render MS Windows system fonts when they are not already embedded in the PDF file (the MuPDF library, as downloaded from the MuPDF site, does not do this). So I have incorporated those Sumatra mods into my version of the MuPDF library, and the next release of k2pdfopt will correctly render this font (or any other MS Windows system font which is not directly embedded in the PDF file).

I re-checked again and found out that, when using the gui mode, enabling native pdf, re-flow and OCR turned off, k2pdfopt will output a file without the correct font, but, when using the commands mode (no gui), the output file with be fine with the Arabic font.

[willus]

Originally Posted by kim1406

I re-checked again and found out that, when using the gui mode, enabling native pdf, re-flow and OCR turned off, k2pdfopt will output a file without the correct font, but, when using the commands mode (no gui), the output file with be fine with the Arabic font.

I was able to reproduce this behavior (native mode output not correctly turned on in GUI). I'll fix it in the next release. In the mean time, you can put -n in the "additional options" box to get native mode output. Thank you for reporting it.