I haven’t used this filter much in the past especially because it is a resource and processor-heavy filter and has even succeeded in crashing my PC at times. I’ve experimented with running it from a bash script and also with applying the effect to various downsized versions of a 24 megapixel image. It turns out downsizing the image to 2/3 of its dimensions restricts the resource-hunger of this particular filter to acceptable proportions but also has the most pleasing end result. Smaller and the image becomes almost unrecognizably abstract, larger and the barbouillage effect becomes hard to notice.
#!/bin/bash
let "nbfiles = $#"
ref=$(kdialog --title "Barbouillage" --progressbar "(Initialising ...)" 15)
qdbus $ref Set org.kde.kdialog.ProgressDialog maximum $nbfiles
compteur=0
for i in "$@";do
if [ -f "$i" ];then
let "compteur +=1"
FILE="$i"
fi
FILE_NO_EXT=${i%.*}
echo $i
echo $local
convert $i -filter lanczos -resize 66% $FILE_NO_EXT-presize.jpg
gmic -i $FILE_NO_EXT-presize.jpg -samj_Barbouillage_Paint_Daub 2,2,100,0.2,1,4,1,0,8 -o $FILE_NO_EXT-presize1.jpg
convert $FILE_NO_EXT-presize1.jpg -filter lanczos -resize 150% $FILE_NO_EXT-barbouillage66.jpg
rm -f $FILE_NO_EXT-presize1.jpg
rm -f $FILE_NO_EXT-presize.jpg
qdbus $ref setLabelText "Barbouillage: image created: `basename "$FILE"` ($compteur of $nbfiles files)"
qdbus $ref Set org.kde.kdialog.ProgressDialog value $compteur
done
qdbus $ref close
Results: