Smoothing out KNote notes

[zsdiogo]

So I've spent the last semester using KNote to take my lecture notes, and I've been pretty happy with it. There are some features (like selection tools) that could be there and aren't, but that's not a big problem.
When I started using my notes to study for the exams, though, I noticed that the way the notes are exported is less than ideal. Although they are exported as PDF, the writing is not vectorised, and zooming a bit on the notes is enough to see the jagging and pixelization on the writing. Moreover, I think the guidelines behind the writing just add to the clutter, and there's no option to export the notes without the guidelines (except of just removing them before exporting)

I eventually found a way to work around this that I would like to share, and it is very simple. Using imagemagick one can remove the background and convert the PDF file to a set of bitmap files. The output (which is just black and white) can then be fed into potrace to vectorise the handwriting. Both programs are open-source. An early bash script for this would be along the lines of:

#!/usr/bin/env bash

FULL_EXT=${1?Error: no name given}
NAME_EXT=$(basename $FULL_EXT)
FULL_NOEXT=${FULL_EXT%.pdf}
NAME_NOEXT=${NAME_EXT%.pdf}
mkdir ~/tmp
convert ${FULL_EXT} -fill white +opaque "#000000" -background white -alpha remove ~/tmp/${NAME_NOEXT}.bmp
a="potrace -b pdf "
for entry in ~/tmp/*
do
	a+=$entry\ 
done
a+="-o ${FULL_NOEXT}.pdf"
$a
rm -r ~/tmp

And the result:


Detail comparison:


One could then construct a routine where the notes are synced from the KNote app to e.g. Dropbox, Dropbox syncs the files to a local folder on the computer, and then a script that monitors the folder for new versions of the files processes the files in the above way. So every time the notes are opened in the computer they would already be automatically processed.

EDIT: I forgot to say that, due to vectorisation, one can indefinitely zoom in the notes without loss of resolution.

[whopper]

bookmarking this thread because I had the same problem.

[retrography]

Great idea. One tip: If you explode that PDF with poppler’s pdfimages, you can simply take the writing layer images and discard the rest (including the background template).

[zsdiogo]

Originally Posted by retrography

Great idea. One tip: If you explode that PDF with poppler’s pdfimages, you can simply take the writing layer images and discard the rest (including the background template).

Yeah, I though about trying to expand the PDF to see if I could do something like that. But luckily all the backgrounds are in some shade of grey, and so it's easy to use imagemagick to solve two problems with one command: convert to bitmap and exclude all non-black colours.

[orpheus2011]

I have installed both imagemagick and potrace through brew on my Mac, but when attempting to run the script in Terminal, I keep getting this error:

bash graf.sh note8.pdf
convert: FailedToExecuteCommand `'gs' -sstdout=%stderr -dQUIET -dSAFER -dBATCH -dNOPAUSE -dNOPROMPT -dMaxBitmap=500000000 -dAlignToPixels=0 -dGridFitTT=2 '-sDEVICE=pngalpha' -dTextAlphaBits=4 -dGraphicsAlphaBits=4 '-r72x72'  '-sOutputFile=/var/folders/65/rnwl3xfs09vdnmb4v36ksh1c0000gn/T/magick-6019unPXRJTYmLWU%d' '-f/var/folders/65/rnwl3xfs09vdnmb4v36ksh1c0000gn/T/magick-6019mC1jfwxXKtfZ' '-f/var/folders/65/rnwl3xfs09vdnmb4v36ksh1c0000gn/T/magick-6019iTtQTQd9b6Vx'' (1) @ error/pdf.c/InvokePDFDelegate/290.
convert: no images defined `/Users/siavush/tmp/note8.bmp' @ error/convert.c/ConvertImageCommand/3300.
potrace: /Users/siavush/tmp/*: No such file or directory

Could anyone advise on what's going wrong?

Many thanks

[zsdiogo]

Originally Posted by orpheus2011

I have installed both imagemagick and potrace through brew on my Mac, but when attempting to run the script in Terminal, I keep getting this error:

bash graf.sh note8.pdf
convert: FailedToExecuteCommand `'gs' -sstdout=%stderr -dQUIET -dSAFER -dBATCH -dNOPAUSE -dNOPROMPT -dMaxBitmap=500000000 -dAlignToPixels=0 -dGridFitTT=2 '-sDEVICE=pngalpha' -dTextAlphaBits=4 -dGraphicsAlphaBits=4 '-r72x72'  '-sOutputFile=/var/folders/65/rnwl3xfs09vdnmb4v36ksh1c0000gn/T/magick-6019unPXRJTYmLWU%d' '-f/var/folders/65/rnwl3xfs09vdnmb4v36ksh1c0000gn/T/magick-6019mC1jfwxXKtfZ' '-f/var/folders/65/rnwl3xfs09vdnmb4v36ksh1c0000gn/T/magick-6019iTtQTQd9b6Vx'' (1) @ error/pdf.c/InvokePDFDelegate/290.
convert: no images defined `/Users/siavush/tmp/note8.bmp' @ error/convert.c/ConvertImageCommand/3300.
potrace: /Users/siavush/tmp/*: No such file or directory

Could anyone advise on what's going wrong?

Many thanks

It sounds like it might be a problem with the dependencies. Do you have ghostscript installed?

[orpheus2011]

Originally Posted by zsdiogo

It sounds like it might be a problem with the dependencies. Do you have ghostscript installed?

Many thanks - it was indeed the missing gs package! Works great now, but could you kindly recommend how to amend the convert command to keep the gridlines?

[zsdiogo]

Originally Posted by orpheus2011

Many thanks - it was indeed the missing gs package! Works great now, but could you kindly recommend how to amend the convert command to keep the gridlines?

Well PoTrace is very good at what is does, but it is also very restrictive in terms of features. If you give it the pic with the grid it will also trace over the grid. What you could do is just extract the grid and keep it somewhere in your PC, and add a line on the script with imagemagick to merge the output PDF with the grid PDF.