10-01-2012 update: I have noticed many hits in google to this post. If you would like to read about how to deal with kanji recognition in your program, head to this post.
I have completed the stroke order database for all JIS X 0208-1990 kanji, to be included in zkanji. That’s 6355 kanji that most Japanese fonts contain, and it took quite some time to finish. (No wonder making such a font file takes 3-4 years.) There have been open projects out there already (or at least one, called kanjiVG – hopefully fixed soon) that attempted the same. I started out using the Taka database years ago which contains a bit more than the Jouyou kanji. (When I started I didn’t know about kanjiVG, and the Taka database didn’t have all Jouyou kanji, but I had to start somewhere.)
The reason why my project is unique (I’m making this up), is that the information stored in the data file for zkanji doesn’t only contain coordinates for strokes and their order, and of course the kanji parts, but also some additional information about the type of stroke. That’s why it is possible (since v0.573) to display these stroke orders with a bit more style.
I have plans to add other information to this data. For example it would be nice if I could show which part of the kanji corresponds to a given ON reading. I hope that it can be done automatically from the data already in zkanji(, partly taken from KANJIDIC, and partly from the stroke order diagrams). I might have to specify other relations between kanji. For example when one kanji is the old form of another, and not used in modern writing anymore. This is not going to happen any time soon though.
The next step? I could release the stroke order information publicly. It is currently not free (can only be released with zkanji), but if someone asks nicely, I might write a post about the data format and even release it with some CC type license.
DISCLAIMER: zkanji doesn’t use data from the Taka database anymore. The stroke order diagrams are all my work.